Search

Aims: To create and validate a weakly supervised artificial intelligence (AI) model for detection of abnormal colorectal histology, including dysplasia and cancer, and prioritise biopsies according to clinical significance (severity of diagnosis)., Materials and Methods: Triagnexia Colorectal, a weakly supervised deep learning model, was developed for the classification of colorectal samples from haematoxylin and eosin (H&E)-stained whole slide images. The model was trained on 24 983 digitised images and assessed by multiple pathologists in a simulated digital pathology environment. The AI application was implemented as part of a point and click graphical user interface to streamline decision-making. Pathologists assessed the accuracy of the AI tool, its value, ease of use and integration into the digital pathology workflow., Results: Validation of the model was conducted on two cohorts: the first, on 100 single-slide cases, achieved micro-average model specificity of 0.984, micro-average model sensitivity of 0.949 and micro-average model F1 score of 0.949 across all classes. A secondary multi-institutional validation cohort, of 101 single-slide cases, achieved micro-average model specificity of 0.978, micro-average model sensitivity of 0.931 and micro-average model F1 score of 0.931 across all classes. Pathologists reflected their positive impressions on the overall accuracy of the AI in detecting colorectal pathology abnormalities., Conclusions: We have developed a high-performing colorectal biopsy AI triage model that can be integrated into a routine digital pathology workflow to assist pathologists in prioritising cases and identifying cases with dysplasia/cancer versus non-neoplastic biopsies., (© 2024 John Wiley & Sons Ltd.)

Using Deep Learning in computer-aided diagnosis systems has been of great interest due to its impressive performance in the general domain and medical domain. However, a notable challenge is the lack of explainability of many advanced models, which poses risks in critical applications such as diagnosing findings in CXR. To address this problem, we propose ItpCtrl-AI, a novel end-to-end interpretable and controllable framework that mirrors the decision-making process of the radiologist. By emulating the eye gaze patterns of radiologists, our framework initially determines the focal areas and assesses the significance of each pixel within those regions. As a result, the model generates an attention heatmap representing radiologists' attention, which is then used to extract attended visual information to diagnose the findings. By allowing the directional input, our framework is controllable by the user. Furthermore, by displaying the eye gaze heatmap which guides the diagnostic conclusion, the underlying rationale behind the model's decision is revealed, thereby making it interpretable. In addition to developing an interpretable and controllable framework, our work includes the creation of a dataset, named Diagnosed-Gaze++, which aligns medical findings with eye gaze data. Our extensive experimentation validates the effectiveness of our approach in generating accurate attention heatmaps and diagnoses. The experimental results show that our model not only accurately identifies medical findings but also precisely produces the eye gaze attention of radiologists. The dataset, models, and source code will be made publicly available upon acceptance., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Trong-Thang Pham reports financial support was provided by National Science Foundation. Trong-Thang Pham reports financial support was provided by National Institutes of Health. Carol Wu reports financial support was provided by National Institutes of Health. Carol Wu reports financial support was provided by National Science Foundation. Jacob Brecheisen reports financial support was provided by National Science Foundation. Hien Nguyen reports financial support was provided by National Science Foundation. Hien Nguyen reports financial support was provided by National Institutes of Health. Zhigang Deng reports financial support was provided by National Science Foundation. Zhigang Deng reports financial support was provided by National Institutes of Health. Ngan Le reports financial support was provided by National Science Foundation. Ngan Le reports financial support was provided by National Institutes of Health. Donald Adjeroh reports financial support was provided by National Science Foundation. Gianfranco Doretto reports financial support was provided by National Science Foundation. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier B.V. All rights reserved.)

Atherosclerosis causes heart disease by forming plaques in arterial walls. IVUS imaging provides a high-resolution cross-sectional view of coronary arteries and plaque morphology. Healthcare professionals diagnose and quantify atherosclerosis physically or using VH-IVUS software. Since manual or VH-IVUS software-based diagnosis is time-consuming, automated plaque characterization tools are essential for accurate atherosclerosis detection and classification. Recently, deep learning (DL) and computer vision (CV) approaches are promising tools for automatically classifying plaques on IVUS images. With this motivation, this manuscript proposes an automated atherosclerotic plaque classification method using a hybrid Ant Lion Optimizer with Deep Learning (AAPC-HALODL) technique on IVUS images. The AAPC-HALODL technique uses the faster regional convolutional neural network (Faster RCNN)-based segmentation approach to identify diseased regions in the IVUS images. Next, the ShuffleNet-v2 model generates a useful set of feature vectors from the segmented IVUS images, and its hyperparameters can be optimally selected by using the HALO technique. Finally, an average ensemble classification process comprising a stacked autoencoder (SAE) and deep extreme learning machine (DELM) model can be utilized. The MICCAI Challenge 2011 dataset was used for AAPC-HALODL simulation analysis. A detailed comparative study showed that the AAPC-HALODL approach outperformed other DL models with a maximum accuracy of 98.33%, precision of 97.87%, sensitivity of 98.33%, and F score of 98.10%., Competing Interests: Declarations. Conflict of interest: The authors declare no competing interests., (© 2024. International Federation for Medical and Biological Engineering.)

The potential and promise of deep learning systems to provide an independent assessment and relieve radiologists' burden in screening mammography have been recognized in several studies. However, the low cancer prevalence, the need to process high-resolution images, and the need to combine information from multiple views and scales still pose technical challenges. Multi-view architectures that combine information from the four mammographic views to produce an exam-level classification score are a promising approach to the automated processing of screening mammography. However, training such architectures from exam-level labels, without relying on pixel-level supervision, requires very large datasets and may result in suboptimal accuracy. Emerging architectures such as Visual Transformers (ViT) and graph-based architectures can potentially integrate ipsi-lateral and contra-lateral breast views better than traditional convolutional neural networks, thanks to their stronger ability of modeling long-range dependencies. In this paper, we extensively evaluate novel transformer-based and graph-based architectures against state-of-the-art multi-view convolutional neural networks, trained in a weakly-supervised setting on a middle-scale dataset, both in terms of performance and interpretability. Extensive experiments on the CSAW dataset suggest that, while transformer-based architecture outperform other architectures, different inductive biases lead to complementary strengths and weaknesses, as each architecture is sensitive to different signs and mammographic features. Hence, an ensemble of different architectures should be preferred over a winner-takes-all approach to achieve more accurate and robust results. Overall, the findings highlight the potential of a wide range of multi-view architectures for breast cancer classification, even in datasets of relatively modest size, although the detection of small lesions remains challenging without pixel-wise supervision or ad-hoc networks., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lia Morra, Fabrizio Lamberti, Andrea Pagnani report financial support was provided by HealthTriage srl. Lia Morra, Francesco Manigrasso, Alessandro Russo, Rosario Milazzo has patent pending to HealthTriage srl. Fredrik Strand reports speaker fees from Lunit and Pfizer. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved.)

Breast cancer is the most common cancer in women. Ultrasound is one of the most used techniques for diagnosis, but an expert in the field is necessary to interpret the test. Computer-aided diagnosis (CAD) systems aim to help physicians during this process. Experts use the Breast Imaging-Reporting and Data System (BI-RADS) to describe tumors according to several features (shape, margin, orientation...) and estimate their malignancy, with a common language. To aid in tumor diagnosis with BI-RADS explanations, this paper presents a deep neural network for tumor detection, description, and classification. An expert radiologist described with BI-RADS terms 749 nodules taken from public datasets. The YOLO detection algorithm is used to obtain Regions of Interest (ROIs), and then a model, based on a multi-class classification architecture, receives as input each ROI and outputs the BI-RADS descriptors, the BI-RADS classification (with 6 categories), and a Boolean classification of malignancy. Six hundred of the nodules were used for 10-fold cross-validation (CV) and 149 for testing. The accuracy of this model was compared with state-of-the-art CNNs for the same task. This model outperforms plain classifiers in the agreement with the expert (Cohen's kappa), with a mean over the descriptors of 0.58 in CV and 0.64 in testing, while the second best model yielded kappas of 0.55 and 0.59, respectively. Adding YOLO to the model significantly enhances the performance (0.16 in CV and 0.09 in testing). More importantly, training the model with BI-RADS descriptors enables the explainability of the Boolean malignancy classification without reducing accuracy., Competing Interests: Declarations. Conflict of interest: The authors declare no competing interests. Ethics Approval: All the data have been obtained from public datasets, so ethics approval was not required., (© 2024. The Author(s).)

Viruses are submicroscopic agents that can infect other lifeforms and use their hosts' cells to replicate themselves. Despite having simplistic genetic structures among all living beings, viruses are highly adaptable, resilient, and capable of causing severe complications in their hosts' bodies. Due to their multiple transmission pathways, high contagion rate, and lethality, viruses pose the biggest biological threat both animal and plant species face. It is often challenging to promptly detect a virus in a host and accurately determine its type using manual examination techniques. However, computer-based automatic diagnosis methods, especially the ones using Transmission Electron Microscopy (TEM) images, have proven effective in instant virus identification. Using TEM images collected from a recent dataset, this article proposes a deep learning-based classification model to identify the virus type within those images. The methodology of this study includes two coherent image processing techniques to reduce the noise present in raw microscopy images and a functional Convolutional Neural Network (CNN) model for classification. Experimental results show that it can differentiate among 14 types of viruses with a maximum of 97.44% classification accuracy and F 1 -score, which asserts the effectiveness and reliability of the proposed method. Implementing this scheme will impart a fast and dependable virus identification scheme subsidiary to the thorough diagnostic procedures., Competing Interests: Declarations. Competing interests: The authors declare no competing interests., (© 2024. The Author(s).)

Synovial sarcoma (SS) is a rare cancer that forms in soft tissues around joints, and early detection is crucial for improving patient survival rates. This study introduces a convolutional neural network (CNN) using an improved AlexNet deep learning classifier to improve SS diagnosis from digital pathological images. Key preprocessing steps, such as dataset augmentation and noise reduction techniques, such as adaptive median filtering (AMF) and histogram equalization were employed to improve image quality. Feature extraction was conducted using the Gray-Level Co-occurrence Matrix (GLCM) and Improved Linear Discriminant Analysis (ILDA), while image segmentation targeted spindle-shaped cells using repetitive phase-level set segmentation (RPLSS). The improved AlexNet architecture features additional convolutional layers and resized input images, leading to superior performance. The model demonstrated significant improvements in accuracy, sensitivity, specificity, and AUC, outperforming existing methods by 3%, 1.70%, 6.08%, and 8.86%, respectively, in predicting SS., (© 2024 IOP Publishing Ltd. All rights, including for text and data mining, AI training, and similar technologies, are reserved.)

Objectives: Artificial intelligence (AI) could be used as an automatically diagnosis method for dental disease due to its accuracy and efficiency. This research proposed a novel convolutional neural network (CNN)-based deep learning (DL) ensemble model for tooth position detection, tooth outline segmentation, tooth tissue segmentation, periodontal bone loss and periodontitis stage prediction using dental panoramic radiographs., Methods: The dental panoramic radiographs of 320 patients during the period January 2020 to December 2023 were collected in our dataset. All images were de-identified without private information. In total, 8462 teeth were included. The algorithms that DL ensemble model adopted include YOLOv8, Mask R-CNN, and TransUNet. The prediction results of DL method were compared with diagnosis of periodontists., Results: The periodontal bone loss degree deviation between the DL method and ground truth drawn by the three periodontists was 5.28%. The overall PCC value of the DL method and the periodontists' diagnoses was 0.832 (P <​ 0.001). ​The ICC value was 0.806 (P <​ 0.001). The total diagnostic accuracy of the DL method was 89.45%., Conclusions: The proposed DL ensemble model in this study shows high accuracy and efficiency in radiographic detection and a valuable adjunct to periodontal diagnosis. The method has strong potential to enhance clinical professional performance and build more efficient dental health services., Clinical Significance: The DL method not only could help dentists for rapid and accurate auxiliary diagnosis and prevent medical negligence, but also could be used as a useful learning resource for inexperienced dentists and dental students., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Purpose To guide the attention of a deep learning (DL) model toward MRI characteristics of brain lesions by incorporating radiology report-derived textual features to achieve interpretable lesion detection. Materials and Methods In this retrospective study, 35 282 brain MRI scans (January 2018 to June 2023) and corresponding radiology reports from center 1 were used for training, validation, and internal testing. A total of 2655 brain MRI scans (January 2022 to December 2022) from centers 2-5 were reserved for external testing. Textual features were extracted from radiology reports to guide a DL model (ReportGuidedNet) focusing on lesion characteristics. Another DL model (PlainNet) without textual features was developed for comparative analysis. Both models identified 15 conditions, including 14 diseases and normal brains. Performance of each model was assessed by calculating macro-averaged area under the receiver operating characteristic curve (ma-AUC) and micro-averaged AUC (mi-AUC). Attention maps, which visualized model attention, were assessed with a five-point Likert scale. Results ReportGuidedNet outperformed PlainNet for all diagnoses on both internal (ma-AUC, 0.93 [95% CI: 0.91, 0.95] vs 0.85 [95% CI: 0.81, 0.88]; mi-AUC, 0.93 [95% CI: 0.90, 0.95] vs 0.89 [95% CI: 0.83, 0.92]) and external (ma-AUC, 0.91 [95% CI: 0.88, 0.93] vs 0.75 [95% CI: 0.72, 0.79]; mi-AUC, 0.90 [95% CI: 0.87, 0.92] vs 0.76 [95% CI: 0.72, 0.80]) testing sets. The performance difference between internal and external testing sets was smaller for ReportGuidedNet than for PlainNet (Δma-AUC, 0.03 vs 0.10; Δmi-AUC, 0.02 vs 0.13). The Likert scale score of ReportGuidedNet was higher than that of PlainNet (mean ± SD: 2.50 ± 1.09 vs 1.32 ± 1.20; P < .001). Conclusion The integration of radiology report textual features improved the ability of the DL model to detect brain lesions, thereby enhancing interpretability and generalizability. Keywords: Deep Learning, Computer-aided Diagnosis, Knowledge-driven Model, Radiology Report, Brain MRI Supplemental material is available for this article. Published under a CC BY 4.0 license.

Automatic image-based severity estimation is an important task in computer-aided diagnosis. Severity estimation by deep learning requires a large amount of training data to achieve a high performance. In general, severity estimation uses training data annotated with discrete (i.e., quantized) severity labels. Annotating discrete labels is often difficult in images with ambiguous severity, and the annotation cost is high. In contrast, relative annotation, in which the severity between a pair of images is compared, can avoid quantizing severity and thus makes it easier. We can estimate relative disease severity using a learning-to-rank framework with relative annotations, but relative annotation has the problem of the enormous number of pairs that can be annotated. Therefore, the selection of appropriate pairs is essential for relative annotation. In this paper, we propose a deep Bayesian active learning-to-rank that automatically selects appropriate pairs for relative annotation. Our method preferentially annotates unlabeled pairs with high learning efficiency from the model uncertainty of the samples. We prove the theoretical basis for adapting Bayesian neural networks to pairwise learning-to-rank and demonstrate the efficiency of our method through experiments on endoscopic images of ulcerative colitis on both private and public datasets. We also show that our method achieves a high performance under conditions of significant class imbalance because it automatically selects samples from the minority classes., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier B.V. All rights reserved.)

Lung cancer (LC) is a life-threatening and dangerous disease all over the world. However, earlier diagnoses and treatment can save lives. Earlier diagnoses of malevolent cells in the lungs responsible for oxygenating the human body and expelling carbon dioxide due to significant procedures are critical. Even though a computed tomography (CT) scan is the best imaging approach in the healthcare sector, it is challenging for physicians to identify and interpret the tumour from CT scans. LC diagnosis in CT scan using artificial intelligence (AI) can help radiologists in earlier diagnoses, enhance performance, and decrease false negatives. Deep learning (DL) for detecting lymph node contribution on histopathological slides has become popular due to its great significance in patient diagnoses and treatment. This study introduces a computer-aided diagnosis for LC by utilizing the Waterwheel Plant Algorithm with DL (CADLC-WWPADL) approach. The primary aim of the CADLC-WWPADL approach is to classify and identify the existence of LC on CT scans. The CADLC-WWPADL method uses a lightweight MobileNet model for feature extraction. Besides, the CADLC-WWPADL method employs WWPA for the hyperparameter tuning process. Furthermore, the symmetrical autoencoder (SAE) model is utilized for classification. An investigational evaluation is performed to demonstrate the significant detection outputs of the CADLC-WWPADL technique. An extensive comparative study reported that the CADLC-WWPADL technique effectively performs with other models with a maximum accuracy of 99.05% under the benchmark CT image dataset., (© 2024. The Author(s).)

Lung cancer screening (LCS) using annual computed tomography (CT) scanning significantly reduces mortality by detecting cancerous lung nodules at an earlier stage. Deep learning algorithms can improve nodule malignancy risk stratification. However, they have typically been used to analyse single time point CT data when detecting malignant nodules on either baseline or incident CT LCS rounds. Deep learning algorithms have the greatest value in two aspects. These approaches have great potential in assessing nodule change across time-series CT scans where subtle changes may be challenging to identify using the human eye alone. Moreover, they could be targeted to detect nodules developing on incident screening rounds, where cancers are generally smaller and more challenging to detect confidently. Here, we show the performance of our Deep learning-based Computer-Aided Diagnosis model integrating Nodule and Lung imaging data with clinical Metadata Longitudinally (DeepCAD-NLM-L) for malignancy prediction. DeepCAD-NLM-L showed improved performance (AUC = 88%) against models utilizing single time-point data alone. DeepCAD-NLM-L also demonstrated comparable and complementary performance to radiologists when interpreting the most challenging nodules typically found in LCS programs. It also demonstrated similar performance to radiologists when assessed on out-of-distribution imaging dataset. The results emphasize the advantages of using time-series and multimodal analyses when interpreting malignancy risk in LCS., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Shahab Aslani reports financial support was provided by Cancer Research UK. Joseph Jacob reports financial support was provided by Wellcome Trust Clinical Research Career Development Fellowship. This work was supported by Cancer Research UK (C68622/A29390). The authors thank the National Cancer Institute for access to NCI’s data collected by National Lung Screening Trial (NLST). The statement contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI. J.J and this research was supported by Wellcome Trust Clinical Research Career Development Fellowship 209553/Z/17/Z. For the purpose of open access, the author has applied a CC-BY public copyright licence to any author accepted manuscript version arising from this submission. This project, J.J, E.G, SH.A, SM.J, A.N and DC.A were also supported by the NIHR UCLH Biomedical Research Centre, UK. The SUMMIT Study is funded by GRAIL LLC. through a research grant awarded to SM.J as Principal Investigator. SM.J is supported by CRUK programme grant (EDDCPGM/100002), and MRC Programme grant (MR/W025051/1). SM.J receives support from the CRUK Lung Cancer Centre of Excellence (C11496/ A30025) and the CRUK City of London Centre, the Rosetrees Trust, the Roy Castle Lung Cancer foundation, the Longfonds BREATH Consortia, MRC UKRMP2 Consortia, the Garfield Weston Trust and University College London Hospitals Charitable Foundation. SM.J’s work is supported by a Stand Up To Cancer-LUNGevity- American Lung Association Lung Cancer Interception Dream Team Translational Research Grant and Johnson and Johnson (grant number: SU2C-AACR-DT23-17 to S.M. Dubinett and A.E. Spira). Stand Up To Cancer is a division of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the Scientific Partner of SU2C.This work was partly undertaken at UCLH/UCL who received a proportion of funding from the Department of Health’s NIHR Biomedical Research Centre’s funding scheme. J.J reports consultancy fees from Boehringer Ingelheim, F. Hoffmann-La Roche, GlaxoSmithKline and NHSX. J.J reports advisory boards in Boehringer Ingelheim and F. Hoffmann-La Roche. J.J reports lecture fees from Boehringer Ingelheim, F. Hoffmann-La Roche, and Takeda. J.J reports grant Funding from GlaxoSmithKline, Wellcome Trust, Microsoft Research and Gilead Sciences. J.J reports patents UK patent application numbers 2113765.8 and GB2211487.0. J.J was supported by Wellcome Trust Clinical Research Career Development Fellowship 209553/Z/17/Z. SM.J has received fees for advisory board membership in the last three years from Bard1 Lifescience. SM.J has received grant income from GRAIL Inc. SM.J is an unpaid member of a GRAIL advisory board. SM.J has received lecture fees for academic meetings from Cheisi and Astra Zeneca. SM.J’s wife works for Astra Zeneca. SM.J reports fees from Astra-Zeneca, Bard1 Bioscience, Achilles Therapeutics, and Jansen unrelated to the submitted work. SM.J received assistance for travel to meetings from Astra Zeneca to American Thoracic Conference 2018 and from Takeda to World Conference Lung Cancer 2019 and is the Investigator Lead on grants from GRAIL Inc, GlaxoSmithKline plc and Owlstone. A.N reports advisory fees from Part-funded by UCLH Biomedical Research Centre (BRC), National Institute for Health Research (NIHR); Member of Advisory Board, Aidence, Artificial Intelligence BV; Co-Investigator, Integration and Analysis of Data Using Artificial Intelligence to Improve Patent Outcomes with Thoracic Diseases (DART) study; Scientific Advisory Board, iDx Lung Trial; Collaborator/Advisory fees Merck Sharp and Dohme (MSD) (UK) Limited; Speaker Fees, Astra Zeneca (AZ) UK Ltd. A.D’s disclosures are fees from Boehringer Ingelheim, Roche, Brainomix, and Vicore. P.A and R.C are the founder of ManasAI, an AI SaaS company specializing in predictive models. P.A and R.C. own equity in ManasAI. SH.A, E.G and DC.A is supported by the National Institute for Health Research University College London Hospitals Biomedical Research Centre. E.CH, J.M and C.H have no competing interests to declare. The remainder of the SUMMIT consortium declare no competing interest with regard to the current manuscript., (Copyright © 2024 The Authors. Published by Elsevier Ltd.. All rights reserved.)

This study aims to develop a super-resolution (SR) algorithm tailored specifically for enhancing the image quality and resolution of early cervical cancer (CC) magnetic resonance imaging (MRI) images. The proposed method is subjected to both qualitative and quantitative analyses, thoroughly investigating its performance across various upscaling factors and assessing its impact on medical image segmentation tasks. The innovative SR algorithm employed for reconstructing early CC MRI images integrates complex architectures and deep convolutional kernels. Training is conducted on matched pairs of input images through a multi-input model. The research findings highlight the significant advantages of the proposed SR method on two distinct datasets at different upscaling factors. Specifically, at a 2× upscaling factor, the sagittal test set outperforms the state-of-the-art methods in the PSNR index evaluation, second only to the hybrid attention transformer, while the axial test set outperforms the state-of-the-art methods in both PSNR and SSIM index evaluation. At a 4× upscaling factor, both the sagittal test set and the axial test set achieve the best results in the evaluation of PNSR and SSIM indicators. This method not only effectively enhances image quality, but also exhibits superior performance in medical segmentation tasks, thereby providing a more reliable foundation for clinical diagnosis and image analysis., (© 2024. The Author(s).)

Convolutional Neural Networks (CNNs) have received substantial attention as a highly effective tool for analyzing medical images, notably in interpreting endoscopic images, due to their capacity to provide results equivalent to or exceeding those of medical specialists. This capability is particularly crucial in the realm of gastrointestinal disorders, where even experienced gastroenterologists find the automatic diagnosis of such conditions using endoscopic pictures to be a challenging endeavor. Currently, gastrointestinal findings in medical diagnosis are primarily determined by manual inspection by competent gastrointestinal endoscopists. This evaluation procedure is labor-intensive, time-consuming, and frequently results in high variability between laboratories. To address these challenges, we introduced a specialized CNN-based architecture called GastroFuse-Net, designed to recognize human gastrointestinal diseases from endoscopic images. GastroFuse-Net was developed by combining features extracted from two different CNN models with different numbers of layers, integrating shallow and deep representations to capture diverse aspects of the abnormalities. The Kvasir dataset was used to thoroughly test the proposed deep learning model. This dataset contained images that were classified according to structures (cecum, z-line, pylorus), diseases (ulcerative colitis, esophagitis, polyps), or surgical operations (dyed resection margins, dyed lifted polyps). The proposed model was evaluated using various measures, including specificity, recall, precision, F1-score, Mathew's Correlation Coefficient (MCC), and accuracy. The proposed model GastroFuse-Net exhibited exceptional performance, achieving a precision of 0.985, recall of 0.985, specificity of 0.984, F1-score of 0.997, MCC of 0.982, and an accuracy of 98.5%.

Background: Transcranial sonography (TCS) plays a crucial role in diagnosing Parkinson's disease. However, the intricate nature of TCS pathological features, the lack of consistent diagnostic criteria, and the dependence on physicians' expertise can hinder accurate diagnosis. Current TCS-based diagnostic methods, which rely on machine learning, often involve complex feature engineering and may struggle to capture deep image features. While deep learning offers advantages in image processing, it has not been tailored to address specific TCS and movement disorder considerations. Consequently, there is a scarcity of research on deep learning algorithms for TCS-based PD diagnosis., Methods: This study introduces a deep learning residual network model, augmented with attention mechanisms and multi-scale feature extraction, termed AMSNet, to assist in accurate diagnosis. Initially, a multi-scale feature extraction module is implemented to robustly handle the irregular morphological features and significant area information present in TCS images. This module effectively mitigates the effects of artifacts and noise. When combined with a convolutional attention module, it enhances the model's ability to learn features of lesion areas. Subsequently, a residual network architecture, integrated with channel attention, is utilized to capture hierarchical and detailed textures within the images, further enhancing the model's feature representation capabilities., Results: The study compiled TCS images and personal data from 1109 participants. Experiments conducted on this dataset demonstrated that AMSNet achieved remarkable classification accuracy (92.79%), precision (95.42%), and specificity (93.1%). It surpassed the performance of previously employed machine learning algorithms in this domain, as well as current general-purpose deep learning models., Conclusion: The AMSNet proposed in this study deviates from traditional machine learning approaches that necessitate intricate feature engineering. It is capable of automatically extracting and learning deep pathological features, and has the capacity to comprehend and articulate complex data. This underscores the substantial potential of deep learning methods in the application of TCS images for the diagnosis of movement disorders., (© 2024. The Author(s).)

The bone marrow overproduces immature cells in the malignancy known as Acute Lymphoblastic Leukemia (ALL). In the United States, about 6500 occurrences of ALL are diagnosed each year in both children and adults, comprising nearly 25% of pediatric cancer cases. Recently, many computer-assisted diagnosis (CAD) systems have been proposed to aid hematologists in reducing workload, providing correct results, and managing enormous volumes of data. Traditional CAD systems rely on hematologists' expertise, specialized features, and subject knowledge. Utilizing early detection of ALL can aid radiologists and doctors in making medical decisions. In this study, Deep Dilated Residual Convolutional Neural Network (DDRNet) is presented for the classification of blood cell images, focusing on eosinophils, lymphocytes, monocytes, and neutrophils. To tackle challenges like vanishing gradients and enhance feature extraction, the model incorporates Deep Residual Dilated Blocks (DRDB) for faster convergence. Conventional residual blocks are strategically placed between layers to preserve original information and extract general feature maps. Global and Local Feature Enhancement Blocks (GLFEB) balance weak contributions from shallow layers for improved feature normalization. The global feature from the initial convolution layer, when combined with GLFEB-processed features, reinforces classification representations. The Tanh function introduces non-linearity. A Channel and Spatial Attention Block (CSAB) is integrated into the neural network to emphasize or minimize specific feature channels, while fully connected layers transform the data. The use of a sigmoid activation function concentrates on relevant features for multiclass lymphoblastic leukemia classification The model was analyzed with Kaggle dataset (16,249 images) categorized into four classes, with a training and testing ratio of 80:20. Experimental results showed that DRDB, GLFEB and CSAB blocks' feature discrimination ability boosted the DDRNet model F1 score to 0.96 with minimal computational complexity and optimum classification accuracy of 99.86% and 91.98% for training and testing data. The DDRNet model stands out from existing methods due to its high testing accuracy of 91.98%, F1 score of 0.96, minimal computational complexity, and enhanced feature discrimination ability. The strategic combination of these blocks (DRDB, GLFEB, and CSAB) are designed to address specific challenges in the classification process, leading to improved discrimination of features crucial for accurate multi-class blood cell image identification. Their effective integration within the model contributes to the superior performance of DDRNet., (© 2024. The Author(s).)

Objectives: Rheumatoid arthritis (RA) is a severe and common autoimmune disease. Conventional diagnostic methods are often subjective, error-prone, and repetitive works. There is an urgent need for a method to detect RA accurately. Therefore, this study aims to develop an automatic diagnostic system based on deep learning for recognizing and staging RA from radiographs to assist physicians in diagnosing RA quickly and accurately., Methods: We develop a CNN-based fully automated RA diagnostic model, exploring five popular CNN architectures on two clinical applications. The model is trained on a radiograph dataset containing 240 hand radiographs, of which 39 are normal and 201 are RA with five stages. For evaluation, we use 104 hand radiographs, of which 13 are normal and 91 RA with five stages., Results: The CNN model achieves good performance in RA diagnosis based on hand radiographs. For the RA recognition, all models achieve an AUC above 90% with a sensitivity over 98%. In particular, the AUC of the GoogLeNet-based model is 97.80%, and the sensitivity is 100.0%. For the RA staging, all models achieve over 77% AUC with a sensitivity over 80%. Specifically, the VGG16-based model achieves 83.36% AUC with 92.67% sensitivity., Conclusion: The presented GoogLeNet-based model and VGG16-based model have the best AUC and sensitivity for RA recognition and staging, respectively. The experimental results demonstrate the feasibility and applicability of CNN in radiograph-based RA diagnosis. Therefore, this model has important clinical significance, especially for resource-limited areas and inexperienced physicians., (© 2024. The Author(s).)

Multimodal medical imaging plays a pivotal role in clinical diagnosis and research, as it combines information from various imaging modalities to provide a more comprehensive understanding of the underlying pathology. Recently, deep learning-based multimodal fusion techniques have emerged as powerful tools for improving medical image classification. This review offers a thorough analysis of the developments in deep learning-based multimodal fusion for medical classification tasks. We explore the complementary relationships among prevalent clinical modalities and outline three main fusion schemes for multimodal classification networks: input fusion, intermediate fusion (encompassing single-level fusion, hierarchical fusion, and attention-based fusion), and output fusion. By evaluating the performance of these fusion techniques, we provide insight into the suitability of different network architectures for various multimodal fusion scenarios and application domains. Furthermore, we delve into challenges related to network architecture selection, handling incomplete multimodal data management, and the potential limitations of multimodal fusion. Finally, we spotlight the promising future of Transformer-based multimodal fusion techniques and give recommendations for future research in this rapidly evolving field., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Ramin Tadayoni, reports financial support was provided by French National Research Agency. Gwenolé Quellec reports financial support was provided by French National Research Agency., (Copyright © 2024 The Author(s). Published by Elsevier Ltd.. All rights reserved.)

Background: Analysis of structures contained in tissue samples and the relevant contextual information is of utmost importance to histopathologists during diagnosis. Cardiac biopsies require in-depth analysis of the relationships between biological structures. Statistical measures are insufficient for determining a model's viability and applicability in the diagnostic process. A deeper understanding of predictions is necessary in order to support histopathologists., Methods: We propose a method for providing supporting information in the form of segmentation of histological structures to histopathologists based on these principles. The proposed method utilizes nuclei type and density information in addition to standard image input provided at two different zoom levels for the semantic segmentation of blood vessels, inflammation, and endocardium in heart tissue., Results: The proposed method was able to reach state-of-the-art segmentation results. The overall quality and viability of the predictions was qualitatively evaluated by two pathologists and a histotechnologist., Conclusions: The decision process of the proposed deep learning model utilizes the provided information sources correctly and simulates the decision process of histopathologists via the usage of a custom-designed attention gate that provides a combination of spatial and encoder attention mechanisms. The implementation is available at https://github.com/mathali/IEDL-segmentation-of-heart-tissue., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024. Published by Elsevier Ltd.)

Drowning diagnosis is a complicated process in the autopsy, even with the assistance of autopsy imaging and the on-site information from where the body was found. Previous studies have developed well-performed deep learning (DL) models for drowning diagnosis. However, the validity of the DL models was not assessed, raising doubts about whether the learned features accurately represented the medical findings observed by human experts. In this paper, we assessed the medical validity of DL models that had achieved high classification performance for drowning diagnosis. This retrospective study included autopsy cases aged 8-91 years who underwent postmortem computed tomography between 2012 and 2021 (153 drowning and 160 non-drowning cases). We first trained three deep learning models from a previous work and generated saliency maps that highlight important features in the input. To assess the validity of models, pixel-level annotations were created by four radiological technologists and further quantitatively compared with the saliency maps. All the three models demonstrated high classification performance with areas under the receiver operating characteristic curves of 0.94, 0.97, and 0.98, respectively. On the other hand, the assessment results revealed unexpected inconsistency between annotations and models' saliency maps. In fact, each model had, respectively, around 30%, 40%, and 80% of irrelevant areas in the saliency maps, suggesting the predictions of the DL models might be unreliable. The result alerts us in the careful assessment of DL tools, even those with high classification performance., (© 2024. The Author(s).)

Objective(s): Chronic endometritis (CE) is a localized mucosal inflammatory disorder associated with female infertility of unknown etiology, endometriosis, tubal factors, repeated implantation failure, and recurrent pregnancy loss, along with atypical uterine bleeding and iron deficiency anemia. Diagnosis of CE has traditionally relied on endometrial biopsy and detection of CD138(+) endometrial stromal plasmacytes. To develop a less invasive diagnostic system for CE, we aimed to construct a deep learning-based convolutional neural network (CNN) model for the automatic detection of endometrial micropolyps (EMiP), a fluid hysteroscopy (F-HSC) finding recognized as tiny protrusive lesions that are closely related to this disease., Study Design: This is an in silico study using archival images of F-HSC performed at an infertility center in a private clinic. A total of 244 infertile women undergoing F-HSC on the days 6-12 of the menstrual cycle between April 2019 and December 2021 with histopathologically-confirmed CE with the aid of immunohistochemistry for CD138 were utilized., Results: The archival F-HSC images of 208 women (78 with EMiP and 130 without EMiP) who met the inclusion criteria were finally subjected to analysis. Following preprocessing of the images, half a set was input into a CNN architecture for training, whereas the remaining images were utilized as the test set to evaluate the performance of the model, which was compared with that of the experienced gynecologists. The sensitivity, specificity, accuracy, precision, and F1-score of the CNN model-aided diagnosis were 93.6 %, 92.3 %, 92.8 %, 88.0 %, and 0.907, respectively. The area under the receiver operating characteristic curves of the CNN model-aided diagnosis (0.930) was at a similar level (p > .05) to the value of conventional diagnosis by three experienced gynecologists (0.927, 0.948, and 0.906)., Conclusion: These findings indicate that our deep learning-based CNN is capable of recognizing EMiP in F-HSC images and holds promise for further development of the computer-aided diagnostic system for CE., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier B.V. All rights reserved.)

Objective . During deep-learning-aided (DL-aided) ultrasound (US) diagnosis, US image classification is a foundational task. Due to the existence of serious speckle noise in US images, the performance of DL models may be degraded. Pre-denoising US images before their use in DL models is usually a logical choice. However, our investigation suggests that pre-speckle-denoising is not consistently advantageous. Furthermore, due to the decoupling of speckle denoising from the subsequent DL classification, investing intensive time in parameter tuning is inevitable to attain the optimal denoising parameters for various datasets and DL models. Pre-denoising will also add extra complexity to the classification task and make it no longer end-to-end. Approach . In this work, we propose a multi-scale high-frequency-based feature augmentation (MSHFFA) module that couples feature augmentation and speckle noise suppression with specific DL models, preserving an end-to-end fashion. In MSHFFA, the input US image is first decomposed to multi-scale low-frequency and high-frequency components (LFC and HFC) with discrete wavelet transform. Then, multi-scale augmentation maps are obtained by computing the correlation between LFC and HFC. Last, the original DL model features are augmented with multi-scale augmentation maps. Main results . On two public US datasets, all six renowned DL models exhibited enhanced F1-scores compared with their original versions (by 1.31%-8.17% on the POCUS dataset and 0.46%-3.89% on the BLU dataset) after using the MSHFFA module, with only approximately 1% increase in model parameter count. Significance . The proposed MSHFFA has broad applicability and commendable efficiency and thus can be used to enhance the performance of DL-aided US diagnosis. The codes are available athttps://github.com/ResonWang/MSHFFA., (© 2024 Institute of Physics and Engineering in Medicine.)

Diagnosing and screening for diabetic retinopathy is a well-known issue in the biomedical field. A component of computer-aided diagnosis that has advanced significantly over the past few years as a result of the development and effectiveness of deep learning is the use of medical imagery from a patient's eye to identify the damage caused to blood vessels. Issues with unbalanced datasets, incorrect annotations, a lack of sample images, and improper performance evaluation measures have negatively impacted the performance of deep learning models. Using three benchmark datasets of diabetic retinopathy, we conducted a detailed comparison study comparing various state-of-the-art approaches to address the effect caused by class imbalance, with precision scores of 93%, 89%, 81%, 76%, and 96%, respectively, for normal, mild, moderate, severe, and DR phases. The analyses of the hybrid modeling, including CNN analysis and SHAP model derivation results, are compared at the end of the paper, and ideal hybrid modeling strategies for deep learning classification models for automated DR detection are identified.

Objective: Ultrasound (US) examination has unique advantages in diagnosing carpal tunnel syndrome (CTS), although identification of the median nerve (MN) and diagnosis of CTS depend heavily on the expertise of examiners. In the aim of alleviating this problem, we developed a one-stop automated CTS diagnosis system (OSA-CTSD) and evaluated its effectiveness as a computer-aided diagnostic tool., Methods: We combined real-time MN delineation, accurate biometric measurements and explainable CTS diagnosis into a unified framework, called OSA-CTSD. We then collected a total of 32,301 static images from US videos of 90 normal wrists and 40 CTS wrists for evaluation using a simplified scanning protocol., Results: The proposed model exhibited better segmentation and measurement performance than competing methods, with a Hausdorff distance (95th percentile) score of 7.21 px, average symmetric surface distance score of 2.64 px, Dice score of 85.78% and intersection over union score of 76.00%. In the reader study, it exhibited performance comparable to the average performance of experienced radiologists in classifying CTS and outperformed inexperienced radiologists in terms of classification metrics (e.g., accuracy score 3.59% higher and F1 score 5.85% higher)., Conclusion: Diagnostic performance of the OSA-CTSD was promising, with the advantages of real-time delineation, automation and clinical interpretability. The application of such a tool not only reduces reliance on the expertise of examiners but also can help to promote future standardization of the CTS diagnostic process, benefiting both patients and radiologists., Competing Interests: Conflict of interest The authors declare no competing interests., (Copyright © 2023. Published by Elsevier Inc.)

Introduction: In developing Computer-Aided Diagnosis (CAD), a Convolutional Neural Network (CNN) has been commonly used as a Deep Learning (DL) model. Although it is still early, DL has excellent potential in implementing computers in medical diagnosis., Methods: This study reviews the use of DL for Anterior Cruciate Ligament (ACL) tear diagnosis. A comprehensive search was performed in PubMed, Embase, and Web of Science databases from 2018 to 2024. The included study criteria used MRI images to evaluate ACL tears, and the diagnosis of ACL tears was performed using the DL model. We summarized the paper by reporting their model accuracy, model comparison with arthroscopy, and explainable., Results: AI implementation in tabular format; we conclude that many medical professionals believe that arthroscopic diagnosis is the most reliable method for diagnosing ACL tears. However, due to its intrusive treatment, CAD is projected to be able to produce similar outcomes from MRI scan results. To gain the trust of physicians and meet the demand for reliable knee injury detection systems, an algorithm for CAD should also meet several criteria, such as being transparent, interpretable, explainable, and easy to use. Therefore, future works should consider creating an Explainable DL model for ACL tear diagnosis. It is also essential to evaluate the performance of this Explainable DL model compared to the gold standard of arthroscopy diagnosis., Conclusion: There are issues regarding the need for Explainable DL in CAD to increase confidence in its result while also highlighting the importance of the involvement of medical practitioners in system design. There is no funding for this work.

Background: Accurate diagnosis and subsequent delineated treatment planning require the experience of clinicians in the handling of their case numbers. However, applying deep learning in image processing is useful in creating tools that promise faster high-quality diagnoses, but the accuracy and precision of 3-D image processing from 2-D data may be limited by factors such as superposition of organs, distortion and magnification, and detection of new pathologies. The purpose of this research is to use radiomics and deep learning to develop a tool for lung cancer diagnosis., Methods: This study applies radiomics and deep learning in the diagnosis of lung cancer to help clinicians accurately analyze the images and thereby provide the appropriate treatment planning. 86 patients were recruited from Bach Mai Hospital, and 1012 patients were collected from an open-source database. First, deep learning has been applied in the process of segmentation by U-NET and cancer classification via the use of the DenseNet model. Second, the radiomics were applied for measuring and calculating diameter, surface area, and volume. Finally, the hardware also was designed by connecting between Arduino Nano and MFRC522 module for reading data from the tag. In addition, the displayed interface was created on a web platform using Python through Streamlit., Results: The applied segmentation model yielded a validation loss of 0.498, a train loss of 0.27, a cancer classification validation loss of 0.78, and a training accuracy of 0.98. The outcomes of the diagnostic capabilities of lung cancer (recognition and classification of lung cancer from chest CT scans) were quite successful., Conclusions: The model provided means for storing and updating patients' data directly on the interface which allowed the results to be readily available for the health care providers. The developed system will improve clinical communication and information exchange. Moreover, it can manage efforts by generating correlated and coherent summaries of cancer diagnoses.

Objective: Normal interictal [ 18 F]FDG-PET can be predicted from the corresponding T1w MRI with Generative Adversarial Networks (GANs). A technique we call SIPCOM (Subtraction Interictal PET Co-registered to MRI) can then be used to compare epilepsy patients' predicted and clinical PET. We assessed the ability of SIPCOM to identify the Resection Zone (RZ) in patients with drug-resistant epilepsy (DRE) with reference to visual and statistical parametric mapping (SPM) analysis., Methods: Patients with complete presurgical work-up and subsequent SEEG and cortectomy were included. RZ localisation, the reference region, was assigned to one of eighteen anatomical brain regions. SIPCOM was implemented using healthy controls to train a GAN. To compare, the clinical PET coregistered to MRI was visually assessed by two trained readers, and a standard SPM analysis was performed., Results: Twenty patients aged 17-50 (32 ± 7.8) years were included, 14 (70%) with temporal lobe epilepsy (TLE). Eight (40%) were MRI-negative. After surgery, 14 patients (70%) had a good outcome (Engel I-II). RZ localisation rate was 60% with SIPCOM vs 35% using SPM (P = 0.015) and vs 85% using visual analysis (P = 0.54). Results were similar for Engel I-II patients, the RZ localisation rate was 64% with SIPCOM vs 36% with SPM. With SIPCOM localisation was correct in 67% in MRI-positive vs 50% in MRI-negative patients, and 64% in TLE vs 43% in extra-TLE. The average number of false-positive clusters was 2.2 ± 1.3 using SIPCOM vs 2.3 ± 3.1 using SPM. All RZs localized with SPM were correctly localized with SIPCOM. In one case, PET and MRI were visually reported as negative, but both SIPCOM and SPM localized the RZ., Significance: SIPCOM performed better than the reference computer-assisted method (SPM) for RZ detection in a group of operated DRE patients. SIPCOM's impact on epilepsy management needs to be prospectively validated., (© 2023 The Authors. Epilepsia Open published by Wiley Periodicals LLC on behalf of International League Against Epilepsy.)

Aim: Although many studies in various fields employ deep learning models, only a few such studies exist in dental imaging. The present article aims to evaluate the effectiveness of convolutional neural network (CNN) algorithms for the detection and diagnosis of the quantitative level of dental restorations using panoramic radiographs by preparing a novel dataset., Materials and Methods: 20,973 panoramic radiographs were used, all labeled into five distinct categories by three dental experts. AlexNet, VGG-16, and variants of ResNet models were trained with the dataset and evaluated for the classification task. Additionally, 10-fold cross-validation (ie, 9 folds were separated for training and 1 fold for validation) and data augmentation were carried out for all experiments., Results: The most successful result was shown by ResNet-101, with an accuracy of 92.7%. Its macro-average AUC was also the highest, at 0.989. Other accuracy results obtained for the dataset were 75.5% for AlexNet, 85.0% for VGG-16, 92.1% for ResNet-18, 91.7% for ResNet-50, and 92.1% for InceptionResNet-v2., Conclusions: An accuracy of 92.7% is a very promising result for a computer-aided diagnostic system. This result proved that the system could assist dentists in providing supportive preliminary information from the moment a patient's first panoramic radiograph is taken. Furthermore, as the introduced dataset is powerful enough, it can be relabeled for different problems and used in different studies.

Microscopic evaluation of hematoxylin and eosin-stained slides is still the diagnostic gold standard for a variety of diseases, including neoplasms. Nevertheless, intra- and interrater variability are well documented among pathologists. So far, computer assistance via automated image analysis has shown potential to support pathologists in improving accuracy and reproducibility of quantitative tasks. In this proof of principle study, we describe a machine-learning-based algorithm for the automated diagnosis of 7 of the most common canine skin tumors: trichoblastoma, squamous cell carcinoma, peripheral nerve sheath tumor, melanoma, histiocytoma, mast cell tumor, and plasmacytoma. We selected, digitized, and annotated 350 hematoxylin and eosin-stained slides (50 per tumor type) to create a database divided into training, n = 245 whole-slide images (WSIs), validation ( n = 35 WSIs), and test sets ( n = 70 WSIs). Full annotations included the 7 tumor classes and 6 normal skin structures. The data set was used to train a convolutional neural network (CNN) for the automatic segmentation of tumor and nontumor classes. Subsequently, the detected tumor regions were classified patch-wise into 1 of the 7 tumor classes. A majority of patches-approach led to a tumor classification accuracy of the network on the slide-level of 95% (133/140 WSIs), with a patch-level precision of 85%. The same 140 WSIs were provided to 6 experienced pathologists for diagnosis, who achieved a similar slide-level accuracy of 98% (137/140 correct majority votes). Our results highlight the feasibility of artificial intelligence-based methods as a support tool in diagnostic oncologic pathology with future applications in other species and tumor types., Competing Interests: Declaration of Conflicting InterestsThe author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

BACKGROUND. Computer-aided diagnosis (CAD) systems for breast ultrasound interpretation have been primarily evaluated at tertiary and/or urban medical centers by radiologists with breast ultrasound expertise. OBJECTIVE. The purpose of this study was to evaluate the usefulness of deep learning-based CAD software on the diagnostic performance of radiologists without breast ultrasound expertise at secondary or rural hospitals in the differentiation of benign and malignant breast lesions measuring up to 2.0 cm on ultrasound. METHODS. This prospective study included patients scheduled to undergo biopsy or surgical resection at any of eight participating secondary or rural hospitals in China of a breast lesion classified as BI-RADS category 3-5 on prior breast ultrasound from November 2021 to September 2022. Patients underwent an additional investigational breast ultrasound, performed and interpreted by a radiologist without breast ultrasound expertise (hybrid body/breast radiologists, either who lacked breast imaging subspecialty training or for whom the number of breast ultrasounds performed annually accounted for less than 10% of all ultrasounds performed annually by the radiologist), who assigned a BI-RADS category. CAD results were used to upgrade reader-assigned BI-RADS category 3 lesions to category 4A and to downgrade reader-assigned BI-RADS category 4A lesions to category 3. Histologic results of biopsy or resection served as the reference standard. RESULTS. The study included 313 patients (mean age, 47.0 ± 14.0 years) with 313 breast lesions (102 malignant, 211 benign). Of BI-RADS category 3 lesions, 6.0% (6/100) were upgraded by CAD to category 4A, of which 16.7% (1/6) were malignant. Of category 4A lesions, 79.1% (87/110) were downgraded by CAD to category 3, of which 4.6% (4/87) were malignant. Diagnostic performance was significantly better after application of CAD, in comparison with before application of CAD, in terms of accuracy (86.6% vs 62.6%, p < .001), specificity (82.9% vs 46.0%, p < .001), and PPV (72.7% vs 46.5%, p < .001) but not significantly different in terms of sensitivity (94.1% vs 97.1%, p = .38) or NPV (96.7% vs 97.0%, p > .99). CONCLUSION. CAD significantly improved radiologists' diagnostic performance, showing particular potential to reduce the frequency of benign breast biopsies. CLINICAL IMPACT. The findings indicate the ability of CAD to improve patient care in settings with incomplete access to breast imaging expertise.

Computer vision and deep learning have the potential to improve medical artificial intelligence (AI) by assisting in diagnosis, prediction, and prognosis. However, the application of deep learning to medical image analysis is challenging due to limited data availability and imbalanced data. While model performance is undoubtedly essential for medical image analysis, model trust is equally important. To address these challenges, we propose TRUDLMIA, a trustworthy deep learning framework for medical image analysis, which leverages image features learned through self-supervised learning and utilizes a novel surrogate loss function to build trustworthy models with optimal performance. The framework is validated on three benchmark data sets for detecting pneumonia, COVID-19, and melanoma, and the created models prove to be highly competitive, even outperforming those designed specifically for the tasks. Furthermore, we conduct ablation studies, cross-validation, and result visualization and demonstrate the contribution of proposed modules to both model performance (up to 21%) and model trust (up to 5%). We expect that the proposed framework will support researchers and clinicians in advancing the use of deep learning for dealing with public health crises, improving patient outcomes, increasing diagnostic accuracy, and enhancing the overall quality of healthcare delivery.

Purpose: Computer-aided diagnosis (CAD), which assists in the interpretation of chest radiographs, is becoming common. However, few studies have evaluated the benefits and pitfalls of CAD in the real world. This study aimed to evaluate the independent performance of commercially available deep learning-based automatic detection (DLAD) software, EIRL Chest X-ray Lung Nodule, in a cohort that included patients with background pulmonary abnormalities often encountered in clinical situations., Methods: Patients with clinically suspected lung cancer for whom chest radiography was performed within a month before or after CT scan between June 2020 and May 2022 in our institution were enrolled. The reference standard was created using a bounding box annotated by two radiologists with reference to the CT. The visibility score, characteristics, location of the pulmonary nodules, presence of overlapping structures or pulmonary disease, and background lung score were manually determined., Results: We included 388 patients. The DLAD software detected 222 of the 322 nodules visible on manual evaluation, with a sensitivity of 0.689 and a false-positive rate of 0.168. The detectability of the DLAD software was significantly lower for small and subsolid and nodules with overlapping structures. The visibility score and sensitivity of detection by the DLAD software were positively correlated. The relationship between the background lung score and detection by the DLAD software was unclear., Conclusion: The standalone performance of DLAD in detecting pulmonary nodules exhibited a sensitivity of 0.689 and a false-positive rate of 0.168. Understanding the characteristics of DLAD is crucial when interpreting chest radiographs with the assistance of the DLAD., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 Elsevier B.V. All rights reserved.)

Liver disease is a potentially asymptomatic clinical entity that may progress to patient death. This study proposes a multi-modal deep neural network for multi-class malignant liver diagnosis. In parallel with the portal venous computed tomography (CT) scans, pathology data is utilized to prognosticate primary liver cancer variants and metastasis. The processed CT scans are fed to the deep dilated convolution neural network to explore salient features. The residual connections are further added to address vanishing gradient problems. Correspondingly, five pathological features are learned using a wide and deep network that gives a benefit of memorization with generalization. The down-scaled hierarchical features from CT scan and pathology data are concatenated to pass through fully connected layers for classification between liver cancer variants. In addition, the transfer learning of pre-trained deep dilated convolution layers assists in handling insufficient and imbalanced dataset issues. The fine-tuned network can predict three-class liver cancer variants with an average accuracy of 96.06% and an Area Under Curve (AUC) of 0.832. To the best of our knowledge, this is the first study to classify liver cancer variants by integrating pathology and image data, hence following the medical perspective of malignant liver diagnosis. The comparative analysis on the benchmark dataset shows that the proposed multi-modal neural network outperformed most of the liver diagnostic studies and is comparable to others., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023. Published by Elsevier Ltd.)

Purpose: Septic arthritis is an infectious disease. Conventionally, the diagnosis of septic arthritis can only be based on the identification of causal pathogens taken from synovial fluid, synovium or blood samples. However, the cultures require several days for the isolation of pathogens. A rapid assessment performed through computer-aided diagnosis (CAD) would bring timely treatment., Methods: A total of 214 non-septic arthritis and 64 septic arthritis images generated by gray-scale (GS) and Power Doppler (PD) ultrasound modalities were collected for the experiment. A deep learning-based vision transformer (ViT) with pre-trained parameters were used for image feature extraction. The extracted features were then combined in machine learning classifiers with ten-fold cross validation in order to evaluate the abilities of septic arthritis classification., Results: Using a support vector machine, GS and PD features can achieve an accuracy rate of 86% and 91%, with the area under the receiver operating characteristic curves (AUCs) being 0.90 and 0.92, respectively. The best accuracy (92%) and best AUC (0.92) was obtained by combining both feature sets., Conclusions: This is the first CAD system based on a deep learning approach for the diagnosis of septic arthritis as seen on knee ultrasound images. Using pre-trained ViT, both the accuracy and computation costs improved more than they had through convolutional neural networks. Additionally, automatically combining GS and PD generates a higher accuracy to better assist the physician's observations, thus providing a timely evaluation of septic arthritis., Competing Interests: Declaration of Competing Interest The authors declare that they have no financial or personal relationships with any people or organizations that could have inappropriately influenced their work., (Copyright © 2023 Elsevier B.V. All rights reserved.)

Purpose: To collect a dataset with adequate laryngoscopy images and identify the appearance of vocal folds and their lesions in flexible laryngoscopy images by objective deep learning models., Methods: We adopted a number of novel deep learning models to train and classify 4549 flexible laryngoscopy images as no vocal fold, normal vocal folds, and abnormal vocal folds. This could help these models recognize vocal folds and their lesions within these images. Ultimately, we made a comparison between the results of the state-of-the-art deep learning models, and another comparison of the results between the computer-aided classification system and ENT doctors., Results: This study exhibited the performance of the deep learning models by evaluating laryngoscopy images collected from 876 patients. The efficiency of the Xception model was higher and steadier than almost the rest of the models. The accuracy of no vocal fold, normal vocal folds, and vocal fold abnormalities on this model were 98.90 %, 97.36 %, and 96.26 %, respectively. Compared to our ENT doctors, the Xception model produced better results than a junior doctor and was near an expert., Conclusion: Our results show that current deep learning models can classify vocal fold images well and effectively assist physicians in vocal fold identification and classification of normal or abnormal vocal folds., Competing Interests: Declaration of competing interest All the authors declared that they have no conflict of interest., (Copyright © 2023 Elsevier Inc. All rights reserved.)

Background and Objective: Current studies based on digital biopsy images have achieved satisfactory results in detecting colon cancer despite their limited visual spectral range. Such methods may be less accurate when applied to samples taken from the tumor margin region or to samples containing multiple diagnoses. In contrast with the traditional computer vision approach, micro-FTIR hyperspectral images quantify the tissue-light interaction on a histochemical level and characterize different tissue pathologies, as they present a unique spectral signature. Therefore, this paper investigates the possibility of using hyperspectral images acquired over micro-FTIR absorbance spectroscopy to characterize healthy, inflammatory, and tumor colon tissues., Methods: The proposed method consists of modeling hyperspectral data into a voxel format to detect the patterns of each voxel using fully connected deep neural network. A web-based computer-aided diagnosis tool for inference is also provided., Results: Our experiments were performed using the K-fold cross-validation protocol in an intrapatient approach and achieved an overall accuracy of 99% using a deep neural network and 96% using a linear support vector machine. Through the experiments, we noticed the high performance of the method in characterizing such tissues using deep learning and hyperspectral images, indicating that the infrared spectrum contains relevant information and can be used to assist pathologists during the diagnostic process., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023. Published by Elsevier B.V.)

Alzheimer's disease (AD) is the most common form of dementia. Computer-aided diagnosis (CAD) can help in the early detection of associated cognitive impairment. The aim of this work is to improve the automatic detection of dementia in MRI brain data. For this purpose, we used an established pipeline that includes the registration, slicing, and classification steps. The contribution of this research was to investigate for the first time, to our knowledge, three current and promising deep convolutional models (ResNet, DenseNet, and EfficientNet) and two transformer-based architectures (MAE and DeiT) for mapping input images to clinical diagnosis. To allow a fair comparison, the experiments were performed on two publicly available datasets (ADNI and OASIS) using multiple benchmarks obtained by changing the number of slices per subject extracted from the available 3D voxels. The experiments showed that very deep ResNet and DenseNet models performed better than the shallow ResNet and VGG versions tested in the literature. It was also found that transformer architectures, and DeiT in particular, produced the best classification results and were more robust to the noise added by increasing the number of slices. A significant improvement in accuracy (up to 7%) was achieved compared to the leading state-of-the-art approaches, paving the way for the use of CAD approaches in real-world applications.

Background: A high false-positive rate remains a technical glitch hindering the broad spectrum of application of deep-learning-based diagnostic tools in routine radiological practice from assisting in diagnosing rib fractures., Purpose: To examine the performance of two versions of deep-learning-based software tools in aiding radiologists in diagnosing rib fractures on chest computed tomography (CT) images., Material and Methods: In total, 123 patients (708 rib fractures) were included in this retrospective study. Two groups of radiologists with different experience levels retrospectively reviewed images for rib fractures in the concurrent mode aided with RibFrac-High Sensitivity (HS) and RibFrac-High Precision (HP). We compared their diagnostic performance against the reference standard in terms of sensitivity and positive predictive value (PPV)., Results: On a per-patient basis, RibFrac-HS exhibited a higher sensitivity compared with RibFrac-HP (mean difference=0.051, 95% CI=0.012-0.090; P  = 0.011), whereas the latter significantly outperformed the former in terms of the PPV (mean difference=0.273, 95% CI=0.238-0.308; P  < 0.0001). The use of RibFrac-HP significantly improved the junior and the senior groups' sensitivities respectively by 0.058 (95% CI=0.033-0.083; P  < 0.0001) and 0.058 (95% CI=0.034-0.081; P  < 0.0001), and decreased the diagnosis time by 206 s (95% CI=191-220; P  < 0.0001) and 79 s (95% CI=67-92; P  < 0.0001), respectively, when compared to no software assistance., Conclusion: The sensitivity and efficiency of radiologists in identifying rib fractures can be improved by using RibFrac-HS and/or RibFrac-HP. With an added module for false-positive suppression, RibFrac-HP maintains the sensitivity and increases the PPV in fracture detection compared to Rib-Frac-HS.

The accuracy of computed tomography angiography (CTA) image interpretation depends on the radiologist. This study aims to develop a new method for automatically detecting intracranial aneurysms from CTA images using deep learning, based on a convolutional neural network (CNN) implemented on the DeepMedic platform. Ninety CTA scans of patients with intracranial aneurysms are collected and divided into two datasets: training (80 subjects) and test (10 subjects) datasets. Subsequently, a deep learning architecture with a three-dimensional (3D) CNN model is implemented on the DeepMedic platform for the automatic segmentation and detection of intracranial aneurysms from the CTA images. The samples in the training dataset are used to train the CNN model, and those in the test dataset are used to assess the performance of the established system. Sensitivity, positive predictive value (PPV), and false positives are evaluated. The overall sensitivity and PPV of this system for detecting intracranial aneurysms from CTA images are 92.3% and 100%, respectively, and the segmentation sensitivity is 92.3%. The performance of the system in the detection of intracranial aneurysms is closely related to their size. The detection sensitivity for small intracranial aneurysms (≤ 3 mm) is 66.7%, whereas the sensitivity of detection for large (> 10 mm) and medium-sized (3-10 mm) intracranial aneurysms is 100%. The deep learning architecture with a 3D CNN model on the DeepMedic platform can reliably segment and detect intracranial aneurysms from CTA images with high sensitivity., (© 2022. The Author(s) under exclusive licence to Society for Imaging Informatics in Medicine.)

Microsatellite instability (MSI) status is an important prognostic marker for various cancers. Furthermore, because immune checkpoint inhibitors are much more effective in tumors with high level of MSI (MSI-H), MSI status is routinely tested in multiple cancer types. Therefore, many studies have tested the feasibility of deep learning (DL)-based prediction of MSI status from hematoxylin and eosin (H&E)-stained tissue slides. In the present study, we attempted a fully automated classification of MSI status in gastric cancer (GC) tissue slides. For frozen and formalin-fixed paraffin-embedded (FFPE) GC tissues from The Cancer Genome Atlas (TCGA), the areas under the curves (AUCs) for the receiver operating characteristic (ROC) curves were 0.893 and 0.902, respectively. The classifier trained with the TCGA FFPE tissues performed well on an external validation Asian FFPE cohort, with an AUC of 0.874. However, the DL-based classifier seems incompatible with cancers from different organs because morphologic features of MSI-H tissues are different. Analysis of histomorphologic features of MSI-H GC tissues suggested that MSI-H GC could largely be divided into two groups: intestinal type tumors with moderate to poor differentiation and diffuse type mucinous tumors. However, the recognizable morphologic features cannot completely explain the good performance of the DL-based classifier. These results indicate that DL could automatically learn the optimal features for discrimination of MSI status in GC tissue slides. This study demonstrated the potential of a DL-based MSI classifier as a screening tool for definitive cases., (© 2022 UICC.)

Breast Cancer (BC) is a major health issue in women of the age group above 45. Identification of BC at an earlier stage is important to reduce the mortality rate. Image-based noninvasive methods are used for early detection and for providing appropriate treatment. Computer-Aided Diagnosis (CAD) schemes can support radiologists in making correct decisions. Computational intelligence paradigms such as Machine Learning (ML) and Deep Learning (DL) have been used in the recent past in CAD systems to accelerate diagnosis. ML techniques are feature driven and require a high amount of domain expertise. However, DL approaches make decisions directly from the image. The current advancement in DL approaches for early diagnosis of BC is the motivation behind this review. This article throws light on various types of CAD approaches used in BC detection and diagnosis. A survey on DL, Transfer Learning, and DL-based CAD approaches for the diagnosis of BC is presented in detail. A comparative study on techniques, datasets, and performance metrics used in state-of-the-art literature in BC diagnosis is also summarized. The proposed work provides a review of recent advancements in DL techniques for enhancing BC diagnosis.

Objective: Accurate segmentation of the lung nodule in computed tomography images is a critical component of a computer-assisted lung cancer detection/diagnosis system. However, lung nodule segmentation is a challenging task due to the heterogeneity of nodules. This study is to develop a hybrid deep learning (H-DL) model for the segmentation of lung nodules with a wide variety of sizes, shapes, margins, and opacities., Materials and Methods: A dataset collected from Lung Image Database Consortium image collection containing 847 cases with lung nodules manually annotated by at least two radiologists with nodule diameters greater than 7 mm and less than 45 mm was randomly split into 683 training/validation and 164 independent test cases. The 50% consensus consolidation of radiologists' annotation was used as the reference standard for each nodule. We designed a new H-DL model combining two deep convolutional neural networks (DCNNs) with different structures as encoders to increase the learning capabilities for the segmentation of complex lung nodules. Leveraging the basic symmetric U-shaped architecture of U-Net, we redesigned two new U-shaped deep learning (U-DL) models that were expanded to six levels of convolutional layers. One U-DL model used a shallow DCNN structure containing 16 convolutional layers adapted from the VGG-19 as the encoder, and the other used a deep DCNN structure containing 200 layers adapted from DenseNet-201 as the encoder, while the same decoder with only one convolutional layer at each level was used in both U-DL models, and we referred to them as the shallow and deep U-DL models. Finally, an ensemble layer was used to combine the two U-DL models into the H-DL model. We compared the effectiveness of the H-DL, the shallow U-DL and the deep U-DL models by deploying them separately to the test set. The accuracy of volume segmentation for each nodule was evaluated by the 3D Dice coefficient and Jaccard index (JI) relative to the reference standard. For comparison, we calculated the median and minimum of the 3D Dice and JI over the individual radiologists who segmented each nodule, referred to as M-Dice, min-Dice, M-JI, and min-JI., Results: For the 164 test cases with 327 nodules, our H-DL model achieved an average 3D Dice coefficient of 0.750 ± 0.135 and an average JI of 0.617 ± 0.159. The radiologists' average M-Dice was 0.778 ± 0.102, and the average M-JI was 0.651 ± 0.127; both were significantly higher than those achieved by the H-DL model (p < 0.05). The radiologists' average min-Dice (0.685 ± 0.139) and the average min-JI (0.537 ± 0.153) were significantly lower than those achieved by the H-DL model (p < 0.05). The results indicated that the H-DL model approached the average performance of radiologists and was superior to the radiologist whose manual segmentation had the min-Dice and min-JI. Moreover, the average Dice and average JI achieved by the H-DL model were significantly higher than those achieved by the individual shallow U-DL model (Dice of 0.745 ± 0.139, JI of 0.611 ± 0.161; p < 0.05) or the individual deep U-DL model alone (Dice of 0.739 ± 0.145, JI of 0.604 ± 0.163; p < 0.05)., Conclusion: Our newly developed H-DL model outperformed the individual shallow or deep U-DL models. The H-DL method combining multilevel features learned by both the shallow and deep DCNNs could achieve segmentation accuracy comparable to radiologists' segmentation for nodules with wide ranges of image characteristics., (© 2022 The Authors. Medical Physics published by Wiley Periodicals LLC on behalf of American Association of Physicists in Medicine.)

Background: Intracranial aneurysms (IAs) are a life-threatening disease. Their rupture can lead to hemorrhagic stroke. Most studies applying deep learning for the detection of aneurysms are based on angiographic images. However, critical diagnostic information such as morphology and aneurysm location are not captured by deep learning algorithms and still require manual assessments., Purpose: Digital subtraction angiography (DSA) is the gold standard for aneurysm diagnosis. To facilitate the fully automatic diagnosis of aneurysms, we proposed a comprehensive system for the detection, morphology measurement, and location classification of aneurysms on three-dimensional DSA images, allowing automatic diagnosis without further human input., Methods: The system comprised three neural networks: a network for aneurysm detection, a network for morphology measurement, and a network for aneurysm location identification. A cross-scale dual-path transformer module was proposed to effectively fuse local and global information to capture aneurysms of varying sizes. A multitask learning approach was also proposed to allow an accurate localization of aneurysm neck for morphology measurement., Results: The cross-scale dual-path transformer module was shown to outperform other state-of-the-art network architectures, improving segmentation, and classification accuracy. The detection network in our system achieved an F2 score of 0.946 (recall 93%, precision 100%), better than the winning team in the Cerebral Aneurysm Detection and Analysis challenge. The measurement network achieved a relative error of less than 10% for morphology measurement, at the same level as human operators. Perfect accuracy (100%) was achieved on aneurysm location classification., Conclusions: We have demonstrated that a comprehensive system can automatically detect, measure morphology and report the aneurysm location of aneurysms without human intervention. This can be a potential tool for the diagnosis of IAs, improving radiologists' performance and reducing their workload., (© 2022 American Association of Physicists in Medicine.)

Introduction: Microvascular invasion (MVI) is a known risk factor for prognosis after R0 liver resection for hepatocellular carcinoma (HCC). The aim of this study was to develop a deep learning prognostic prediction model by incorporating a new factor of MVI area to the other independent risk factors., Methods: Consecutive patients with HCC who underwent R0 liver resection from January to December 2016 at the Eastern Hepatobiliary Surgery Hospital were included in this retrospective study. For patients with MVI detected on resected specimens, they were divided into two groups according to the size of the maximal MVI area: the small-MVI group and the large-MVI group., Results: Of 193 patients who had MVI in the 337 HCC patients, 130 patients formed the training cohort and 63 patients formed the validation cohort. The large-MVI group of patients had worse overall survival (OS) when compared with the small-MVI group (p = 0.009). A deep learning model was developed based on the following independent risk factors found in this study: MVI stage, maximal MVI area, presence/absence of cirrhosis, and maximal tumor diameter. The areas under the receiver operating characteristic of the deep learning model for the 1-, 3-, and 5-year predictions of OS were 80.65, 74.04, and 79.44, respectively, which outperformed the traditional COX proportional hazards model., Conclusion: The deep learning model, by incorporating the maximal MVI area as an additional prognostic factor to the other previously known independent risk factors, predicted more accurately postoperative long-term OS for HCC patients with MVI after R0 liver resection., (© 2022. Asian Pacific Association for the Study of the Liver.)

The class distribution of a training dataset is an important factor which influences the performance of a deep learning-based system. Understanding the optimal class distribution is therefore crucial when building a new training set which may be costly to annotate. This is the case for histological images used in cancer diagnosis where image annotation requires domain experts. In this paper, we tackle the problem of finding the optimal class distribution of a training set to be able to train an optimal model that detects cancer in histological images. We formulate several hypotheses which are then tested in scores of experiments with hundreds of trials. The experiments have been designed to account for both segmentation and classification frameworks with various class distributions in the training set, such as natural, balanced, over-represented cancer, and over-represented non-cancer. In the case of cancer detection, the experiments show several important results: (a) the natural class distribution produces more accurate results than the artificially generated balanced distribution; (b) the over-representation of non-cancer/negative classes (healthy tissue and/or background classes) compared to cancer/positive classes reduces the number of samples which are falsely predicted as cancer (false positive); (c) the least expensive to annotate non-ROI (non-region-of-interest) data can be useful in compensating for the performance loss in the system due to a shortage of expensive to annotate ROI data; (d) the multi-label examples are more useful than the single-label ones to train a segmentation model; and (e) when the classification model is tuned with a balanced validation set, it is less affected than the segmentation model by the class distribution of the training set., (© 2022. The Author(s) under exclusive licence to Society for Imaging Informatics in Medicine.)

Chest X-ray (CXR) imaging is one of the most common diagnostic imaging techniques in clinical diagnosis and is usually used for radiological examinations to screen for thorax diseases. In this paper, we propose a novel computer-aided diagnosis (CAD) system based on a hybrid deep learning model composed of a convolutional neural network (CNN) and a graph neural network (GNN). The system is intended to explore implicit correlations between thorax diseases to aid in the multilabel chest X-ray image classification task, which we term ‶CheXGAT‶. Specifically, the proposed CheXGAT framework comprises two main modules: an image representation learning (IRL) module and a graph representation learning (GRL) module. We employ the IRL module to learn high-level visual representation features from the CXR image. From the GRL module, the self-attention mechanism aggregates neighborhood features from the graphic structure to enhance the implicit correlation between thorax diseases. We adopted a data-driven method to create a disease correlation matrix that works on the message passing and aggregation process for the nodes in the GRL module. After end-to-end training, the GRL module enhances the correlation between thorax diseases to improve diagnosis performance. We performed experiments on the NIH Chest X-ray14 dataset, which contains 112,120 frontal-view radiographs; each image has multiple thorax disease labels. In the experimental results, the average AUC score of our proposed CheXGAT model reached 0.8266, and the AUC scores of Emphysema and Hernia reached 0.9447 and 0.9313, respectively. In addition, we visualized explanations via a gradient-based localization method in our proposed CheXGAT framework. Compared with previous studies, the experimental results show the competitive performance of our framework. Specifically, we propose a CAD system that uses a hybrid model to help radiologists identify CXR images from 14 chest diseases for clinical diagnosis., Competing Interests: Declaration of competing interest The authors declare that they have no financial and personal relationships with other people or organizations that could inappropriately influence their work., (Copyright © 2022 Elsevier B.V. All rights reserved.)

Purpose: To develop a novel multimodal data fusion model by incorporating computed tomography (CT) images and clinical variables based on deep learning for predicting the invasiveness risk of stage I lung adenocarcinoma that manifests as ground-glass nodules (GGNs) and compare the diagnostic performance of it with that of radiologists., Methods: A total of 1946 patients with solitary and histopathologically confirmed GGNs with maximum diameter less than 3 cm were retrospectively enrolled. The training dataset containing 1704 GGNs was augmented by resampling, scaling, random cropping, and so forth, to generate new training data. A multimodal data fusion model based on residual learning architecture and two multilayer perceptron with attention mechanism combining CT images with patient general data and serum tumor markers was built. The distance-based confidence scores (DCS) were calculated and compared among multimodal data models with different combinations. An observer study was conducted and the prediction performance of the fusion algorithms was compared with that of the two radiologists by an independent testing dataset with 242 GGNs., Results: Among the whole GGNs, 606 GGNs are confirmed as invasive adenocarcinoma (IA) and 1340 are non-IA. The proposed novel multimodal data fusion model combining CT images, patient general data, and serum tumor markers achieved the highest accuracy (88.5%), area under a ROC curve (0.957), F1 (81.5%), F1 weighted (81.9%), and Matthews correlation coefficient (73.2%) for classifying between IA and non-IA GGNs, which was even better than the senior radiologist's performance (accuracy, 86.1%). In addition, the DCSs for multimodal data suggested that CT image had a stronger influence (0.9540) quantitatively than general data (0.6726) or tumor marker (0.6971)., Conclusion: This study demonstrated that the feasibility of integrating different types of data including CT images and clinical variables, and the multimodal data fusion model yielded higher performance for distinguishing IA from non-IA GGNs., (© 2022 American Association of Physicists in Medicine.)

Computer-aided diagnosis (CAD) systems can be used to process breast ultrasound (BUS) images with the goal of enhancing the capability of diagnosing breast cancer. Many CAD systems operate by analyzing the region-of-interest (ROI) that contains the tumor in the BUS image using conventional texture-based classification models and deep learning-based classification models. Hence, the development of these systems requires automatic methods to localize the ROI that contains the tumor in the BUS image. Deep learning object-detection models can be used to localize the ROI that contains the tumor, but the ROI generated by one model might be better than the ROIs generated by other models. In this study, a new method, called the edge-based selection method, is proposed to analyze the ROIs generated by different deep learning object-detection models with the goal of selecting the ROI that improves the localization of the tumor region. The proposed method employs edge maps computed for BUS images using the recently introduced Dense Extreme Inception Network (DexiNed) deep learning edge-detection model. To the best of our knowledge, our study is the first study that has employed a deep learning edge-detection model to detect the tumor edges in BUS images. The proposed edge-based selection method is applied to analyze the ROIs generated by four deep learning object-detection models. The performance of the proposed edge-based selection method and the four deep learning object-detection models is evaluated using two BUS image datasets. The first dataset, which is used to perform cross-validation evaluation analysis, is a private dataset that includes 380 BUS images. The second dataset, which is used to perform generalization evaluation analysis, is a public dataset that includes 630 BUS images. For both the cross-validation evaluation analysis and the generalization evaluation analysis, the proposed method obtained the overall ROI detection rate, mean precision, mean recall, and mean F1-score values of 98%, 0.91, 0.90, and 0.90, respectively. Moreover, the results show that the proposed edge-based selection method outperformed the four deep learning object-detection models as well as three baseline-combining methods that can be used to combine the ROIs generated by the four deep learning object-detection models. These findings suggest the potential of employing our proposed method to analyze the ROIs generated using different deep learning object-detection models to select the ROI that improves the localization of the tumor region.

Objective: We aimed to compare the deep learning-based (VSBONE BSI) and atlas-based (BONENAVI) segmentation accuracy that have been developed to measure the bone scan index based on skeletal segmentation., Methods: We retrospectively conducted bone scans for 383 patients with prostate cancer. These patients were divided into two groups: 208 patients were injected with 99m Tc-hydroxymethylene diphosphonate processed by VSBONE BSI, and 175 patients were injected with 99m Tc-methylene diphosphonate processed by BONENAVI. Three observers classified the skeletal segmentations as either a "Match" or "Mismatch" in the following regions: the skull, cervical vertebrae, thoracic vertebrae, lumbar vertebrae, pelvis, sacrum, humerus, rib, sternum, clavicle, scapula, and femur. Segmentation error was defined if two or more observers selected "Mismatch" in the same region. We calculated the segmentation error rate according to each administration group and evaluated the presence of hot spots suspected bone metastases in "Mismatch" regions. Multivariate logistic regression analysis was used to determine the association between segmentation error and variables like age, uptake time, total counts, extent of disease, and gamma cameras., Results: The regions of "Mismatch" were more common in the long tube bones for VSBONE BSI and in the pelvis and axial skeletons for BONENAVI. Segmentation error was observed in 49 cases (23.6%) with VSBONE BSI and 58 cases (33.1%) with BONENAVI. VSBONE BSI tended that "Mismatch" regions contained hot spots suspected of bone metastases in patients with multiple bone metastases and showed that patients with higher extent of disease (odds ratio = 8.34) were associated with segmentation error in multivariate logistic regression analysis., Conclusions: VSBONE BSI has a potential to be higher segmentation accuracy compared with BONENAVI. However, the segmentation error in VSBONE BSI occurred dependent on bone metastases burden. We need to be careful when evaluating multiple bone metastases using VSBONE BSI., (© 2022. The Author(s) under exclusive licence to The Japanese Society of Nuclear Medicine.)

Background: Deep learning is a state-of-the-art technology that has rapidly become the method of choice for medical image analysis. Its fast and robust object detection, segmentation, tracking, and classification of pathophysiological anatomical structures can support medical practitioners during routine clinical workflow. Thus, deep learning-based applications for diseases diagnosis will empower physicians and allow fast decision-making in clinical practice., Highlight: Deep learning can be more robust with various features for differentiating classes, provided the training set is large and diverse for analysis. However, sufficient medical images for training sets are not always available from medical institutions, which is one of the major limitations of deep learning in medical image analysis. This review article presents some solutions for this issue and discusses efforts needed to develop robust deep learning-based computer-aided diagnosis applications for better clinical workflow in endoscopy, radiology, pathology, and dentistry., Conclusion: The introduction of deep learning-based applications will enhance the traditional role of medical practitioners in ensuring accurate diagnoses and treatment in terms of precision, reproducibility, and scalability., Competing Interests: Conflicts of interest M.T. is an employee of Medmain Inc. The author declares no competing financial interests associated with this publication. This review article did not receive any financial supports from funding agencies in the public, commercial, or not-for-profit sectors., (Copyright © 2022 Japanese Association for Oral Biology. Published by Elsevier B.V. All rights reserved.)