Search

Abstract Background Automatic diagnosis and brain hemorrhage segmentation in Computed Tomography (CT) may be helpful in assisting the neurosurgeon in developing treatment plans that improve the patient’s chances of survival. Because medical segmentation of images is important and performing operations manually is challenging, many automated algorithms have been developed for this purpose, primarily focusing on certain image modalities. Whenever a blood vessel bursts, a dangerous medical condition known as intracranial hemorrhage (ICH) occurs. For best results, quick action is required. That being said, identifying subdural (SDH) and epidural haemorrhages (EDH) is a difficult task in this field and calls for a new, more precise detection method. Methods This work uses a head CT scan to detect cerebral bleeding and distinguish between two types of dural hemorrhages using deep learning techniques. This paper proposes a rich segmentation approach to segment both SDH and EDH by enhancing segmentation efficiency with a better feature extraction procedure. This method incorporates Spatial attention- based CSR (convolution-SE-residual) Unet, for rich segmentation and precise feature extraction. Results According to the study’s findings, the CSR based Spatial network performs better than the other models, exhibiting impressive metrics for all assessed parameters with a mean dice coefficient of 0.970 and mean IoU of 0.718, while EDH and SDH dice scores are 0.983 and 0.969 respectively. Conclusions The CSR Spatial network experiment results show that it can perform well regarding dice coefficient. Furthermore, Spatial Unet based on CSR may effectively model the complicated in segmentations and rich feature extraction and improve the representation learning compared to alternative deep learning techniques, of illness and medical treatment, to enhance the meticulousness in predicting the fatality.

Abstract Objective This study aims to assess the consistency of various CT-FFR software, to determine the reliability of current CT-FFR software, and to measure relevant influence factors. The goal is to build a solid foundation of enhanced workflow and technical principles that will ultimately improve the accuracy of measurements of coronary blood flow reserve fractions. This improvement is critical for assessing the level of ischemia in patients with coronary heart disease. Methods 103 participants were chosen for a prospective research using coronary computed tomography angiography (CCTA) assessment. Heart rate, heart rate variability, subjective picture quality, objective image quality, vascular shifting length, and other factors were assessed. CT-FFR software including K software and S software are used for CT-FFR calculations. The consistency of the two software is assessed using paired-sample t-tests and Bland-Altman plots. The error classification effect is used to construct the receiver operating characteristic curve. Results The CT-FFR measurements differed significantly between the K and S software, with a statistical significance of P

Abstract Background It is challenging to correctly identify and diagnose breast nonmass lesions. This study aimed to explore the multimodal ultrasound features associated with malignant breast nonmass lesions (NMLs), and evaluate their combined diagnostic performance. Methods This retrospective analysis was conducted on 573 breast NMLs, including 309 were benign and 264 were malignant, their multimodal ultrasound features (B-mode, color Doppler and strain elastography) were assessed by two experienced radiologists. Univariate and multivariate logistic regression analysises were used to explore multimodal ultrasound features associated with malignancy, and a nomogram was developed. Diagnostic performance and clinical utility were evaluated and validated by the receiver operating characteristic (ROC) curve, calibration curve and decision curve in the training and validation cohorts. Results Multimodal ultrasound features including linear (odds ratio [OR] = 4.69) or segmental distribution (OR = 7.67), posterior shadowing (OR = 3.14), calcification (OR = 7.40), hypovascularity (OR = 0.38), elasticity scored 4 (OR = 7.00) and 5 (OR = 15.77) were independent factors associated with malignant breast NMLs. The nomogram based on these features exhibited diagnostic performance in the training and validation cohorts were comparable to that of experienced radiologists, with superior specificity (89.4%, 89.5% vs. 81.2%) and positive predictive value (PPV) (89.2%, 90.4% vs. 82.4%). The nomogram also demonstrated good calibration in both training and validation cohorts (all P > 0.05). Decision curve analysis indicated that interventions guided by the nomogram would be beneficial across a wide range of threshold probabilities (0.05-1 in the training cohort and 0.05–0.93 in the validation cohort). Conclusions The combined use of linear or segmental distribution, posterior shadowing, calcification, hypervascularity and high elasticity score, displayed as a nomogram, demonstrated satisfied diagnostic performance for malignant breast NMLs, which may contribute to the imaging interpretation and clinical management of tumors.

Abstract Background Conventional MR acceleration techniques, such as compressed sensing, parallel imaging, and half Fourier often face limitations, including noise amplification, reduced signal-to-noise ratio (SNR) and increased susceptibility to artifacts, which can compromise image quality, especially in high-speed acquisitions. Artificial intelligence (AI)-assisted compressed sensing (ACS) has emerged as a novel approach that combines the conventional techniques with advanced AI algorithms. The objective of this study was to examine the imaging quality of the ACS approach by qualitative and quantitative analysis for brain, spine, kidney, liver, and knee MR imaging, as well as compare the performance of this method with conventional (non-ACS) MR imaging. Methods This study included 50 subjects. Three radiologists independently assessed the quality of MR images based on artefacts, image sharpness, overall image quality and diagnostic efficacy. SNR, contrast-to-noise ratio (CNR), edge content (EC), enhancement measure (EME), scanning time were used for quantitative evaluation. The Cohen’s kappa correlation coefficient (k) was employed to measure radiologists’ inter-observer agreement, and the Mann Whitney U-test used for comparison between non-ACS and ACS. Results The qualitative analysis of three radiologists demonstrated that ACS images showed superior clinical information than non-ACS images with a mean k of ~ 0.70. The images acquired with ACS approach showed statistically higher values (p

Abstract Background To construct and compare merged models integrating clinical factors, MRI-based radiomics features and deep learning (DL) models for predicting pathological complete response (pCR) to neoadjuvant chemoradiotherapy (nCRT) in patients with locally advanced rectal cancer (LARC). Methods Totally 197 patients with LARC administered surgical resection after nCRT were assigned to cohort 1 (training and test sets); meanwhile, 52 cases were assigned to cohort 2 as a validation set. Radscore and DL models were established for predicting pCR applying pre- and post-nCRT MRI data, respectively. Different merged models integrating clinical factors, Radscore and DL model were constituted. Their predictive performances were validated and compared by receiver operating characteristic (ROC) and decision curve analyses (DCA). Results Merged models were established integrating selected clinical factors, Radscore and DL model for pCR prediction. The areas under the ROC curves (AUCs) of the pre-nCRT merged model were 0.834 (95% CI: 0.737–0.931) and 0.742 (95% CI: 0.650–0.834) in test and validation sets, respectively. The AUCs of the post-nCRT merged model were 0.746 (95% CI: 0.636–0.856) and 0.737 (95% CI: 0.646–0.828) in test and validation sets, respectively. DCA showed that the pretreatment algorithm could yield enhanced clinically benefit than the post-nCRT approach. Conclusions The pre-nCRT merged model including clinical factors, Radscore and DL model constitutes an effective non-invasive tool for pCR prediction in LARC.

Abstract Background This study aimed to conduct a systematic review and meta-analysis to summarize the available evidence comparing the diagnostic accuracy of periapical radiography (PA) and cone-beam computed tomography (CBCT) for detection of vertical root fractures (VRFs). Methods A search was conducted in PubMed, Scopus, and Web of Science for articles published regarding all types of human teeth. Data were analyzed by Comprehensive Meta-Analysis statistical software V3 software program. The I2 statistic was applied to analyze heterogeneity among the studies. Results Twenty-three articles met the criteria for inclusion in the systematic review and 16 for the meta-analysis. The sensitivity and specificity for detection of VRFs were calculated to be 0.51 and 0.87, respectively for PA radiography, and 0.70 and 0.84, respectively for CBCT. Conclusions The sensitivity of CBCT was higher than PA radiography; however, difference between the specificity of the two modalities was not statistically significant.

Abstract Background The increasing use of CT imaging in emergency departments, despite efforts of reducing low-value imaging, is not fully understood, especially during and after the COVID-19 pandemic. The aim of this study was to investigate the impact of COVID-19 pandemic related measures on trends and volume in CT examinations requested in the emergency department. Methods CT examinations of the head, chest, and/or abdomen-pelvis (n = 161,008), and chest radiographs (n = 113,240) performed at our tertiary care hospital between 01/2014 and 12/2023 were retrospectively analyzed. CT examinations (head, chest, abdomen, dual-region and polytrauma) and chest radiographs requested by the emergency department during (03/2020-03/2022) and after the COVID-19 pandemic (04/2022-12/2023) were compared to a pre-pandemic control period (02/2018-02/2020). Analyses included CT examinations per emergency department visit, and prediction models based on pre-pandemic trends and inpatient data. A regular expressions text search algorithm determined the most common clinical questions. Results The usage of dual-region and chest CT examinations were higher during (+ 116,4% and + 115.8%, respectively; p

Abstract Objective Subtraction angiographies are calculated using a native and a contrast-enhanced 3D angiography images. This minimizes both bone and metal artifacts and results in a pure image of the vessels. However, carrying out the examination twice means double the radiation dose for the patient. With the help of generative AI, it could be possible to simulate subtraction angiographies from contrast-enhanced 3D angiographies and thus reduce the need for another dose of radiation without a cutback in quality. We implemented this concept by using conditional generative adversarial networks. Methods We selected all 3D subtraction angiographies from our PACS system, which had performed between 01/01/2018 and 12/31/2022 and randomly divided them into training, validation, and test sets (66%:17%:17%). We adapted the pix2pix framework to work on 3D data and trained a conditional generative adversarial network with 621 data sets. Additionally, we used 158 data sets for validation and 164 for testing. We evaluated two test sets with (n = 72) and without artifacts (n = 92). Five (blinded) neuroradiologists compared these datasets with the original subtraction dataset. They assessed similarity, subjective image quality, and severity of artifacts. Results Image quality and subjective diagnostic accuracy of the virtual subtraction angiographies revealed no significant differences compared to the original 3D angiographies. While bone and movement artifact level were reduced, artifact level caused by metal implants differed from case to case between both angiographies without one group being significant superior to the other. Conclusion Conditional generative adversarial networks can be used to simulate subtraction angiographies in clinical practice, however, new artifacts can also appear as a result of this technology.

Abstract Objective We aim to develop a predictive model for lymphovascular invasion (LVI) in patients with invasive breast cancer (IBC), using magnetic resonance imaging (MRI)-based radiomics features. Methods A total of 204 patients with IBC admitted to our hospital were included in this retrospective study. The data was split into training and validation sets at a 7:3 ratio. Feature normalization was conducted, followed by feature selection using ANOVA, correlation analysis, and LASSO in the training set. The final step involved building a logistic regression model. The LVI prediction models were established by single sequence image and combined different sequence images as follows: A: prediction model based on the optimal sequence in the 7-phase enhanced MRI scans; B: prediction model based on the optimal sequences in the sequences T1WI, T2WI, and DWI; and C: the combined model based on the optimal sequences selected from A and B. Subjects' work characteristic curves (ROC) and decision curves (DCA) were plotted to determine the extent to which they predicted LVI performance in the training and validation sets. Simultaneously, nomogram models were constructed by integrating radiomics features and independent risk factors. In addition, an additional 16 patients from the center between January and August 2024 were collected as the Nomogram external validation set. The ROC and DCA were used to evaluate the performance of the model. Results In the enhanced images, Model A built based on the enhanced 2-phase achieved the best average AUC, with a validation set of 0.764. Model B built based on the T2WI had better results, with a validation set of 0.693. Model C built by combining enhanced 2-phase and T2WI sequences had a mean AUC of 0.705 in the validation set. In addition, the tumor size, whether the tumor boundary was clear or not, and whether there was a coelom in the tumor tissue had a statistically significant effect on the LVI of IBC, and a clinical-radiomics nomogram was established. DCAs as well as Nomogram also indicate that Model A has good clinical utility. The AUC of the nomogram in the training set, internal validation set, and external validation set were 0.703, 0.615, and 0.609, respectively. The DCA also showed that the radiomics nomogram combined with clinical factors had good predictive ability for LVI. Conclusion In IBC, MRI radiomics can serve as a noninvasive predictor of LVI. The clinical-MRI radiomics model, as an efficient visual prognostic tool, shows promise in forecasting LVI. This highlights the significant potential of pre-radiomics prediction in enhancing treatment strategies.

Abstract Background This study aimed to evaluate the effectiveness of DeepLabv3+with Squeeze-and-Excitation (DeepLabv3+SE) architectures for segmenting the choroid in optical coherence tomography (OCT) images of patients with diabetic retinopathy. Methods A total of 300 B-scans were selected from 21 patients with mild to moderate diabetic retinopathy. Six DeepLabv3+SE variants, each utilizing a different pre-trained convolutional neural network (CNN) for feature extraction, were compared. Segmentation performance was assessed using the Jaccard index, Dice score (DSC), precision, recall, and F1-score. Binarization and Bland-Altman analysis were employed to evaluate the agreement between automated and manual measurements of choroidal area, luminal area (LA), and Choroidal Vascularity Index (CVI). Results DeepLabv3+SE with EfficientNetB0 achieved the highest segmentation performance, with a Jaccard index of 95.47, DSC of 98.29, precision of 98.80, recall of 97.41, and F1-score of 98.10 on the validation set. Bland-Altman analysis indicated good agreement between automated and manual measurements of LA and CVI. Conclusions DeepLabv3+SE with EfficientNetB0 demonstrates promise for accurate choroid segmentation in OCT images. This approach offers a potential solution for automated CVI calculation in diabetic retinopathy patients. Further evaluation of the proposed method on a larger and more diverse dataset can strengthen its generalizability and clinical applicability.

Abstract Purpose The objective of this study was to evaluate the feasibility of using Artificial Intelligence (AI) to measure the long-diameter of tumors for evaluating treatment response. Methods Our study included 48 patients with lung-specific target lesions and conducted 277 measurements. The radiologists recorded the long-diameter in axial imaging plane of the target lesions for each measurement. Meanwhile, AI software was utilized to measure the long-diameter in both the axial imaging plane and in three dimensions (3D). Statistical analyses including the Bland-Altman plot, Spearman correlation analysis, and paired t-test to ascertain the accuracy and reliability of our findings. Results The Bland-Altman plot showed that the AI measurements had a bias of -0.28 mm and had limits of agreement ranging from − 13.78 to 13.22 mm (P = 0.497), indicating agreement with the manual measurements. However, there was no agreement between the 3D measurements and the manual measurements, with P

Abstract Rationale and objective To investigate the MR characteristics of phlegmonous stage and abscess stage primary spinal epidural abscess. Materials and methods This study retrospectively analyzed the clinical and imaging characteristics of 27 cases of pathologically confirmed primary spinal epidural abscess. Predisposing conditions of all patients were collected. All patients underwent conventional magnetic resonance imaging, while fifteen patients also underwent post-contrast magnetic resonance imaging. Results The initial symptoms included back pain in 25 patients, fever in 18, motor deficit in five, and sensory changes in 13. Underlying diseases included distant site of infection in seven, injection therapy in five, neoplasm in five, chronic inflammatory disease in five, diabetes mellitus in four, alcoholism in three, metabolic disorder in three, hepatopathy in three, and obesity in two. Abscess location was ventral epidural space in 15 patients (55.6%) and dorsal epidural space in 12 (44.4%). On T1-weighted image, the abscess was hypointense to the spinal cord in 23 patients (85%) and isointense in four (15%). All abscesses were hyperintense to the spinal cord on T2-weighted image. Among the 15 patients who underwent contrast-enhanced imaging, ring enhancement was present in 13 and homogeneous enhancement in two. Adjacent vertebrae body edema was present in four patients. The abscess was purely intraspinal in 25 patients (92.6%). Paraspinal extension was present in two (7.4%). Conclusion Primary spinal epidural abscess patients have one or more predisposing conditions. Phlegmonous stage primary spinal epidural abscess appears isointense on T1WI and hyperintense on T2WI and enhancement is homogeneous. Abscess stage primary spinal epidural abscess hyperintense on T2WI and hypointense on T1WI and ring enhancement. Presence of vertebral body edema is an important sign to help diagnose primary spinal epidural abscess.

Abstract Objective To evaluate the effect of the contrast-enhancement-boost (CE-boost) postprocessing technique on improving the image quality of obese patients in computed tomography pulmonary angiography (CTPA) compared to hybrid iterative reconstruction (HIR) and model-based iterative reconstruction (MBIR) algorithms. Methods This prospective study was conducted on 100 patients who underwent CTPA for suspected pulmonary embolism. Non-obese patients with a body mass index (BMI) under 25 were designated as group 1, while obese patients (group 2) had a BMI exceeding 25. The CE-boost images were generated by subtracting non-contrast HIR images from contrast-enhanced HIR images to improve the visibility of pulmonary arteries further. The CT value, image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were quantitatively assessed. Two chest radiologists independently reviewed the CT images (5, best; 1, worst) across three subjective characteristics including diagnostic confidence, subjective image noise, and vascular contrast. The Friedman test and Dunn-Bonferroni correction were used for statistical analysis. Results The CE-boost had significantly higher CT values than HIR and MBIR in both groups (all p

Abstract Background Invasive pulmonary aspergillosis (IPA) is a serious fungal infection. However, current diagnostic methods have limitations. The purpose of this study was to use artificial intelligence to achieve a more accurate diagnosis of IPA. Methods Totally 263 patients (148 cases of IPA, 115 cases of non-IPA) were retrospectively enrolled from a single institution and randomly divided into training and test sets at a ratio of 7:3. Clinic-radiological independent risk factors for IPA were screened using univariate analysis and multivariate logistic regression analysis, after which a clinic-radiological model was constructed. The optimal radiomics features were extracted and screened based on CT images to construct the radiomics label score (Rad-score) and radiomics model. The optimal DL features were extracted and screened using four pre-trained convolutional neural networks, respectively, followed by the construction of the DL label score (DL-score) and DL model. Then, the radiomics-DL model was constructed. Finally, the combined model was constructed based on clinic-radiological independent risk factors, the Rad-score, and the DL-score. LR was adopted as the classifier. Receiver operating characteristic (ROC) curves were drawn, and the areas under the curve (AUC) were calculated to evaluate the efficacy of each model in predicting IPA. Additionally, based on the best-performing model on the LR classifier, four other machine learning (ML) classifiers were constructed to evaluate the predictive value for IPA. Results The AUC of the clinic-radiological model for predicting IPA in the training and test sets was 0.845 and 0.765, respectively. The AUC of the radiomics-DL and combined models in the training set was 0.871 and 0.932, while in the test set was 0.851 and 0.881, respectively. The combined model showed better predictive performance than all other models. DCA showed that taking 0.00–1.00 as the threshold, the clinical benefit of the combined model was higher than that of all other models. Then, the combined model was trained on four other machine learning classifiers, all of which achieved AUC values above 0.80 in the test set, showing good performance in predicting IPA. Conclusion Clinic, CT radiomics, and DL combined model could be used to predict IPA effectively.

Abstract Background The aim of this study was to conduct a systematic review and meta-analysis to comprehensively evaluate the performance and methodological quality of artificial intelligence (AI) in predicting recurrence after single first-line treatment for liver cancer. Methods A rigorous and systematic evaluation was conducted on the AI studies related to recurrence after single first-line treatment for liver cancer, retrieved from the PubMed, Embase, Web of Science, Cochrane Library, and CNKI databases. The area under the curve (AUC), sensitivity (SENC), and specificity (SPEC) of each study were extracted for meta-analysis. Results Six percutaneous ablation (PA) studies, 16 surgical resection (SR) studies, and 5 transarterial chemoembolization (TACE) studies were included in the meta-analysis for predicting recurrence after hepatocellular carcinoma (HCC) treatment, respectively. Four SR studies and 2 PA studies were included in the meta-analysis for recurrence after intrahepatic cholangiocarcinoma (ICC) and colorectal cancer liver metastasis (CRLM) treatment. The pooled SENC, SEPC, and AUC of AI in predicting recurrence after primary HCC treatment via PA, SR, and TACE were 0.78, 0.90, and 0.92; 0.81, 0.77, and 0.86; and 0.73, 0.79, and 0.79, respectively. The values for ICC treated with SR and CRLM treated with PA were 0.85, 0.71, 0.86 and 0.69, 0.63,0.74, respectively. Conclusion This systematic review and meta-analysis demonstrates the comprehensive application value of AI in predicting recurrence after a single first-line treatment of liver cancer, with satisfactory results, indicating the clinical translation potential of AI in predicting recurrence after liver cancer treatment.

Abstract Background Most patients with osteoporosis experience vertebral compression fracture (VCF), which significantly reduces their quality of life. These patients are at a high risk of secondary VCF regardless of treatment. Thus, accurate diagnosis of VCF is important for treating and preventing new fractures. We aimed to investigate the diagnostic and predictive value of quantitative bone imaging techniques for fresh VCF. Methods From November 2021 to March 2023, 34 patients with VCF were enrolled in this study, all of whom underwent routine 99mTc-MDP whole-body bone planar scan and local SPECT/CT imaging. The maximum standard uptake value (SUVmax) of 57 fresh VCF, 57 normal adjacent vertebrae, and 19 old VCF were measured. Based on the site of the fracture, fresh VCFs were regrouped into the intervertebral-type group and the margin-type group. Meanwhile, 52 patients who had no bone metastasis or VCFs in their bone scan were assigned to the control group. The SUVmax of 110 normal vertebral bodies and 10 old VCFs in the control group were measured. Results The median SUVmax of fresh VCF was 19.80, which was significantly higher than the SUVmax of other groups. The receiver operator characteristic (ROC) curve showed that the cut-off value of SUVmax was 9.925 for diagnosing fresh VCF. The SUVmax in the intervertebral-type group was significantly higher than that in the margin-type group (P = 0.04). The SUVmax of normal vertebrae was higher among patients than among the control group (P

Abstract Background ‘Kimura’s disease (KD) is a rare chronic inflammatory disorder of unknown etiology and is difficult to diagnose due to poor clinical presentation and imaging features. Few studies on characteristics of 18F-FDG PET/CT of KD have been reported. This study aimed to observe the reliable characteristics and usefulness of 18F-FDG PET/CT for the evaluation of consecutive patients with KD. Methods The clinical data and 18F-FDG PET/CT imaging findings of 8 patients with pathologically confirmed KD were reviewed retrospectively.18F-FDG PET/CT images were evaluated visually and semiquantitatively by measuring the maximum standardized uptake value (SUVmax). The correlations between clinical data and 18F-FDG PET/CT features were analyzed by simple linear regression. Results This study included 7 males and one female ranging in age from 17 to 79 years. The longest diameter of lesions ranged from 0.8 cm to 4.8 cm, and regional or generalized lymphadenopathy was found in all 8 patients with eosinophilia, while subcutaneous masses and salivary gland involvement concurrently were found in 4 patients. 18F-FDG PET/CT revealed that these involved lesions had high 18F-FDG uptake with SUVmax > 2.5 (2.6 to 6.3). Moreover, the margin of the lesions was well defined in 6 cases and ill defined in 2 cases, and homogeneous density and 18F-FDG uptake were both found in all these lesions. There was negative correlation between eosinophils and SUVmax (R 2 = 0.538). Conclusions Kimura’s disease should be considered when 18F-FDG PET/CT is characterized by homogeneous lesions of regional or generalized lymphadenopathy, accompanied with subcutaneous masses and salivary gland involvement concurrently, especially in patients with eosinophilia.

Abstract Background The cost of labeling to collect training data sets using deep learning is especially high in medical applications compared to other fields. Furthermore, due to variances in images depending on the computed tomography (CT) devices, a deep learning based segmentation model trained with a certain device often does not work with images from a different device. Methods In this study, we propose an efficient learning strategy for deep learning models in medical image segmentation. We aim to overcome the difficulties of segmentation in CT images by training a VNet segmentation model which enables rapid labeling of organs in CT images with the model obtained by transfer learning using a small number of manually labeled images, called SEED images. We established a process for generating SEED images and conducting transfer learning a model. We evaluate the performance of various segmentation models such as vanilla UNet, UNETR, Swin-UNETR and VNet. Furthermore, assuming a scenario that a model is repeatedly trained with CT images collected from multiple devices, in which is catastrophic forgetting often occurs, we examine if the performance of our model degrades. Results We show that transfer learning can train a model that does a good job of segmenting muscles with a small number of images. In addition, it was confirmed that VNet shows better performance when comparing the performance of existing semi-automated segmentation tools and other deep learning networks to muscle and liver segmentation tasks. Additionally, we confirmed that VNet is the most robust model to deal with catastrophic forgetting problems. Conclusion In the 2D CT image segmentation task, we confirmed that the CNN-based network shows better performance than the existing semi-automatic segmentation tool or latest transformer-based networks.

Abstract Objective To evaluate value of Nomogram prediction model based on CTA imaging features for selecting treatment methods for isolated superior mesenteric artery dissection (ISMAD). Methods Symptomatic ISMAD patients were randomly divided into a training set and a validation set in a 7:3 ratio. In the training set, relevant risk factors for conservative treatment failure in ISMAD patients were analyzed, and a Nomogram prediction model for treatment outcome of ISMAD was constructed with risk factors. The predictive value of the model was evaluated. Results Low true lumen residual ratio (TLRR), long dissection length, and large arterial angle (superior mesenteric artery [SMA]/abdominal aorta [AA]) were identified as independent high-risk factors for conservative treatment failure (P

Abstract Background To investigate whether radiomics models derived from pretreatment CT could help to predict response to immunotherapy in oral squamous cell carcinoma (OSCC). Methods Retrospectively, a total of 40 patients with measurable OSCC were included. The patients were divided into responder group and non-responder group according to the comparison of pre-treatment and post-treatment CT findings. Radiomics features were extracted from pre-treatment CT images, and optimal features were selected by univariate analysis and the least absolute shrinkage and selection operator (LASSO) regression analysis. Neural network, support vector machine, random forest and logistic regression models were used to predict response to immunotherapy in OSCC, and leave-one-out cross validation was employed to assess the performance of the classifiers. The area under the curve (AUC), accuracy, sensitivity and specificity were calculated to quantify the predictive efficacy. Results A total of 7 features were selected to build models upon machine learning methods. By comparing different machine learning based models, the neural network model achieved the best predictive ability, with an AUC of 0.864, an accuracy of 82.5%, a sensitivity of 82.5%, and a specificity of 82.5%. Conclusions The pretreatment CT-based radiomics model showed good performance in predicting response to immunotherapy in OSCC. Pretreatment CT-based radiomics model might provide an alternative approach for the selection of patients who benefit from immunotherapy.

Abstract Background In recent years, radiomics has been shown to be an effective tool for the diagnosis and prediction of diseases. Existing evidence suggests that imaging features play a key role in predicting the recurrence of lumbar disk herniation (rLDH). Thus, this study aimed to evaluate the risk of rLDH in patients undergoing percutaneous endoscopic lumbar discectomy (PELD) using radiomics to facilitate the development of more rational surgical and perioperative management strategies. Method This was a retrospective case-control study involving 487 patients who underwent PELD at the L4/5 level. The rLDH and negative groups were matched using propensity score matching (PSM). A total of 1409 radiomic features were extracted from preoperative lumbar MRI images using intraclass correlation coefficient (ICC) analysis, t-test, and LASSO analysis. Afterward, 6 predictive models were constructed and evaluated using ROC curve analysis, AUC, specificity, sensitivity, confusion matrix, and 2 repeated 3-fold cross-validations. Lastly, the Shapley Additive Explanation (SHAP) analysis provided visual explanations for the models. Results Following screening and matching, 128 patients were included in both the recurrence and control groups. Moreover, 18 of the extracted radiomic features were selected for generating six models, which achieved an AUC of 0.551–0.859 for predicting rLDH. Among these models, SVM, RF, and XG Boost exhibited superior performances. Finally, cross-validation revealed that their accuracy was 0.674–0.791, 0.647–0.729, and 0.674–0.718. Conclusion Radiomics based on MRI can be used to predict the risk of rLDH, offering more comprehensive guidance for perioperative treatment by extracting imaging information that cannot be visualized with the naked eye. Meanwhile, the accuracy and generalizability of the model can be improved in the future by incorporating more data and conducting multicenter studies.

Abstract This paper presents a non-contact and unrestrained respiration monitoring system based on the optical triangulation technique. The proposed system consists of a red-green-blue (RGB) camera and a line laser installed to face the frontal thorax of a human body. The underlying idea of the work is that the camera and line laser are mounted in opposite directions, unlike other research. By applying the proposed image processing algorithm to the camera image, laser coordinates are extracted and converted to world coordinates using the optical triangulation method. These converted world coordinates represent the height of the thorax of a person. The respiratory rate is measured by analyzing changes of the thorax surface depth. To verify system performance, the camera and the line laser are installed on the head and foot sides of a bed, respectively, facing toward the center of the bed. Twenty healthy volunteers were enrolled and underwent measurement for 100s. Evaluation results show that the optical triangulation-based image processing method demonstrates non-inferior performance to a commercial patient monitoring system with a root-mean-squared error of 0.30rpm and a maximum error of 1rpm ( $$p > 0.05$$ p > 0.05 ), which implies the proposed non-contact system can be a useful alternative to the conventional healthcare method.

Abstract Early screening methods for the thyroid gland include palpation and imaging. Although palpation is relatively simple, its effectiveness in detecting early clinical signs of the thyroid gland may be limited, especially in children, due to the shorter thyroid growth time. Therefore, this constitutes a crucial foundational work. However, accurately determining the location and size of the thyroid gland in children is a challenging task. Accuracy depends on the experience of the ultrasound operator in current clinical practice, leading to subjective results. Even among experts, there is poor agreement on thyroid identification. In addition, the effective use of ultrasound machines also relies on the experience of the ultrasound operator in current clinical practice. In order to extract sufficient texture information from pediatric thyroid ultrasound images while reducing the computational complexity and number of parameters, this paper designs a novel U-Net-based network called DC-Contrast U-Net, which aims to achieve better segmentation performance with lower complexity in medical image segmentation. The results show that compared with other U-Net-related segmentation models, the proposed DC-Contrast U-Net model achieves higher segmentation accuracy while improving the inference speed, making it a promising candidate for deployment in medical edge devices in clinical applications in the future.

Abstract Eating disorders (EDs), including anorexia nervosa (AN), bulimia nervosa (BN), binge-eating disorder (BED), and pica, are psychobehavioral conditions characterized by abnormal eating behaviors and an excessive preoccupation with weight and body shape. This review examines changes in brain regions and functional connectivity in ED patients over the past decade (2013–2023) using resting-state functional magnetic resonance imaging (rs-fMRI). Key findings highlight alterations in brain networks such as the default mode network (DMN), central executive network (CEN), and emotion regulation network (ERN). In individuals with AN, there is reduced functional connectivity in areas associated with facial information processing and social cognition, alongside increased connectivity in regions linked to sensory stimulation, aesthetic judgment, and social anxiety. Conversely, BED patients show diminished connectivity in the dorsal anterior cingulate cortex within the salience network and increased connectivity in the posterior cingulate cortex and medial prefrontal cortex within the DMN. These findings suggest that rs-fMRI could serve as a valuable biomarker for assessing brain function and predicting treatment outcomes in EDs, paving the way for personalized therapeutic strategies.

Abstract Background Muscle invasive bladder cancer (MIBC) treatment combines systemic therapy and radical cystectomy (RC) or local (chemo-)radiotherapy. Response to systemic therapy is an important outcome predictor but is difficult to assess pre-operatively. Methods We analyzed multiparametric MRI (mpMRI) in consecutive MIBC patients receiving cisplatin-based neo-adjuvant chemotherapy at our institution. Two readers, blinded for pathological outcome, independently scored mpMRI before and after 2 and 4 cycles using both a qualitative 3-step method and nacVI-RADS. We analyzed accuracy of mpMRI scores to predict pathologic complete response (pCR) and inter-observer agreement. Results We analyzed 46 patients receiving NAC, 6 patients did not undergo RC after NAC and were excluded. Eleven out of 40 (28%) patients showed a pCR. mpMRI could be assessed in over 90% of patients. Radiologic complete response (rCR) using both methods was significantly associated with pCR, with an overall specificity of 96% and sensitivity of 36% and a high inter-observer agreement. rCR as assessed by the 3-step score was significantly associated with disease free survival (DFS) benefit. Conclusion The use of nacVI-RADS can predict pCR after NAC with high specificity but low sensitivity and a high inter-observer agreement. A 3-step score adds value in determining local residual disease, rCR assessed by this method could correlate with DFS benefit. mpMRI scores should be prospectively assessed in future trials of multimodal management of MIBC and can be a predictive asset in routine clinical management.

Abstract Objective To evaluate the performance of a semi-automated artificial intelligence (AI) software program (CerebralDoc® system) in aneurysm detection and morphological measurement. Methods In this study, 354 cases of computed tomographic angiography (CTA) were retrospectively collected in our hospital. Among them, 280 cases were diagnosed with aneurysms by either digital subtraction angiography (DSA) and CTA (DSA group, n = 102), or CTA-only (non-DSA group, n = 178). The presence or absence of aneurysms, as well as their location and related morphological features determined by AI were evaluated using DSA and radiologist findings. Besides, post-processing image quality from AI and radiologists were also rated and compared. Results In the DSA group, AI achieved a sensitivity of 88.24% and an accuracy of 81.97%, whereas radiologists achieved a sensitivity of 95.10% and an accuracy of 84.43%, using DSA results as the gold standard. The AI in the non-DSA group achieved 81.46% sensitivity and 76.29% accuracy, as per the radiologists’ findings. The comparison of position consistency results showed better performance under loose criteria than strict criteria. In terms of morphological characteristics, both the DSA and the non-DSA groups agreed well with the diagnostic results for neck width and maximum diameter, demonstrating excellent ICC reliability exceeding 0.80. The AI-generated images exhibited superior quality compared to the standard software for post-processing, while also demonstrating a significantly reduced processing time. Conclusions The AI-based aneurysm detection rate demonstrates a commendable performance, while the extracted morphological parameters exhibit a remarkable consistency with those assessed by radiologists, thereby showcasing significant potential for clinical application.

Abstract Background The study aimed to identify the optimal model for predicting rectal cancer liver metastasis (RCLM). This involved constructing various prediction models to aid clinicians in early diagnosis and precise decision-making. Methods A retrospective analysis was conducted on 193 patients diagnosed with rectal adenocarcinoma were randomly divided into training set (n = 136) and validation set (n = 57) at a ratio of 7:3. The predictive performance of three models was internally validated by 10-fold cross-validation in the training set. Delineation of the tumor region of interest (ROI) was performed, followed by the extraction of radiomics features from the ROI. The least absolute shrinkage and selection operator (LASSO) regression algorithm and multivariate Cox analysis were employed to reduce the dimensionality of radiomics features and identify significant features. Logistic regression was employed to construct three prediction models: clinical, radiomics, and combined models (radiomics + clinical). The predictive performance of each model was assessed and compared. Results KRAS mutation emerged as an independent predictor of liver metastasis, yielding an odds ratio (OR) of 8.296 (95%CI: 3.471–19.830; p

Abstract Background Cervical spondylotic myelopathy (CSM) is the most common chronic spinal cord injury with poor surgical and neurologic recovery in the advanced stages of the disease. DTI parameters can serve as important biomarkers for CSM prognosis. The study aimed to investigate the predictive value of dynamic diffusion tensor imaging (DTI) for the postoperative outcomes of CSM. Methods One hundred and five patients with CSM who underwent surgery were included in this study. Patients were assessed using the Modified Japanese Orthopedic Association Score (mJOA) before and one year after surgery and then divided into groups with good (≥ 50%) and poor (

Abstract Objective To evaluate early bone marrow microvascular changes in alloxan-induced diabetic rabbits using IDEAL-IQ fat quantification, texture analysis based on DCE-MRI Ktrans map, and metabolomics. Materials and methods 24 male Japanese rabbits were randomly divided into diabetic (n = 12) and control (n = 12) groups. All rabbits underwent sagittal MRI of the lumbar vertebrae at the 0th,4th, 8th, 12th, and 16th week, respectively. The fat fraction (FF) ratio and quantitative permeability of the lumbar bone marrow was measured. Texture parameters were extracted from DCE-MRI Ktrans map. At 16th week, lumbar vertebrae 5 and 6 were used for histological analysis. Lumbar vertebra 7 was crushed to obtain bone marrow for metabolomics research. Results The FF ratio and Ktrans of the lumbar bone marrow in diabetic group were increased significantly at 16th week (t = 2.226, P = 0.02; Z = -2.721, P

Abstract Purpose To investigate the application value of support vector machine (SVM) model based on diffusion-weighted imaging (DWI), dynamic contrast-enhanced (DCE) and amide proton transfer- weighted (APTW) imaging in predicting isocitrate dehydrogenase 1(IDH-1) mutation and Ki-67 expression in glioma. Methods The DWI, DCE and APTW images of 309 patients with glioma confirmed by pathology were retrospectively analyzed and divided into the IDH-1 group (IDH-1(+) group and IDH-1(-) group) and Ki-67 group (low expression group (Ki-67 ≤ 10%) and high expression group (Ki-67 > 10%)). All cases were divided into the training set, and validation set according to the ratio of 7:3. The training set was used to select features and establish machine learning models. The SVM model was established with the data after feature selection. Four single sequence models and one combined model were established in IDH-1 group and Ki-67 group. The receiver operator characteristic (ROC) curve was used to evaluate the diagnostic performance of the model. Validation set data was used for further validation. Results Both in the IDH-1 group and Ki-67 group, the combined model had better predictive efficiency than single sequence model, although the single sequence model had a better predictive efficiency. In the Ki-67 group, the combined model was built from six selected radiomics features, and the AUC were 0.965 and 0.931 in the training and validation sets, respectively. In the IDH-1 group, the combined model was built from four selected radiomics features, and the AUC were 0.997 and 0.967 in the training and validation sets, respectively. Conclusion The radiomics model established by DWI, DCE and APTW images could be used to detect IDH-1 mutation and Ki-67 expression in glioma patients before surgery. The prediction performance of the radiomics model based on the combination sequence was better than that of the single sequence model.

Abstract Background Accurate detection of Hepatocellular carcinoma (HCC) feeding vessels during transcatheter arterial chemoembolization (TACE) is important for an effective treatment, while limiting non-target embolization. This study aimed to investigate the feasibility and accuracy of pre-TACE three dimensional (3D) CT angiography for tumor-feeding vessels detection compared to DSA. Methods Sixty-nine consecutive patients referred for TACE from May 2022 to May 2023 were included. (3D) CT images were reconstructed from the pre-TACE diagnostic multiphasic contrast enhanced CT images and compared with non-selective digital subtraction angiography (DSA) images obtained during TACE for detection of HCC feeding vessels. A “Ground truth” made by consensus between observers after reviewing all available pre-TACE CT images, and DSA and CBCT images during TACE to detect the true feeding vessels was the gold standard. Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), accuracy and ROC curve with AUC were calculated for each modality and compared. Results A total of 136 active HCCs were detected in the 69 consecutive patients included in the study. 185 feeding arteries were detected by 3D CT and DSA and included in the analysis. 3D CT detection of feeding arteries revealed mean sensitivity, specificity, PPV, NPV and accuracy of 91%, 71%, 98%, 36%, and 90%, respectively, with mean AUC = 0.81. DSA detection of feeding arteries revealed mean sensitivity, specificity, PPV, NPV, and accuracy of 80%, 58%, 96.5%, 16.5% and 78%, respectively, with mean AUC = 0.69. Conclusions Pre-TACE 3D CT angiography has shown promise in improving the detection of HCC feeding vessels compared to DSA. However, further studies are required to confirm these findings across different clinical settings and patient populations. Trial registration This study was prospectively registered at Clinicaltrials.gov with ID NCT05304572; Date of registration: 2-4-2022.

Abstract Background Breast cancer is one of the leading diseases worldwide. According to estimates by the National Breast Cancer Foundation, over 42,000 women are expected to die from this disease in 2024. Objective The prognosis of breast cancer depends on the early detection of breast micronodules and the ability to distinguish benign from malignant lesions. Ultrasonography is a crucial radiological imaging technique for diagnosing the illness because it allows for biopsy and lesion characterization. The user’s level of experience and knowledge is vital since ultrasonographic diagnosis relies on the practitioner’s expertise. Furthermore, computer-aided technologies significantly contribute by potentially reducing the workload of radiologists and enhancing their expertise, especially when combined with a large patient volume in a hospital setting. Method This work describes the development of a hybrid CNN system for diagnosing benign and malignant breast cancer lesions. The models InceptionV3 and MobileNetV2 serve as the foundation for the hybrid framework. Features from these models are extracted and concatenated individually, resulting in a larger feature set. Finally, various classifiers are applied for the classification task. Results The model achieved the best results using the softmax classifier, with an accuracy of over 95%. Conclusion Computer-aided diagnosis greatly assists radiologists and reduces their workload. Therefore, this research can serve as a foundation for other researchers to build clinical solutions.

Abstract Objectives To establish a nomogram for differentiating malignant and benign focal liver lesions (FLLs) using ultrasomics features derived from contrast-enhanced ultrasound (CEUS). Methods 527 patients were retrospectively enrolled. On the training cohort, ultrasomics features were extracted from CEUS and b-mode ultrasound (BUS). Automatic feature selection and model development were performed using the Ultrasomics-Platform software, outputting the corresponding ultrasomics scores. A nomogram based on the ultrasomics scores from artery phase (AP), portal venous phase (PVP) and delayed phase (DP) of CEUS, and clinical factors were established. On the validation cohort, the diagnostic performance of the nomogram was assessed and compared with seniorexpert and resident radiologists. Results In the training cohort, the AP, PVP and DP scores exhibited better differential performance than BUS score, with area under the curve (AUC) of 84.1-85.1% compared with the BUS (74.6%, P

Abstract Background In recent years, as deep learning has received widespread attention in the field of heart disease, some studies have explored the potential of deep learning based on coronary angiography (CAG) or coronary CT angiography (CCTA) images in detecting the extent of coronary artery stenosis. However, there is still a lack of a systematic understanding of its diagnostic accuracy, impeding the advancement of intelligent diagnosis of coronary artery stenosis. Therefore, we conducted this study to review the accuracy of image-based deep learning in detecting coronary artery stenosis. Methods We retrieved PubMed, Cochrane, Embase, and Web of Science until April 11, 2023. The risk of bias in the included studies was appraised using the QUADAS-2 tool. We extracted the accuracy of deep learning in the test set and performed subgroup analyses by binary and multiclass classification scenarios. We performed a subgroup analysis based on different degrees of stenosis and applied a double arcsine transformation to process the data. The analysis was done by using R. Results Our systematic review finally included 18 studies, involving 3568 patients and 13,362 images. In the included studies, deep learning models were constructed based on CAG and CCTA. In binary classification tasks, the accuracy for detecting > 25%, > 50% and > 70% degrees of stenosis at the vessel level were 0.81 (95% CI: 0.71–0.85), 0.73 (95% CI: 0.58–0.88) and 0.61 (95% CI: 0.56–0.65), respectively. In multiclass classification tasks, the accuracy for detecting 0–25%, 25–50%, 50–70%, and 70–100% degrees of stenosis at the vessel level were 0.78 (95% CI: 0.73–0.84), 0.86 (95% CI: 0.78–0.93), 0.83 (95% CI: 0.70–0.97), and 0.70 (95% CI: 0.42–0.98), respectively. Conclusions Our study shows that deep learning models based on CAG and CCTA appear to be relatively accurate in diagnosing different degrees of coronary artery stenosis. However, for various degrees of stenosis, their accuracy still needs to be further improved.

Abstract Objective To evaluate the prediction value of Dual-energy CT (DECT)-based quantitative parameters and radiomics model in preoperatively predicting muscle invasion in bladder cancer (BCa). Materials and methods A retrospective study was performed on 126 patients with BCa who underwent DECT urography (DECTU) in our hospital. Patients were randomly divided into training and test cohorts with a ratio of 7:3. Quantitative parameters derived from DECTU were identified through univariate and multivariate logistic regression analysis to construct a DECT model. Radiomics features were extracted from the 40, 70, 100 keV and iodine-based material-decomposition (IMD) images in the venous phase to construct radiomics models from individual and combined images using a support vector machine classifier, and the optimal performing model was chosen as the final radiomics model. Subsequently, a fusion model combining the DECT parameters and the radiomics model was established. The diagnostic performances of all three models were evaluated through receiver operating characteristic (ROC) curves and the clinical usefulness was estimated using decision curve analysis (DCA). Results The normalized iodine concentration (NIC) in DECT was an independent factor in diagnosing muscle invasion of BCa. The optimal multi-image radiomics model had predictive performance with an area-under-the-curve (AUC) of 0.867 in the test cohort, better than the AUC = 0.704 with NIC. The fusion model showed an increased level of performance, although the difference in AUC (0.893) was not statistically significant. Additionally, it demonstrated superior performance in DCA. For lesions smaller than 3 cm, the fusion model showed a high predictive capability, achieving an AUC value of 0.911. There was a slight improvement in model performance, although the difference was not statistically significant. This improvement was observed when comparing the AUC values of the DECT and radiomics models, which were 0.726 and 0.884, respectively. Conclusion The proposed fusion model combing NIC and the optimal multi-image radiomics model in DECT showed good diagnostic capability in predicting muscle invasiveness of BCa.

Abstract The analysis of the psoas muscle in morphological and functional imaging has proved to be an accurate approach to assess sarcopenia, i.e. a systemic loss of skeletal muscle mass and function that may be correlated to multifactorial etiological aspects. The inclusion of sarcopenia assessment into a radiological workflow would need the implementation of computational pipelines for image processing that guarantee segmentation reliability and a significant degree of automation. The present study utilizes three-dimensional numerical schemes for psoas segmentation in low-dose X-ray computed tomography images. Specifically, here we focused on the level set methodology and compared the performances of two standard approaches, a classical evolution model and a three-dimension geodesic model, with the performances of an original first-order modification of this latter one. The results of this analysis show that these gradient-based schemes guarantee reliability with respect to manual segmentation and that the first-order scheme requires a computational burden that is significantly smaller than the one needed by the second-order approach.

Abstract Recently emerged SAM-Med2D represents a state-of-the-art advancement in medical image segmentation. Through fine-tuning the Large Visual Model, Segment Anything Model (SAM), on extensive medical datasets, it has achieved impressive results in cross-modal medical image segmentation. However, its reliance on interactive prompts may restrict its applicability under specific conditions. To address this limitation, we introduce SAM-AutoMed, which achieves automatic segmentation of medical images by replacing the original prompt encoder with an improved MobileNet v3 backbone. The performance on multiple datasets surpasses both SAM and SAM-Med2D. Current enhancements on the Large Visual Model SAM lack applications in the field of medical image classification. Therefore, we introduce SAM-MedCls, which combines the encoder of SAM-Med2D with our designed attention modules to construct an end-to-end medical image classification model. It performs well on datasets of various modalities, even achieving state-of-the-art results, indicating its potential to become a universal model for medical image classification.

Abstract Breast cancer prediction and diagnosis are critical for timely and effective treatment, significantly impacting patient outcomes. Machine learning algorithms have become powerful tools for improving the prediction and diagnosis of breast cancer. The Breast Cancer Prediction and Diagnosis Model (BCPM), which utilises machine learning techniques to improve the precision and efficiency of breast cancer diagnosis and prediction, is presented in this paper. BCPM collects comprehensive and high-quality data from diverse sources, including electronic medical records, clinical trials, and public datasets. Through rigorous pre-processing, the data is cleaned, inconsistencies are addressed, and missing values are handled. Feature scaling techniques are applied to normalize the data, ensuring fair comparison and equal importance among different features. Furthermore, feature-selection algorithms are utilized to identify the most relevant features that contribute to breast cancer projection and diagnosis, optimizing the model’s efficiency. The BCPM employs numerous machine learning methods, such as logistic regression, random forests, decision trees, support vector machines, and neural networks, to generate accurate models. Area under the curve (AUC), sensitivity, specificity, and accuracy are only some of the metrics used to assess a model’s performance once it has been trained on a subset of data. The BCPM holds promise in improving breast cancer prediction and diagnosis, aiding in personalized treatment planning, and ultimately taming patient results. By leveraging machine learning algorithms, the BCPM contributes to ongoing efforts in combating breast cancer and saving lives.

Abstract Background This study aims to utilize the deep learning method of VB-Net to locate and segment the trigeminal nerve, and employ radiomics methods to distinguish between CTN patients and healthy individuals. Methods A total of 165 CTN patients and 175 healthy controls, matched for gender and age, were recruited. All subjects underwent magnetic resonance scans. VB-Net was used to locate and segment the bilateral trigeminal nerve of all subjects, followed by the application of radiomics methods for feature extraction, dimensionality reduction, feature selection, model construction, and model evaluation. Results On the test set for trigeminal nerve segmentation, our segmentation parameters are as follows: the mean Dice Similarity Coefficient (mDCS) is 0.74, the Average Symmetric Surface Distance (ASSD) is 0.64 mm, and the Hausdorff Distance (HD) is 3.34 mm, which are within the acceptable range. Analysis of CTN patients and healthy controls identified 12 features with larger weights, and there was a statistically significant difference in Rad_score between the two groups (p

Abstract Purpose To evaluate the predictive capabilities of MRI-based radiomics for detecting lymphovascular space invasion (LVSI) in patients diagnosed with endometrial carcinoma (EC). Materials and methods A retrospective analysis was conducted on 160 female patients diagnosed with EC. The radiomics model including T2-weighted and dynamic contrast-enhanced MRI (DCE-MRI) images was established. Additionally, a conventional MRI model, which incorporated MRI-reported FIGO stage, deep myometrial infiltration (DMI), adnexal involvement, and vaginal/parametrial involvement, was established. Finally, a combined model was created by integrating the radiomics signature and conventional MRI characteristics. The predictive performance was validated by the area under the curve (AUC) of the receiver operating characteristic (ROC) curves. A stratified analysis was conducted to compare the differences between the three models by Delong test. Results In predicting LVSI, the radiomics model outperformed the clinical model in the training cohort (AUC: 0.899 vs. 0.8862) but not in the test cohort (AUC: 0.812 vs. 0.8758). The combined model demonstrated superior performance in both the training and test cohorts (training cohort: AUC = 0.934, 95% CI: 0.8807–0.9873; testing cohort: AUC = 0.905, 95% CI: 0.7679-1). Conclusions The combined model exhibited utility in preoperatively predicting LVSI in patients with EC, offering potential benefits for clinical decision-making.

Abstract Background Several studies have advocated the use of ultrasound to diagnose distal forearm fractures in children. However, there is limited data on the diagnostic accuracy of ultrasound for distal forearm fractures when conducted by pediatric surgeons or trainees who manage orthopedic injuries in children. The objective of this study was to determine the diagnostic accuracy of point-of-care ultrasound (POCUS) for pediatric distal forearm fractures when conducted by pediatric surgeons and trainees after minimal training. Methods This diagnostic study was conducted in a tertiary hospital emergency department in Germany. Participants were children and adolescents under 15 years of age who presented to the emergency department with an acute, suspected, isolated distal forearm fracture requiring imaging. Pediatric surgeons and trainees, after minimal training for sonographic fracture diagnosis, performed 6-view distal forearm POCUS on each participant prior to X-ray imaging. All data was retrospectively collected from the hospital’s routine digital patient files. The primary outcome was the diagnostic accuracy of POCUS compared to X-ray as the reference standard. Results From February to June 2021, 146 children under 15 met all inclusion and exclusion criteria, and 106 data sets were available for analysis. Regarding the presence of a fracture, X-ray and Wrist-POCUS showed the same result in 99.1%, with 83/106 (78.3%) fractures detected in both modalities and one suspected buckle fracture on POCUS not confirmed in the radiographs. Wrist-POCUS had a sensitivity of 100% (95% CI [0.956, 1]) and a specificity of 95.8% (95% CI [0.789, 0.999]) compared to radiographs. In 6 cases, there were minor differences regarding a concomitant ulnar buckle. The amount of prior ultrasound training had no influence on the accuracy of Wrist-POCUS for diagnosing distal forearm fractures. All fractures were reliably diagnosed even when captured POCUS images did not meet all quality criteria. Conclusion Pediatric surgeons and trainees, after minimal training in POCUS, had excellent diagnostic accuracy for distal forearm fractures in children and adolescents using POCUS compared to X-ray.

Abstract Background The impression section integrates key findings of a radiology report but can be subjective and variable. We sought to fine-tune and evaluate an open-source Large Language Model (LLM) in automatically generating impressions from the remainder of a radiology report across different imaging modalities and hospitals. Methods In this institutional review board-approved retrospective study, we collated a dataset of CT, US, and MRI radiology reports from the University of California San Francisco Medical Center (UCSFMC) (n = 372,716) and the Zuckerberg San Francisco General (ZSFG) Hospital and Trauma Center (n = 60,049), both under a single institution. The Recall-Oriented Understudy for Gisting Evaluation (ROUGE) score, an automatic natural language evaluation metric that measures word overlap, was used for automatic natural language evaluation. A reader study with five cardiothoracic radiologists was performed to more strictly evaluate the model’s performance on a specific modality (CT chest exams) with a radiologist subspecialist baseline. We stratified the results of the reader performance study based on the diagnosis category and the original impression length to gauge case complexity. Results The LLM achieved ROUGE-L scores of 46.51, 44.2, and 50.96 on UCSFMC and upon external validation, ROUGE-L scores of 40.74, 37.89, and 24.61 on ZSFG across the CT, US, and MRI modalities respectively, implying a substantial degree of overlap between the model-generated impressions and impressions written by the subspecialist attending radiologists, but with a degree of degradation upon external validation. In our reader study, the model-generated impressions achieved overall mean scores of 3.56/4, 3.92/4, 3.37/4, 18.29 s,12.32 words, and 84 while the original impression written by a subspecialist radiologist achieved overall mean scores of 3.75/4, 3.87/4, 3.54/4, 12.2 s, 5.74 words, and 89 for clinical accuracy, grammatical accuracy, stylistic quality, edit time, edit distance, and ROUGE-L score respectively. The LLM achieved the highest clinical accuracy ratings for acute/emergent findings and on shorter impressions. Conclusions An open-source fine-tuned LLM can generate impressions to a satisfactory level of clinical accuracy, grammatical accuracy, and stylistic quality. Our reader performance study demonstrates the potential of large language models in drafting radiology report impressions that can aid in streamlining radiologists’ workflows.

Abstract Background Magnetic Resonance Imaging (MRI) is extensively utilized in clinical diagnostics and medical research, yet the imaging process is often compromised by noise interference. This noise arises from various sources, leading to a reduction in image quality and subsequently hindering the accurate interpretation of image details by clinicians. Traditional denoising methods typically assume that noise follows a Gaussian distribution, thereby neglecting the more complex noise types present in MRI images, such as Rician noise. As a result, denoising remains a challenging and practical task. Method The main research work of this paper focuses on modifying mask information based on a global mask mapper. The mask mapper samples all blind spot pixels on the denoised image and maps them to the same channel. By incorporating perceptual loss, it utilizes all available information to improve performance while avoiding identity mapping. During the denoising process, the model may mistakenly remove some useful information as noise, resulting in a loss of detail in the denoised image. To address this issue, we train a generative adversarial network (GAN) with adaptive hybrid attention to restore the detailed information in the denoised MRI images. Result The two-stage model NRAE shows an improvement of nearly 1.4 dB in PSNR and approximately 0.1 in SSIM on clinical datasets compared to other classic models. Specifically, compared to the baseline model, PSNR is increased by about 0.6 dB, and SSIM is only 0.015 lower. From a visual perspective, NRAE more effectively restores the details in the images, resulting in richer and clearer representation of image details. Conclusion We have developed a deep learning-based two-stage model to address noise issues in medical MRI images. This method not only successfully reduces noise signals but also effectively restores anatomical details. The current results indicate that this is a promising approach. In future work, we plan to replace the current denoising network with more advanced models to further enhance performance.

Abstract Background Aim to validate the diagnostic efficacy of radiomics models for predicting various degrees of cognitive impairment in patients with cerebral small vessel disease (CSVD). Methods Participants were divided into mild cognitive impairment group (mild-CSVD group) and sever cognitive impairment group (sever-CSVD group) according to Montreal Cognitive Assessment (MoCA) performance, 98 gender-age-education matched subjects served as normal controls. Radiomic features were extracted from the segmented hippocampus using PyRadiomics. The feature preprocessing involved replacing missing values with the mean, applying stratified random sampling to allocate subjects into training (80%) and testing (20%) sets, ensuring balance among the three classes (normal controls, mild-CSVD group, and sever-CSVD group). A feature selection method was applied to identify discriminative radiomic features, with the optimal texture feature chosen for developing diagnostic models. Performance was evaluated in both the training and testing sets using receiver operating characteristic (ROC) curve analysis. Results The radiomics model achieved an accuracy of 0.625, an AUC of 0.593, a sensitivity of 0.828, and a specificity of 0.316 in distinguishing mild-CSVD group from normal controls. When distinguishing mild-CSVD group from sever-CSVD group, the radiomics model reached an accuracy of 0.683, an AUC of 0.660, a sensitivity of 0.167, and a specificity of 0.897. Similarly, in distinguishing sever-CSVD group from normal controls, the radiomics model exhibited an accuracy of 0.781, an AUC of 0.818, a sensitivity of 0.538, and a specificity of 0.947. Conclusion Radiomics model based on hippocampal texture had disparities in the diagnostic efficacy of radiomics models in predicting various degrees of cognitive impairment in patients with CSVD.

Abstract Background Kidney biopsy is the standard of care for the diagnosis of various kidney diseases. In particular, chronic histopathologic lesions, such as interstitial fibrosis and tubular atrophy, can provide prognostic information regarding chronic kidney disease progression. In this study, we aimed to evaluate historadiological correlations between CT-based radiomic features and chronic histologic changes in native kidney biopsies and to construct and validate a radiomics-based prediction model for chronicity grade. Methods We included patients aged ≥ 18 years who underwent kidney biopsy and abdominal CT scan within a week before kidney biopsy. Left kidneys were three-dimensionally segmented using a deep learning model based on the 3D Swin UNEt Transformers architecture. We additionally defined isovolumic cortical regions of interest near the lower pole of the left kidneys. Shape, first-order, and high-order texture features were extracted after resampling and kernel normalization. Correlations and diagnostic metrics between extracted features and chronic histologic lesions were examined. A machine learning-based radiomic prediction model for moderate chronicity was developed and compared according to the segmented regions of interest (ROI). Results Overall, moderate correlations with statistical significance (P

Abstract Objective Alzheimer’s disease (AD) is a type of neurological illness that significantly impacts individuals’ daily lives. In the intelligent diagnosis of AD, 3D networks require larger computational resources and storage space for training the models, leading to increased model complexity and training time. On the other hand, 2D slices analysis may overlook the 3D structural information of MRI and can result in information loss. Approach We propose a multi-slice attention fusion and multi-view personalized fusion lightweight network for automated AD diagnosis. It incorporates a multi-branch lightweight backbone to extract features from sagittal, axial, and coronal view of MRI, respectively. In addition, we introduce a novel multi-slice attention fusion module, which utilizes a combination of global and local channel attention mechanism to ensure consistent classification across multiple slices. Additionally, a multi-view personalized fusion module is tailored to assign appropriate weights to the three views, taking into account the varying significance of each view in achieving accurate classification results. To enhance the performance of the multi-view personalized fusion module, we utilize a label consistency loss to guide the model’s learning process. This encourages the acquisition of more consistent and stable representations across all three views. Main results The suggested strategy is efficient in lowering the number of parameters and FLOPs, with only 3.75M and 4.45G respectively, and accuracy improved by 10.5% to 14% in three tasks. Moreover, in the classification tasks of AD vs. CN, AD vs. MCI and MCI vs. CN, the accuracy of the proposed method is 95.63%, 86.88% and 85.00%, respectively, which is superior to the existing methods. Conclusions The results show that the proposed approach not only excels in resource utilization, but also significantly outperforms the four comparison methods in terms of accuracy and sensitivity, particularly in detecting early-stage AD lesions. It can precisely capture and accurately identify subtle brain lesions, providing crucial technical support for early intervention and treatment.

Abstract Background Spectral imaging of photon-counting detector CT (PCD-CT) scanners allows for generating virtual non-contrast (VNC) reconstruction. By analyzing 12 abdominal organs, we aimed to test the reliability of VNC reconstructions in preserving HU values compared to real unenhanced CT images. Methods Our study included 34 patients with pancreatic cystic neoplasm (PCN). The VNC reconstructions were generated from unenhanced, arterial, portal, and venous phase PCD-CT scans using the Liver-VNC algorithm. The observed 11 abdominal organs were segmented by the TotalSegmentator algorithm, the PCNs were segmented manually. Average densities were extracted from unenhanced scans (HUunenhanced), postcontrast (HUpostcontrast) scans, and VNC reconstructions (HUVNC). The error was calculated as HUerror=HUVNC–HUunenhanced. Pearson’s or Spearman’s correlation was used to assess the association. Reproducibility was evaluated by intraclass correlation coefficients (ICC). Results Significant differences between HUunenhanced and HUVNC[unenhanced] were found in vertebrae, paraspinal muscles, liver, and spleen. HUVNC[unenhanced] showed a strong correlation with HUunenhanced in all organs except spleen (r = 0.45) and kidneys (r = 0.78 and 0.73). In all postcontrast phases, the HUVNC had strong correlations with HUunenhanced in all organs except the spleen and kidneys. The HUerror had significant correlations with HUunenhanced in the muscles and vertebrae; and with HUpostcontrast in the spleen, vertebrae, and paraspinal muscles in all postcontrast phases. All organs had at least one postcontrast VNC reconstruction that showed good-to-excellent agreement with HUunenhanced during ICC analysis except the vertebrae (ICC: 0.17), paraspinal muscles (ICC: 0.64–0.79), spleen (ICC: 0.21–0.47), and kidneys (ICC: 0.10–0.31). Conclusions VNC reconstructions are reliable in at least one postcontrast phase for most organs, but further improvement is needed before VNC can be utilized to examine the spleen, kidneys, and vertebrae.

Abstract Background To design a pulmonary ground-glass nodules (GGN) classification method based on computed tomography (CT) radiomics and machine learning for prediction of invasion in early-stage ground-glass opacity (GGO) pulmonary adenocarcinoma. Methods This retrospective study included pulmonary GGN patients who were histologically confirmed to have adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma cancer (IAC) from 2020 to 2023. CT images of all patients were automatically segmented and 107 radiomic features were obtained for each patient. Classification models were developed using random forest (RF) and cross-validation, including three one-versus-others models and one three-class model. For each model, features were ranked by normalized Gini importance, and a minimal subset was selected with a cumulative importance exceeding 0.9. These selected features were then used to train the final models. The models’ performance metrics, including area under the curve (AUC), accuracy, sensitivity, and specificity, were computed. AUC and accuracy were compared to determine the final optimal method. Results The study comprised 193 patients (mean age 54 ± 11 years, 65 men), including 65 AIS, 54 MIA, and 74 IAC, divided into one training cohort (N = 154) and one test cohort (N = 39). The final three-class RF model outperformed three individual one-versus-others models in distinguishing each class from the other two. For the multiclass classification model, the AUC, accuracy, sensitivity, and specificity were 0.87, 0.79, 0.62, and 0.88 for AIS; 0.90, 0.79, 0.54, and 0.89 for MIA; and 0.87, 0.69, 0.73, and 0.67 for IAC, respectively. Conclusions A radiomics-based multiclass RF model could effectively differentiate three types of pulmonary GGN, which enabled early diagnosis of GGO pulmonary adenocarcinoma.

Abstract Objective We aimed to probe the diagnostic value of transvaginal color Doppler ultrasound (TV-CDU) parameters in cesarean scar pregnancy (CSP) and CSP sub-types, and the relevant factors affecting patients’ surgical effects. Methods Seventy-five CSP patients (all requested termination of pregnancy) were selected as the observation group, and 75 normal pregnant women with a history of cesarean section were selected as the control group. All the study subjects underwent TV-CDU and their cesarean scar muscle (CSM) thickness, minimum sagittal muscle thickness and resistance index (RI) of blood flow in the anterior wall of the lower uterine segment were calculated. The diagnostic value of CSM, minimum sagittal muscle thickness, and RI for CSP and CSP sub-types was analyzed. The patients in the observation group were grouped into the effective group and the ineffective group according to whether the surgical treatment was effective or not, and the independent factors affecting CSP efficacy were analyzed. Results The observation group had lower CSM, minimum sagittal muscle thickness and RI than the control group. CSM, RI, and minimum sagittal thickness in patients with type II CSP were lower than those in patients with type I, and these indicators in patients with type III were lower than those in patients with type II. The area under the curve (AUC) of CSM, RI and minimum sagittal muscle thickness in combination for CSP diagnosis and the AUC for CSP sub-types were higher than those of each indicator alone. Gestational sac size and CSM were independent factors affecting CSP treatment. Conclusion Changes in TV-CDU parameters facilitates CSP diagnosis after cesarean section. CSM, minimum sagittal muscle thickness changes, and RI in combination possesses high value for CSP diagnosis and CSP sub-types. Gestational sac size and CSM are independent factors affecting CSP treatment.

Abstract Background Radiotherapy (RT) is effective for cervical cancer but causes late side effects (SE) to nearby organs. These late SE occur more than 3 months after RT and are rated by clinical findings to determine their severity. While imaging studies describe late gastrointestinal (GI) SE, none demonstrate the correlation between the findings and the toxicity grading. In this study, we demonstrated the late GI toxicity prevalence, CT findings, and their correlation. Methods We retrospectively studied uterine cervical cancer patients treated with RT between 2015 and 2018. Patient characteristics and treatment(s) were obtained from the hospital’s databases. Late RTOG/EORTC GI SE and CT images were obtained during the follow-up. Post-RT GI changes were reviewed from CT images using pre-defined criteria. Risk ratios (RR) were calculated for CT findings, and multivariable log binomial regression determined adjusted RRs. Results This study included 153 patients, with a median age of 57 years (IQR 49–65). The prevalence of ≥ grade 2 RTOG/EORTC late GI SE was 33 (27.5%). CT findings showed 91 patients (59.48%) with enhanced bowel wall (BW) thickening, 3 (1.96%) with bowel obstruction, 7 (4.58%) with bowel perforation, 6 (3.92%) with fistula, 0 (0%) with bowel ischemia, and 0 (0%) with GI bleeding. Adjusted RRs showed that enhanced BW thickening (RR 9.77, 95% CI 2.64–36.07, p = 0.001), bowel obstruction (RR 5.05, 95% CI 2.30–11.09, p