Search

This study aims to improve early lung cancer detection by creating a sophisticated Computer-Aided Diagnosis (CAD) system. This system employs advanced image processing techniques such as adaptive dynamic histogram equalization (ADHE), Local Binary Pattern (LBP), and Tsallis thresholding to effectively reduce noise, analyze textures, and segment regions. It also includes the InceptiMultiLayer-Net (IML-Net), an advanced version of the Inception V3 architecture designed to capture complex features in medical images. The IML-Net includes a multiclass Error-Correcting Output Codes (ECOC) Support Vector Machine (SVM) classifier, which improves the system's ability to handle complex classification tasks. The system also employs statistical features such as mean, variance, energy, entropy, and correlation to fully describe the characteristics of segmented regions. With an impressive 99.573% accuracy in identifying lung cancer-affected regions, as well as a sensitivity of 99.46% and a specificity of 99.24%, this CAD system has significant potential as an early lung cancer detection tool. These findings highlight the system's ability to assist clinicians in making accurate diagnoses, ultimately improving patient outcomes significantly. [ABSTRACT FROM AUTHOR]

Image fusion is a process of merging or combining two or more images into a single image. This process helps to provide more information by retaining salient information from the source images. In recent years, image fusion research has extended promptly as imaging technologies are advanced. The rapid advancement of imaging modalities, particularly in the medical field, has resulted in the creation of an increasing number of fusion algorithms. Wavelets are the most applied transforms as they provide spatial and spectral representations. The performance of each wavelet family differs in the fusion with respect to its characteristics. To determine the best family of wavelet-based fusion algorithms, it is suggested to use an analytical technique to compare various wavelets used in brain image fusion. In this study, Magnetic Resonance Imaging (MRI) and Computer Tomography (CT) images are used to create fused images. In which, Magnetic Resonance (MR) images have higher spatial resolution and better soft tissue characterization, CT images emphasize three-dimensional imaging and have minimal scan times and high imaging resolutions. It is proposed to find the best-performing wavelet filter bank to support hybrid fusion framework. A total of 87 CT and MR images are collected from different sources, and wavelet transform is applied to them to fuse. The performance of each wavelet family is evaluated on the basis of the Standard Deviation (SD) and Entropy (Epy). [ABSTRACT FROM AUTHOR]

Ultrasonography is widely used to screen thyroid tumors because it is safe, easy to use, and low-cost. However, it is simultaneously affected by speckle noise and other artifacts, so early detection of thyroid abnormalities becomes difficult for the radiologist. Therefore, various researchers continuously address the limitations of sonography and improve the diagnosis potential of US images for thyroid tissue from the last three decays. Accordingly, the present study extensively reviewed various CAD systems used to classify thyroid tumor US (TTUS) images related to datasets, despeckling algorithms, segmentation algorithms, feature extraction and selection, assessment parameters, and classification algorithms. After the exhaustive review, the achievements and challenges have been reported, and build a road map for the new researchers. [ABSTRACT FROM AUTHOR]

In this study, a computer-aided diagnosis (CAD) system for whole-slide images (WSIs) with breast cancer was proposed to achieve lesion detection and classification. Different from the previously proposed works that performed segmentation directly and applied the segmentation metrics for performance evaluation, here we computed the accuracy in patch-level, region-level, and slide-level to access our proposed framework to meet the clinical requirements, instead of simply regarding it as an AI segmentation task. To achieve that for WSIs, lesion detection was conducted at high magnification (× 40) first for region proposal by patching using the modified AlexNet model; and then, several patches were sampled from the detected regions at lower magnification (× 10) to observe cells' patterns for lesion classification by the model of ResNet50. With such fashion, the deep features being distinguishing in lesion classification from the convolutional neural networks (CNN) were further analyzed in this work to provide comprehensive interpretability for the proposed CAD system based on pathological knowledge. For our experiments, a total of 186 slides of WSIs were collected and classified into three categories: Non-Carcinoma, Ductal Carcinoma in Situ (DCIS), and Invasive Ductal Carcinoma (IDC). The slide-level accuracy rate for the three-category classification reached 90.8% (99/109) through a fivefold cross-validation and achieved 94.8% (73/77) on the testing set. In addition, the learned features reflect the clinical insights in lesion classification for explainable AI, which enhanced the reliability and validity of our proposed system. [ABSTRACT FROM AUTHOR]

Introduction: A Computer-Assisted Detection (CAD) System for classification into malignant-benign classes using CT images is proposed. Methods: Two methods that use the fractal dimension (FD) as a measure of the lung nodule contour irregularities (Box counting and Power spectrum) were implemented. The LIDC-IDRI database was used for this study. Of these, 100 slices belonging to 100 patients were analyzed with both methods. Results: The performance between both methods was similar with an accuracy higher than 90%. Little overlap was obtained between FD ranges for the different malignancy grades with both methods, being slightly better in Power spectrum. Box counting had one more false positive than Power spectrum. Discussion: Both methods are able to establish a boundary between the high and low malignancy degree. To further validate these results and enhance the performance of the CAD system, additional studies will be necessary. [ABSTRACT FROM AUTHOR]

Copyright of Knowledge: International Journal is the property of Institute for Knowledge Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

IntroductionA Computer-Assisted Detection (CAD) System for classification into malignant-benign classes using CT images is proposed.MethodsTwo methods that use the fractal dimension (FD) as a measure of the lung nodule contour irregularities (Box counting and Power spectrum) were implemented. The LIDC-IDRI database was used for this study. Of these, 100 slices belonging to 100 patients were analyzed with both methods.ResultsThe performance between both methods was similar with an accuracy higher than 90%. Little overlap was obtained between FD ranges for the different malignancy grades with both methods, being slightly better in Power spectrum. Box counting had one more false positive than Power spectrum.DiscussionBoth methods are able to establish a boundary between the high and low malignancy degree. To further validate these results and enhance the performance of the CAD system, additional studies will be necessary.

The novel coronavirus disease, or COVID-19, is a hazardous disease. It is endangering the lives of many people living in more than two hundred countries. It directly affects the lungs. In general, two main imaging modalities, i.e., computed tomography (CT) and chest x-ray (CXR) are used to achieve a speedy and reliable medical diagnosis. Identifying the coronavirus in medical images is exceedingly difficult for diagnosis, assessment, and treatment. It is demanding, time-consuming, and subject to human mistakes. In biological disciplines, excellent performance can be achieved by employing artificial intelligence (AI) models. As a subfield of AI, deep learning (DL) networks have drawn considerable attention than standard machine learning (ML) methods. DL models automatically carry out all the steps of feature extraction, feature selection, and classification. This study has performed comprehensive analysis of coronavirus classification using CXR and CT imaging modalities using DL architectures. Additionally, we have discussed how transfer learning is helpful in this regard. Finally, the problem of designing and implementing a system using computer-aided diagnostic (CAD) to find COVID-19 using DL approaches highlighted a future research possibility.

In general, breast cancer is a fatal disease; however, early detection can significantly reduce the risk of death. A physician's experience in detecting and diagnosing breast cancеr can be aided by automated feature extraction and classification procedures. Clinical exams and imaging studies are typically used to make a diagnosis of breast cancer. Mammography is by far the most common imaging technique used to detect the early warning signs of breast cancer. The goal of this paper is to design a computer-aided diagnosis/ detection (CAD) system by utilizing image processing techniques. These techniques will represent the first stage in the system, and they will significantly contribute to improving diagnostic accuracy. Next is the "Histogram of oriented gradients (HOG)" technique, which was used to extract features. The final stage involves applying machine learning techniques (MLT), in this case the support vector machine (SVM), a widely used method for detecting breast cancer using mammograms. In testing, the proposed model was found to be 94.74% accurate. [ABSTRACT FROM AUTHOR]

Breast cancer is largely responsible for female mortality across globe. Infrared imaging helps to screen early abnormal signs based on the difference in contralateral temperatures of the breasts and can be used to improve the patient survival rate. Image data is huge to process as it is. In this work, 15 biostatistical features are extracted from the breast region. Using feature selection to achieve high performance prediction, the designed three-layer back propagation artificial neural network (ANN) employs 9 significant features to classify the thermograms as malignant or benign. For this research work, thermal images from the public Visual Lab dataset have been used. The best performance evaluation metrics, viz., accuracy, sensitivity and specificity obtained are 93.8%, 90% and 95.5%, respectively for the model with 10 neurons in the hidden layer. The outcome is promising with value of overall Area Under the Curve greater than 0.9 for both classes. The design of ANN with gradient descent algorithm used in this study outperforms the other neural network models in the literature indicating that a well-designed neural network can boost the capability of thermography to predict breast abnormalities. [ABSTRACT FROM AUTHOR]

Copyright of Acta Scientiarum Naturalium Universitatis Sunyatseni / Zhongshan Daxue Xuebao is the property of Sun-Yat-Sen University and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Auxetic materials possess negative Poisson's ratio, by which they have some unique and extraordinary properties such as improved toughness and resilience, shear and impact resistance, fracture resistance, synclastic behavior, form-fitting ability, and sound absorption, as well as vibration dampening capabilities. Weaving technology has made it possible to fabricate auxetic woven fabrics. This research involves computational modeling of rotating rectangles and chiral geometry, and thereafter their fabric production using the Arahne CAD system. And the developed fabric samples were characterized for Poisson's ratio with the help of a video extensometer. Auxeticity was successfully introduced into the fabric structure, and good agreement was observed between the model and experimental values. [ABSTRACT FROM AUTHOR]

Breast cancer is the second leading cause of death in females. As such, women have high incidence and mortality rates of breast cancer. The incidence rate has been on the rise over time. The earlier breast cancer is caught, the better it shows prognosis and the lower the mortality rate is. For this reason, many researchers and medical doctors have heeded a lot of attention to the CAD systems to detect and classify breast cancer. They have proposed a myriad of methods and techniques. Among them, the CAD system based on artificial intelligence (AI) can process plenty of information fast, and its performance is evaluated to be high. As an AI algorithm, YOLO has excellent detection performance and can detect objects effectively in real time. In this paper, we proposed an anomaly detection model of mammography using a YOLOv4-based histogram. In terms of breast cancer diagnosis, mammography features a fast diagnosis time and an inexpensive cost. For this reason, it is often applied to breast cancer diagnosis. Mammography, however, generates an image only with brightness values, so that a mammogram image has a lot of noise and image edges are dim. To enhance these image edges, we create a difference through histogram and brightness range control and threshold-based region removal methods and expand the single channel of mammogram images using the generated images. Through the expansion, the image edges are enhanced and converted into a single channel again and are learned through YOLO. For performance evaluation, the method proposed in this study is compared with ResNet18, ResNet50, GoogleNet, and VGG16. According to an experiment, the proposed method had the highest accuracy, or 95.74%, followed by GoogleNet (89.9%), VGG16 (88.93%), ResNet50 (87.77%), and ResNet18 (87.67%) in order. [ABSTRACT FROM AUTHOR]

Background: This study aimed to determine the effect of multimedia-based and traditional teaching methods on the quality of dental student preparation by evaluating its smoothness, occlusal reduction, and the presence of undercut in the pre-clinic period. Methods: This study was conducted on 60 pre-clinical dental students, who were divided into two groups of A and B. Group A was trained through Multimedia-based teaching methods, including PowerPoint, instructor demonstration, andprocedural videos, and group B was trained by traditional education methods, which only included instructor demonstration. The computer-aided design (CAD) system was used to evaluate the preparation factors of smoothness, presence of undercuts, and occlusal reduction on the second premolar and first molar teeth. Results: A significant difference was found between the frequency of smoothness in two education groups for teeth 5 and 6 (P = 0.026, P = 0.022). However, there was no significant difference between the frequency distribution of occlusal reduction in the two education groups (P = 0.383 and 0.168, for teeth 5 and 6, respectively) and was no significant difference between the undercut frequency in the two education groups (P = 0.365 and 0.078 for teeth 5 and 6, respectively). Conclusions: Based on the results, multimedia-based education can effectively promote two challenging preparation factors, including occlusal reduction and smoothness among pre-clinical students. [ABSTRACT FROM AUTHOR]

Deep Learning (DL) methods, such as Convolution Neural Networks (CNNs), have shown great potential in diagnosing complex diseases. Among these diseases, Rhegmatogenous Retinal Detachment (RRD) stands out as a critical condition necessitating precise diagnosis and postoperative Visual Acuity (VA) prediction. This research introduces a DL-based Computer-Aided Diagnosis (CAD) system that utilizes Optical Coherence Tomography (OCT) images for both the diagnosis of RRD and the prediction of postoperative VA. The CAD system utilizes DL techniques and a diverse dataset, including OCT images of patients with RRD from the Hedi Raies Ophthalmology Institute of Tunis and a large public dataset of normal subjects OCT. Preprocessing steps, such as image cropping, enhancement, denoising, and resizing, are applied to the tomographic images. Data oversampling and augmentation techniques address class imbalance and improve the dataset by generating additional samples. Various DL models, including pre-trained CNN models (VGG-16, Inception-V3, Inception-ResNet-V2), Bilinear (BCNN) (BCNN (VGG – 16)2 and BCNN (Inception-V3)2), and a custom CNN architecture, are implemented for RRD diagnosis and postoperative VA prediction. The experimental outcomes demonstrate the effectiveness of the proposed CAD system in accurately diagnosing RRD and predicting postoperative VA. The system achieves high accuracy, with 99.87% for diagnosing RRD and 98.06% for predicting postoperative VA using the BCNN (VGG – 16)2 model. The developed CAD system represents a significant advancement in the field of RRD and postoperative VA prediction. By combining DL and OCT imaging, the system provides automated and accurate diagnosis, showing potential in improving patient care and treatment decisions.

Introduction: Computer-aided diagnostic (CAD) programs for malignancy risk stratification from ultrasound (US) imaging of thyroid nodules are being validated both experimentally and in real-world practice. However, they have not been tested for reliability in analyzing difficult or unclear images. Methods: US images with indeterminate characteristics were evaluated by five observers with different experience in US examination and by a commercial CAD program. The nodules, on which the observers widely agreed, were considered concordant and, if there was little agreement, not concordant or difficult to ass ess. The diagnostic performance of the readers and the CAD program was calculated and compared in both groups of nodule images. Results: In the group of concordant thyroid nodules (n = 37), the clinicians and the CAD system obtained similar levels of accuracy (77.0% vs 74.2%, res pectively; P = 0.7) and no differences were found in sensitivity (SEN) (95.0% vs 87.5%, P = 0.2), specificity (SPE) (45.5 vs 49.4, respectively; P = 0.7), positive predictive value (PPV) (75.2% vs 77.7%, respectively; P = 0.8), nor negative predictive value (NPV) (85.6 vs 77.7, respectively; P = 0.3). When analyzing the non-concordant nodules (n = 43), the CAD system presented a decrease in accuracy of 4.2%, which was significantly lower th an that observed by the experts (19.9%, P = 0.02). Conclusions: Clinical observers are similar to the CAD system in the US assessment of the risk of thyroid nodules. However, the AI system for thyroid nodules AmCAD-UT® showed more reliability in the analysis of unclear or misleading images.

In the world, one in eight women will develop breast cancer. Men can also develop it, but less frequently. This condition starts with uncontrolled cell division brought on by a change in the genes that regulate cell division and growth, which leads to the development of a nodule or tumour. These tumours can be either benign, which poses no health risk, or malignant, also known as cancerous, which puts patients' lives in jeopardy and has the potential to spread. The most common way to diagnose this problem is via mammograms. This kind of examination enables the detection of abnormalities in breast tissue, such as masses and microcalcifications, which are thought to be indicators of the presence of disease. This study aims to determine how histogram-based image enhancement methods affect the classification of mammograms into five groups: benign calcifications, benign masses, malignant calcifications, malignant masses, and healthy tissue, as determined by a CAD system of automatic mammography classification using convolutional neural networks. Both Contrast-limited Adaptive Histogram Equalization (CAHE) and Histogram Intensity Windowing (HIW) will be used (CLAHE). By improving the contrast between the image's background, fibrous tissue, dense tissue, and sick tissue, which includes microcalcifications and masses, the mammography histogram is modified using these procedures. In order to help neural networks, learn, the contrast has been increased to make it easier to distinguish between various types of tissue. The proportion of correctly classified images could rise with this technique. Using Deep Convolutional Neural Networks, a model was developed that allows classifying different types of lesions. The model achieved an accuracy of 62%, based on mini-MIAS data. The final goal of the project is the creation of an update algorithm that will be incorporated into the CAD system and will enhance the automatic identification and categorization of microcalcifications and masses. As a result, it would be possible to increase the possibility of early disease identification, which is important because early discovery increases the likelihood of a cure to almost 100%. [ABSTRACT FROM AUTHOR]

Brain tumors (BTs) are an uncommon but fatal kind of cancer. Therefore, the development of computer-aided diagnosis (CAD) systems for classifying brain tumors in magnetic resonance imaging (MRI) has been the subject of many research papers so far. However, research in this sector is still in its early stage. The ultimate goal of this research is to develop a lightweight effective implementation of the U-Net deep network for use in performing exact real-time segmentation. Moreover, a simplified deep convolutional neural network (DCNN) architecture for the BT classification is presented for automatic feature extraction and classification of the segmented regions of interest (ROIs). Five convolutional layers, rectified linear unit, normalization, and max-pooling layers make up the DCNN's proposed simplified architecture. The introduced method was verified on multimodal brain tumor segmentation (BRATS 2015) datasets. Our experimental results on BRATS 2015 acquired Dice similarity coefficient (DSC) scores, sensitivity, and classification accuracy of 88.8%, 89.4%, and 88.6% for high-grade gliomas. When it comes to segmenting BRATS 2015 BT images, the performance of our proposed CAD framework is on par with existing state-of-the-art methods. However, the accuracy achieved in this study for the classification of BT images has improved upon the accuracy reported in prior studies. Image classification accuracy for BRATS 2015 BT has been improved from 88% to 88.6%. [ABSTRACT FROM AUTHOR]

Abstract Background The health crisis resulting from the global COVID-19 pandemic highlighted more than ever the need for rapid, reliable and safe methods of diagnosis and monitoring of respiratory diseases. To study pulmonary involvement in detail, one of the most common resources is the use of different lung imaging modalities (like chest radiography) to explore the possible affected areas. Methods The study of patient characteristics like sex and age in pathologies of this type is crucial for gaining knowledge of the disease and for avoiding biases due to the clear scarcity of data when developing representative systems. In this work, we performed an analysis of these factors in chest X-ray images to identify biases. Specifically, 11 imbalance scenarios were defined with female and male COVID-19 patients present in different proportions for the sex analysis, and 6 scenarios where only one specific age range was used for training for the age factor. In each study, 3 different approaches for automatic COVID-19 screening were used: Normal vs COVID-19, Pneumonia vs COVID-19 and Non-COVID-19 vs COVID-19. The study was validated using two public chest X-ray datasets, allowing a reliable analysis to support the clinical decision-making process. Results The results for the sex-related analysis indicate this factor slightly affects the system in the Normal VS COVID-19 and Pneumonia VS COVID-19 approaches, although the identified differences are not relevant enough to worsen considerably the system. Regarding the age-related analysis, this factor was observed to be influencing the system in a more consistent way than the sex factor, as it was present in all considered scenarios. However, this worsening does not represent a major factor, as it is not of great magnitude. Conclusions Multiple studies have been conducted in other fields in order to determine if certain patient characteristics such as sex or age influenced these deep learning systems. However, to the best of our knowledge, this study has not been done for COVID-19 despite the urgency and lack of COVID-19 chest x-ray images. The presented results evidenced that the proposed methodology and tested approaches allow a robust and reliable analysis to support the clinical decision-making process in this pandemic scenario.

This paper presents an approach for developing the programming and offline simulation systems for low-cost industrial robots in the MatLab/Simulink environment. The approach is presented in the example of a virtual model of a 4-axis robot with hybrid kinematics intended for manipulation tasks. The industrial robot with hybrid kinematics consists of the well-known 5R planar parallel mechanism to which two serial axes have been added. The programming system developed in a MatLab environment involves generating G-code programs based on given pick and place points. The virtual model included in the simulation system is configured in the Simulink environment based on the CAD model of the robot and its kinematic structure. The kinematic model and the inverse kinematic problem have to be included in the virtual model to realize the motion of the virtual robot. The system of programming and simulation has been verified through several examples that include object manipulation to perform various tasks.

This paper describes the simulation of a 45° angle collision of a modelled electric scooter frame including suspension with a non-deformable wall using the finite element method. Includes creation of a CAD geometric model using the surface method with the necessary components and constraints modelled in Abaqus CAE. During the crash test simulations, the frame was found to pose no danger to the driver in a 45° angle crash simulation at 45 km/h. However, in a 45° angle impact at 85 km/h there is a risk of crushing the rider's leg.

Purpose: Many deep learning methods have been developed for pulmonary lesion detection in chest computed tomography (CT) images. However, these methods generally target one particular lesion type, that is, pulmonary nodules. In this work, we intend to develop and evaluate a novel deep learning method for a more challenging task, detecting various benign and malignant mediastinal lesions with wide variations in sizes, shapes, intensities, and locations in chest CT images. Methods: Our method for mediastinal lesion detection contains two main stages: (a) size‐adaptive lesion candidate detection followed by (b) false‐positive (FP) reduction and benign–malignant classification. For candidate detection, an anchor‐free and one‐stage detector, namely 3D‐CenterNet is designed to locate suspicious regions (i.e., candidates with various sizes) within the mediastinum. Then, a 3D‐SEResNet‐based classifier is used to differentiate FPs, benign lesions, and malignant lesions from the candidates. Results: We evaluate the proposed method by conducting five‐fold cross‐validation on a relatively large‐scale dataset, which consists of data collected on 1136 patients from a grade A tertiary hospital. The method can achieve sensitivity scores of 84.3% ± 1.9%, 90.2% ± 1.4%, 93.2% ± 0.8%, and 93.9% ± 1.1%, respectively, in finding all benign and malignant lesions at 1/8, 1/4, ½, and 1 FPs per scan, and the accuracy of benign–malignant classification can reach up to 78.7% ± 2.5%. Conclusions: The proposed method can effectively detect mediastinal lesions with various sizes, shapes, and locations in chest CT images. It can be integrated into most existing pulmonary lesion detection systems to promote their clinical applications. The method can also be readily extended to other similar 3D lesion detection tasks. [ABSTRACT FROM AUTHOR]

Computer-aided diagnosis (CAD) system may be utilized as assistants for doctors and radiologists for the detection of disease. CAD systems using deep learning approaches are promising in diagnosing brain tumors but due to their computationally intensive nature, they are resilient to deploy in real-time scenarios where speed, as well as accuracy, is required. Further, it is necessary for the deep-learning models to capture multi-scale information as the task brain-tumor classification requires modelling pixel-to-pixel relationship and spatial-contexts in tumor-affected regions. To this end, this paper introduces the representational feature learning powers of deep, efficient and lighter deep learning architectures based on novel weight initialization and layers freezing for brain tumor classification. We use five different weight initialization and freezing configurations, four from the domain of transfer learning and the remaining being random initialization. These configurations are applied over different parameters and memory efficient architectures. Results suggest that when architecture is initiated adequately with correct weight initialization configuration based on the number of trainable parameters and architectural depth, performance obtained is optimal. Experimentation over eight different CNN architectures and five different weight initialization configurations was conducted and therefore training and evaluation of 40 deep learning frameworks was carried out. From the comprehensive experimental analyses of classification performances over three classes of brain tumor, it is evident that DenseNet201 based transfer learning model with initial 5 convolution layers frozen attains state-of-the-art accuracy of 98.22% while the lightweight models of MobileNet outperform many other models attaining the highest 97.87% accuracy for transfer learning configuration with initial 3 convolutional layers frozen while sizing only 42.6 MBs. The DenseNet201 model utilizes densely-flowing skip connections which in-turn allows the model to utilize the features learning from different spatial-contexts to formulate understanding of features in accordance with current receptive fields. With the 5 convolutional layer frozen transfer-learning scheme the same architecture achieves a performance gain of 0.88% over the state-of-art-methods. Further, the efficacy of the random initialization paradigm for brain tumor classification is investigated, results suggest that the random initialization framework can be promising if the number of trainable parameters is kept in accordance with training data quantities. [ABSTRACT FROM AUTHOR]

With the help of machine learning, many of the problems that have plagued mammography in the past have been solved. Effective prediction models need many normal and tumor samples. For medical applications such as breast cancer diagnosis framework, it is difficult to gather labeled training data and construct effective learning frameworks. Transfer learning is an emerging strategy that has recently been used to tackle the scarcity of medical data by transferring pre-trained convolutional network knowledge into the medical domain. Despite the well reputation of the transfer learning based on the pre-trained Convolutional Neural Networks (CNN) for medical imaging, several hurdles still exist to achieve a prominent breast cancer classification performance. In this paper, we attempt to solve the Feature Dimensionality Curse (FDC) problem of the deep features that are derived from the transfer learning pre-trained CNNs. Such a problem is raised due to the high space dimensionality of the extracted deep features with respect to the small size of the available medical data samples. Therefore, a novel deep learning cascaded feature selection framework is proposed based on the pre-trained deep convolutional networks as well as the univariate-based paradigm. Deep learning models of AlexNet, VGG, and GoogleNet are randomly selected and used to extract the shallow and deep features from the INbreast mammograms, whereas the univariate strategy helps to overcome the dimensionality curse and multicollinearity issues for the extracted features. The optimized key features via the univariate approach are statistically significant (p-value ≤ 0.05) and have good capability to efficiently train the classification models. Using such optimal features, the proposed framework could achieve a promising evaluation performance in terms of 98.50% accuracy, 98.06% sensitivity, 98.99% specificity, and 98.98% precision. Such performance seems to be beneficial to develop a practical and reliable computer-aided diagnosis (CAD) framework for breast cancer classification. [ABSTRACT FROM AUTHOR]

Automatic segmentation of breast tumor ultrasound images can provide doctors with objective and efficient references for lesions and regions of interest. Both dataset optimization and model structure optimization are crucial for achieving optimal image segmentation performance, and it can be challenging to satisfy the clinical needs solely through model structure enhancements in the context of insufficient breast tumor ultrasound datasets for model training. While significant research has focused on enhancing the architecture of deep learning models to improve tumor segmentation performance, there is a relative paucity of work dedicated to dataset augmentation. Current data augmentation techniques, such as rotation and transformation, often yield insufficient improvements in model accuracy. The deep learning methods used for generating synthetic images, such as GANs is primarily applied to produce visually natural-looking images. Nevertheless, the accuracy of the labels for these generated images still requires manual verification, and the images exhibit a lack of diversity. Therefore, they are not suitable for the training datasets augmentation of image segmentation models. This study introduces a novel dataset augmentation approach that generates synthetic images by embedding tumor regions into normal images. We explore two synthetic methods: one using identical backgrounds and another with varying backgrounds. Through experimental validation, we demonstrate the efficiency of the synthetic datasets in enhancing the performance of image segmentation models. Notably, the synthetic method utilizing different backgrounds exhibits superior improvement compared to the identical background approach. Our findings contribute to medical image analysis, particularly in tumor segmentation, by providing a practical and effective dataset augmentation strategy that can significantly improve the accuracy and reliability of segmentation models., Competing Interests: Declaration of competing interest None Declared., (Copyright © 2024 Elsevier Ltd. All rights reserved.)