Computer-aided Diagnosis Applications Research Articles

Cross-modality transfer learning with knowledge infusion for diabetic retinopathy grading.

Ultra-wide-field (UWF) fundus photography represents an emerging retinal imaging technique offering a broader field of view, thus enhancing its utility in screening and diagnosing various eye diseases, notably diabetic retinopathy (DR). However, the application of computer-aided diagnosis for DR using UWF images confronts two major challenges. The first challenge arises from the limited availability of labeled UWF data, making it daunting to train diagnostic models due to the high cost associated with manual annotation of medical images. Secondly, existing models' performance requires enhancement due to the absence of prior knowledge to guide the learning process. By leveraging extensively annotated datasets within the field, which encompass large-scale, high-quality color fundus image datasets annotated at either image-level or pixel-level, our objective is to transfer knowledge from these datasets to our target domain through unsupervised domain adaptation. Our approach presents a robust model for assessing the severity of diabetic retinopathy (DR) by leveraging unsupervised lesion-aware domain adaptation in ultra-wide-field (UWF) images. Furthermore, to harness the wealth of detailed annotations in publicly available color fundus image datasets, we integrate an adversarial lesion map generator. This generator supplements the grading model by incorporating auxiliary lesion information, drawing inspiration from the clinical methodology of evaluating DR severity by identifying and quantifying associated lesions. We conducted both quantitative and qualitative evaluations of our proposed method. In particular, among the six representative DR grading methods, our approach achieved an accuracy (ACC) of 68.18% and a precision (pre) of 67.43%. Additionally, we conducted extensive experiments in ablation studies to validate the effectiveness of each component of our proposed method. In conclusion, our method not only improves the accuracy of DR grading, but also enhances the interpretability of the results, providing clinicians with a reliable DR grading scheme.

Adipose Tissue Segmentation after Lung Slice Localization in Chest CT Images Based on ConvBiGRU and Multi-Module UNet.

The distribution of adipose tissue in the lungs is intricately linked to a variety of lung diseases, including asthma, chronic obstructive pulmonary disease (COPD), and lung cancer. Accurate detection and quantitative analysis of subcutaneous and visceral adipose tissue surrounding the lungs are essential for effectively diagnosing and managing these diseases. However, there remains a noticeable scarcity of studies focusing on adipose tissue within the lungs on a global scale. Thus, this paper introduces a ConvBiGRU model for localizing lung slices and a multi-module UNet-based model for segmenting subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT), contributing to the analysis of lung adipose tissue and the auxiliary diagnosis of lung diseases. In this study, we propose a bidirectional gated recurrent unit (BiGRU) structure for precise lung slice localization and a modified multi-module UNet model for accurate SAT and VAT segmentations, incorporating an additive weight penalty term for model refinement. For segmentation, we integrate attention, competition, and multi-resolution mechanisms within the UNet architecture to optimize performance and conduct a comparative analysis of its impact on SAT and VAT. The proposed model achieves satisfactory results across multiple performance metrics, including the Dice Score (92.0% for SAT and 82.7% for VAT), F1 Score (82.2% for SAT and 78.8% for VAT), Precision (96.7% for SAT and 78.9% for VAT), and Recall (75.8% for SAT and 79.1% for VAT). Overall, the proposed localization and segmentation framework exhibits high accuracy and reliability, validating its potential application in computer-aided diagnosis (CAD) for medical tasks in this domain.

Computer-aided diagnostic system with automated deep learning method based on the AutoGluon framework improved the diagnostic accuracy of early esophageal cancer.

There have been studies on the application of computer-aided diagnosis (CAD) in the endoscopic diagnosis of early esophageal cancer (EEC), but there is still a significant gap from clinical application. We developed an endoscopic CAD system for EEC based on the AutoGluon framework, aiming to explore the feasibility of automatic deep learning (DL) in clinical application. The endoscopic pictures of normal esophagus, esophagitis, and EEC were collected from The First Affiliated Hospital of Soochow University (September 2015 to December 2021) and the Norwegian HyperKvasir database. All images of non-cancerous esophageal lesions and EEC in this study were pathologically examined. There were three tasks: task A was normal vs. lesion classification under non-magnifying endoscopy (n=932 vs. 1,092); task B was non-cancer lesion vs. EEC classification under non-magnifying endoscopy (n=594 vs. 429); and task C was non-cancer lesion vs. EEC classification under magnifying endoscopy (n=505 vs. 824). In all classification tasks, we took 100 pictures as the verification set, and the rest comprised as the training set. The CAD system was established based on the AutoGluon framework. Diagnostic performance of the model was compared with that of endoscopists grouped according to years of experience (senior >15 years; junior <5 years). Model evaluation indicators included accuracy, recall rate, precision, F1 value, interpretation time, and the area under the receiver operating characteristic (ROC) curve (AUC). In tasks A and B, the accuracies of medium-performance CAD and high-performance CAD were lower than those of junior doctors and senior doctors. In task C, the medium-performance and high-performance CAD accuracies were close to those of junior doctors and senior doctors. The high-performance CAD model outperformed the junior doctors in both task A (0.850 vs. 0.830) and task C (0.840 vs. 0.830) in sensitivity comparison, but there was still a large gap between high-performance CAD models and doctors in sensitivity comparison. In task A, with the aid of CAD pre-interpretation, the accuracy of junior and senior physicians were significantly improved (from 0.880 to 0.915 and from 0.920 to 0.945, respectively); the time spent on film reading was significantly shortened (junior: from 11.3 to 8.7 s; senior: from 6.7 to 5.5 s). In task C, with the aid of CAD pre-interpretation, the accuracy of junior and senior physicians were significantly improved (from 0.850 to 0.865 and from 0.915 to 0.935, respectively); the reading time was significantly shortened (junior: from 9.5 to 7.7 s; senior: from 5.6 to 3.0 s). The CAD system based on the AutoGluon framework can assist doctors to improve the diagnostic accuracy and reading time of EEC under endoscopy. This study reveals that automatic DL methods are promising in clinical application.

The application of computer-aided diagnosis in Breast Imaging Reporting and Data System ultrasound training for residents-a randomized controlled study.

The consistency of Breast Imaging Reporting and Data System (BI-RADS) classification among experienced radiologists is different, which is difficult for inexperienced radiologists to master. This study aims to explore the value of computer-aided diagnosis (CAD) (AI-SONIC breast automatic detection system) in the BI-RADS training for residents. A total of 12 residents who participated in the first year and the second year of standardized resident training in Ningbo No. 2 Hospital from May 2020 to May 2021 were randomly divided into 3 groups (Group 1, Group 2, Group 3) for BI-RADS training. They were asked to complete 2 tests and questionnaires at the beginning and end of the training. After the first test, the educational materials were given to the residents and reviewed during the breast imaging training month. Group 1 studied independently, Group 2 studied with CAD, and Group 3 was taught face-to-face by experts. The test scores and ultrasonographic descriptors of the residents were evaluated and compared with those of the radiology specialists. The trainees' confidence and recognition degree of CAD were investigated by questionnaire. There was no statistical significance in the scores of residents in the first test among the 3 groups (P=0.637). After training and learning, the scores of all 3 groups of residents were improved in the second test (P=0.006). Group 2 (52±7.30) and Group 3 (54±5.16) scored significantly higher than Group 1 (38±3.65). The consistency of ultrasonographic descriptors and final assessments between the residents and senior radiologists were improved (κ3 > κ2 > κ1), with κ2 and κ3 >0.4 (moderately consistent with experts), and κ1 =0.225 (fairly agreed with experts). The results of the questionnaire showed that the trainees had increased confidence in BI-RADS classification, especially Group 2 (1.5 to 3.5) and Group 3 (1.25 to 3.75). All trainees agreed that CAD was helpful for BI-RADS learning (Likert scale score: 4.75 out of 5) and were willing to use CAD as an aid (4.5, max. 5). The AI-SONIC breast automatic detection system can help residents to quickly master BI-RADS, improve the consistency between residents and experts, and help to improve the confidence of residents in the classification of BI-RADS, which may have potential value in the BI-RADS training for radiology residents. Chinese Clinical Trial Registry (ChiCTR2400081672).

P489 Systematic review and meta-analysis of the accuracy of computer-aided diagnosis in identifying endoscopic remission in Ulcerative Colitis

Abstract Background Endoscopic remission (ER), the absence of active inflammation during endoscopy, is a key therapeutic target in Ulcerative Colitis (UC). ER is assessed during colonoscopy using standardised scoring systems. However, there remains significant inter-operator variability in assessing ER, for which the application of computer-aided diagnosis (CADx) could be a potential solution. A previous systematic review and meta-analysis on this topic was limited to literature up to June 2022 and did not assess the methodological quality of studies or heterogeneity amongst results. We aimed to update this review. Methods Searches were performed on six electronic databases and relevant reference lists up to May 2023. Data was extracted and screened independently by two reviewers. Methodological assessment was performed using the QUADAS-2 tool. An ‘image selection’ domain was added to the tool given the importance of endoscopic image quality in training and testing data on CADx’s accuracy. Meta-analysis was performed using a bivariate model. Heterogeneity was assessed through visual inspection of forest plots and receiver-operating characteristic curves, subgroup analyses and meta-regression. The overall quality of evidence was assessed using GRADE methodology. Results 17 studies (16 full-text articles and one abstract) were identified. 15 studies adopted a cross-sectional design. All studies used expert endoscopists scoring of ER as the reference standard. The Mayo Endoscopic Score was the most commonly used scoring system (13/17). CADx assessed ER in real-time in one study, but was applied to images and videos obtained from previous endoscopic procedures in all other studies. Inception-v3 was the most common CADx system used (6/17). ‘Patient selection’ was the domain of greatest concern due to a considerable number of studies using images from the same public database and under-reporting participant characteristics. All studies had a low risk of bias (ROB) in the ‘image selection’ domain. One study was deemed overall high ROB, and two were deemed overall unclear ROB. Sensitivity analysis was performed by removing these studies and the abstract, and pooled estimates were obtained from the remaining 13 studies. CADx had a pooled sensitivity of 90.3% and specificity of 94.6% for diagnosing ER. The overall quality of evidence was moderate due to the failure of a significant proportion of studies to report sufficient participant characteristics. Conclusion There is moderate quality evidence that CADx has high sensitivity and specificity for diagnosing ER in pre-clinical studies. Further evaluation is required in randomised controlled trials to evaluate its efficacy in real-time and human-computer interactions.

Spatial dispersal in chemostat systems with agent-based asymmetry

This paper considers chemostat consumer-resource systems in which the consumer disperses between spatial patches and acts as an agent. Using the theory of dynamical systems and the method of graphs, we demonstrate global stability of steady states in the model. It is shown that dispersal of the consumer and clustering mode of patches are important for persistence of the consumer, while asymmetry in the dispersal determines the final distribution of consumer in the system. Moreover, appropriate asymmetry could make the consumer persist in sink patches, and could make the consumer approach a total population abundance higher than that without dispersal. A novel finding of this work is that agent-based selection on asymmetry could make the consumer evolve to the maximal abundance. Numerical simulations on the model reproduce radiograph/CT of COVID-19 patients in both figures and animations. Our research has potential applications in computer-aided diagnosis and treatment.

RegFSC-Net: Medical Image Registration via Fourier Transform with Spatial Reorganization and Channel Refinement Network.

Medical image registration is crucial in medical image analysis applications. Recently, U-Net-style networks have been commonly used for unsupervised image registration, predicting dense displacement fields in full-resolution space. However, this process is resourceintensive and time-consuming for high-resolution volumetric image data. To address this challenge, this paper proposes a novel model named RegFSC-Net, which utilizes Fourier transform with spatial reorganization (SR) and channel refinement (CR) network for registration. We embed efficient feature extraction modules SR and CR modules into the encoder, and adopt a parameter-free model to drive the decoder to improve the U-shaped network. Precisely, RegFSC-Net does not directly predict the fullresolution displacement field in space but learns the lowdimensional representation of the displacement field in the bandlimited Fourier domain, which is beneficial in reducing network parameters, memory usage, and computational costs. Experimental results show that RegFSC-Net outperforms various state-of-the-art methods. Specifically, in comparison to the widely recognized Transformer-based method TransMorph, RegFSC-Net utilizes only around 8.2% of its parameters, resulting in a 1.95% higher Dice score and significantly faster inference speeds of 126.67% and 419.99% on GPU and CPU, respectively. Furthermore, we also designed three variants of RegFSC-Net and demonstrated their potential applications in computer-aided diagnosis.

Application of computer-aided diagnosis to predict malignancy in BI-RADS 3 breast lesions

PurposeTo evaluate the ability of computer-aided diagnosis (CAD) system (S-Detect) to identify malignancy in ultrasound (US) -detected BI-RADS 3 breast lesions. Materials and methods148 patients with 148 breast lesions categorized as BI-RADS 3 were included in the study between January 2021 and September 2022. The malignancy rate, accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC) were calculated. ResultsIn this study, 143 breast lesions were found to be benign, and 5 breast lesions were malignant (malignancy rate, 3.4 %, 95 % confidence interval (CI): 0.5–6.3). The malignancy rate rose significantly to 18.2 % (4/22, 95 % CI: 2.1–34.3) in the high-risk group with a “possibly malignant” CAD result (p = 0.017). With a “possibly benign” CAD result, the malignancy rate decreased to 0.8 % (1/126, 95 % CI: 0–2.2) in the low-risk group (p = 0.297). The AUC, sensitivity, specificity, accuracy, PPV, and NPV of the CAD system in BI-RADS 3 breast lesions were 0.837 (95 % CI: 77.7–89.6), 80.0 % (95 % CI: 73.6–86.4), 87.4 % (95 % CI: 82.0–92.7), 87.2 % (95 % CI: 81.8–92.6), 18.2 % (95 % CI: 2.1–34.3) and 99.2 % (95 % CI: 97.8–100.0), respectively. ConclusionsCAD system (S-Detect) enables radiologists to distinguish a high-risk group and a low-risk group among US-detected BI-RADS 3 breast lesions, so that patients in the low-risk group can receive follow-up without anxiety, while those in the high-risk group with a significantly increased malignancy rate should actively receive biopsy to avoid delayed diagnosis of breast cancer.

Diseases diagnosis based on artificial intelligence and ensemble classification

BackgroundIn recent years, Computer Aided Diagnosis (CAD) has become an important research area that attracted a lot of researchers. In medical diagnostic systems, several attempts have been made to build and enhance CAD applications to avoid errors that can cause dangerously misleading medical treatments. The most exciting opportunity for promoting the performance of CAD system can be accomplished by integrating Artificial Intelligence (AI) in medicine. This allows the effective automation of traditional manual workflow, which is slow, inaccurate and affected by human errors. AimsThis paper aims to provide a complete Computer Aided Disease Diagnosis (CAD2) strategy based on Machine Learning (ML) techniques that can help clinicians to make better medical decisions. MethodsThe proposed CAD2 consists of three main sequential phases, namely; (i) Outlier Rejection Phase (ORP), (ii) Feature Selection Phase (FSP), and (iii) Classification Phase (CP). ORP is implemented to reject outliers using new Outlier Rejection Technique (ORT) that contains two sequential stages called Fast Outlier Rejection (FOR) and Accurate Outlier Rejection (AOR). The most informative features are selected through FSP using Hybrid Selection Technique (HST). HST includes two main stages called Quick Selection Stage (QS2) using fisher score as a filter method and Precise Selection Stage (PS2) using a Hybrid Bio-inspired Optimization (HBO) technique as a wrapper method. Finally, actual diagnose takes place through CP, which relies on Ensemble Classification Technique (ECT). ResultsThe proposed CAD2 has been tested experimentally against recent disease diagnostic strategies using two different datasets in which the first contains several diseases, while the second includes data for Covid-19 patients only. Experimental results have proven the high efficiency of the proposed CAD2 in terms of accuracy, error, precision, and recall compared with other competitors. Additionally, CAD2 strategy provides the best Wilcoxon signed rank test and Friedman test measurements against other strategies according to both datasets. ConclusionIt is concluded that CAD2 strategy based on ORP, FSP, and CP gave an accurate diagnosis compared to other strategies because it gave the highest accuracy and the lowest error and implementation time.

AERMNet: Attention-enhanced relational memory network for medical image report generation

Background and objectivesThe automatic generation of medical image diagnostic reports can assist doctors in reducing their workload and improving the efficiency and accuracy of diagnosis. However, among the most existing report generation models, there are problems that the weak correlation between generated words and the lack of contextual information in the report generation process. MethodsTo address the above problems, we propose an Attention-Enhanced Relational Memory Network (AERMNet) model, where the relational memory module is continuously updated by the words generated in the previous time step to strengthen the correlation between words in generated medical image report. And the double LSTM with interaction module reduces the loss of context information and makes full use of feature information. Thus, more accurate disease information can be generated by AERMNet for medical image reports. ResultsExperimental results on four medical datasets Fetal heart (FH), Ultrasound, IU X-Ray and MIMIC-CXR, show that our proposed method outperforms some of the previous models with respect to language generation metrics (Cider improving by 2.4% on FH, Bleu1 improving by 2.4% on Ultrasound, Cider improving by 16.4% on IU X-Ray, Bleu2 improving by 9.7% on MIMIC-CXR). ConclusionsThis work promotes the development of medical image report generation and expands the prospects of computer-aided diagnosis applications. Our code is released at https://github.com/llttxx/AERMNET.

BTSC-TNAS: A neural architecture search-based transformer for brain tumor segmentation and classification

Glioblastoma (GBM), isolated brain metastasis (SBM), and primary central nervous system lymphoma (PCNSL) possess a high level of similarity in histomorphology and clinical manifestations on multimodal MRI. Such similarities have led to challenges in the clinical diagnosis of these three malignant tumors. However, many existing models solely focus on either the task of segmentation or classification, which limits the application of computer-aided diagnosis in clinical diagnosis and treatment. To solve this problem, we propose a multi-task learning transformer with neural architecture search (NAS) for brain tumor segmentation and classification (BTSC-TNAS). In the segmentation stage, we use a nested transformer U-shape network (NTU-NAS) with NAS to directly predict brain tumor masks from multi-modal MRI images. In the tumor classification stage, we use the multiscale features obtained from the encoder of NTU-NAS as the input features of the classification network (MSC-NET), which are integrated and corrected by the classification feature correction enhancement (CFCE) block to improve the accuracy of classification. The proposed BTSC-TNAS achieves an average Dice coefficient of 80.86% and 87.12% for the segmentation of tumor region and the maximum abnormal region in clinical data respectively. The model achieves a classification accuracy of 0.941. The experiments performed on the BraTS 2019 dataset show that the proposed BTSC-TNAS has excellent generalizability and can provide support for some challenging tasks in the diagnosis and treatment of brain tumors.

Segmentation of Retinal Images Using Improved Segmentation Network, MesU-Net

Given the immense importance of medical image segmentation and the challenges associated with manual execution, a diverse range of automated medical image segmentation methods have been developed, primarily focusing on specific modalities of images. This paper introduces an innovative segmentation algorithm that effectively segments exudates, hemorrhages, microaneurysms, and blood vessels within retinal images using an enhanced MesNet (MesU-Net) model. By combining the MES-Net model with the U-Net model, this approach achieves accurate results in a shorter period. Consequently, it holds significant potential for clinical application in computer-aided diagnosis. The IDRID and DRIVE datasets are utilized to assess the efficacy of the proposed model for retinal segmentation. The presented method attains segmentation accuracy rates of 97.6%, 98.1%, 99.2%, and 83.7% for exudates, hemorrhages, microaneurysms, and blood vessels, respectively. This proposed model also holds promise for extension to address other medical image segmentation challenges in the future.

Toward Source-Free Cross Tissues Histopathological Cell Segmentation via Target-Specific Finetuning.

Recognition and quantitative analytics of histopathological cells are the golden standard for diagnosing multiple cancers. Despite recent advances in deep learning techniques that have been widely investigated for the automated segmentation of various types of histopathological cells, the heavy dependency on specific histopathological image types with sufficient supervised annotations, as well as the limited access to clinical data in hospitals, still pose significant challenges in the application of computer-aided diagnosis in pathology. In this paper, we focus on the model generalization of cell segmentation towards cross-tissue histopathological images. Remarkably, a novel target-specific finetuning-based self-supervised domain adaptation framework is proposed to transfer the cell segmentation model to unlabeled target datasets, without access to source datasets and annotations. When performed on the target unlabeled histopathological image set, the proposed method only needs to tune very few parameters of the pre-trained model in a self-supervised manner. Considering the morphological properties of pathological cells, we introduce two constraint terms at both local and global levels into this framework to access more reliable predictions. The proposed cross-domain framework is validated on three different types of histopathological tissues, showing promising performance in self-supervised cell segmentation. Additionally, the whole framework can be further applied to clinical tools in pathology without accessing the original training image data. The code and dataset are released at: https://github.com/NeuronXJTU/SFDA-CellSeg.

Ridge operator-assisted delineation of capsulorhexis border for cataract surgery.

With the continuous development of machine vision and imaging technology and its application in computer-aided diagnosis, it is clinically important to use computer technology to assist physicians in accurate cataract surgery. The capsulorhexis directly affects the outcome of cataract surgery, therefore, we design a method to automatically determine the virtual boundary of capsulorhexis for cataract surgery planning and tracking in-vivo to help surgeons achieve a more ideal capsulotomy geometry. In this study, an effective method was proposed to detect and display the location of capsulorhexis in cataract videos in-vivo. The initial step was locating the entire eye area by analyzing the connected components of the mirror reflective points in the image in the cataract surgery video. Then, an operator was designed for ridge edge variation and used to extract pupil edge features. Lastly, circular Hough transform was used to detect the pupillary margin and calculate the boundary between the scleral limbus and the virtual capsulorhexis border in accordance with the pupillary margin and finally displayed it in-vivo during cataract surgery. The method was tested on eight videos of cataract surgery and the results showed that 98.52% accuracy was achieved in the localization of the specular reflection point. We compared the proposed operator with the Sobel, Scharr, Laplace and Canny operators and the results showed that our operator achieved the smallest mean square error with the greatest structural similarity. The analysis demonstrated that the proposed operator outperformed other operators in detection and achieved satisfactory results in the videos of actual cataract surgeries.

AAPM task group report 273: Recommendations on best practices for ai and machine learning for computer-aided diagnosis in medical imaging.

Rapid advances in artificial intelligence (AI) and machine learning, and specifically in deep learning (DL) techniques, have enabled broad application of these methods in health care. The promise of the DL approach has spurred further interest in computer-aided diagnosis (CAD) development and applications using both 'traditional' machine learning methods and newer DL-based methods. We use the term CAD-AI to refer to this expanded clinical decision support environment that uses traditional and DL-based AI methods. Numerous studies have been published to date on the development of machine learning tools for computer-aided, or AI-assisted, clinical tasks. However, most of these machine learning models are not ready for clinical deployment. It is of paramount importance to ensure that a clinical decision support tool undergoes proper training and rigorous validation of its generalizability and robustness before adoption for patient care in the clinic. To address these important issues, the American Association of Physicists in Medicine (AAPM) Computer-Aided Image Analysis Subcommittee (CADSC) is charged, in part, to develop recommendations on practices and standards for the development and performance assessment of computer-aided decision support systems. The committee has previously published two opinion papers on the evaluation of CAD systems and issues associated with user training and quality assurance of these systems in the clinic. With machine learning techniques continuing to evolve and CAD applications expanding to new stages of the patient care process, the current task group report considers the broader issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal is to bring attention to the proper training and validation of machine learning algorithms that may improve their generalizability and reliability and accelerate the adoption of CAD-AI systems for clinical decision support. This article is protected by copyright. All rights reserved.

Fidelity-driven Optimization Reconstruction and Details Preserving Guided Fusion for Multi-Modality Medical Image

By integrating effective features of multi-modality medical images to provide richer information, multi-modality medical image fusion has been substantially used in computer-aided diagnosis applications. However, many existing fusion schemes do not consider how to eliminate the effects of the noise in source medical images and cannot provide enough details and textures for disease diagnosis. To address the problems above, we propose a new fidelity-driven optimization (FDO) reconstruction and details preserving guided-based fusion method for multi-modality medical images. To overcome the influence of noise in multi-modality medical images, a rank coefficient optimization method of low-rank approximation based on weighted mean curvature is proposed to reconstruct multi-modality medical image. Moreover, we propose an iterative detail preserving guided fusion (DPGF) method to integrate more textures and detail information of source multi-modality medical images, while ensuring high signal-to noise ratios. The experimental results show that the proposed method outperforms some of the state-of-the-art fusion methods. Specifically, the extensive experiments prove that our method has high robustness for noisy medical images, which also indicates the application prospects in diagnosis applications. Supplemental material and codes of this work are publicly available at <uri>https://github.com/VCMHE/FDO_DPGF</uri>.

Interpretable classification of pathology whole-slide images using attention based context-aware graph convolutional neural network

Background and ObjectiveWhole slide image (WSI) classification and lesion localization within giga-pixel slide are challenging tasks in computational pathology that requires context-aware representations of histological features to adequately infer nidus. The existing weakly supervised learning methods mainly treat different locations in the slide as independent regions and cannot learn potential nonlinear interactions between instances based on i.i.d assumption, resulting in the model unable to effectively utilize context-ware information to predict the labels of WSIs and locate the region of interest (ROI). MethodsHere, we propose an interpretable classification model named bidirectional Attention-based Multiple Instance Learning Graph Convolutional Network (ABMIL-GCN), which hierarchically aggregates context-aware features of instances into a global representation in a topology fashion to predict the slide labels and localize the region of lymph node metastasis in WSIs. ResultsWe verified the superiority of this method on the Camelyon16 dataset, and the results show that the average predicted ACC and AUC of the proposed model after flooding optimization can reach 90.89% and 0.9149, respectively. The average accuracy and ACC score are improved by more than 7% and 4% compared with the existing state-of-the-art algorithms. ConclusionsThe results demonstrate that context-aware GCN outperforms existing weakly supervised learning methods by introducing spatial correlations between the neighbor image patches, which also addresses the ‘accuracy-interpretability trade-off’ problem. The framework provides a novel paradigm for the clinical application of computer-aided diagnosis and intelligent systems.

Multi-label classification of Alzheimer's disease stages from resting-state fMRI-based correlation connectivity data and deep learning

Alzheimer's disease is a neurodegenerative condition that gradually impairs cognitive abilities. Recently, various neuroimaging modalities and machine learning methods have surfaced to diagnose Alzheimer's disease. Resting-state fMRI is a neuroimaging modality that has been widely utilized to study brain activity related to neurodegenerative diseases. In literature, the previous studies are limited to the binary classification of Alzheimer's disease and Mild Cognitive Impairment. The application of computer-aided diagnosis for the numerous advancing phases of Alzheimer's disease, on the other hand, remains understudied. This research analyzes and presents methods for multi-label classification of six Alzheimer's stages using rs-fMRI and deep learning. The proposed model solves the multi-class classification problem by extracting the brain's functional connectivity networks from rs-fMRI data and employing two deep learning approaches, Stacked Sparse Autoencoder and Brain Connectivity Graph Convolutional Network. The suggested models' results were assessed using the k-fold cross-validation approach, and an average accuracy of 77.13% and 84.03% was reached for multi-label classification using Stacked Sparse Autoencoders and Brain Connectivity Based Convolutional Network, respectively. An analysis of brain regions was also performed by using the network's learned weights, leading to the conclusion that the precentral gyrus, frontal gyrus, lingual gyrus, and supplementary motor area are the significant brain regions of interest.

Validation of Combined Deep Learning Triaging and Computer-Aided Diagnosis in 2901 Breast MRI Examinations From the Second Screening Round of the Dense Tissue and Early Breast Neoplasm Screening Trial.

Computer-aided triaging (CAT) and computer-aided diagnosis (CAD) of screening breast magnetic resonance imaging have shown potential to reduce the workload of radiologists in the context of dismissing normal breast scans and dismissing benign disease in women with extremely dense breasts. The aim of this study was to validate the potential of integrating CAT and CAD to reduce workload and workup on benign lesions in the second screening round of the DENSE trial, without missing cancer. We included 2901 breast magnetic resonance imaging scans, obtained from 8 hospitals in the Netherlands. Computer-aided triaging and CAD were previously developed on data from the first screening round. Computer-aided triaging dismissed examinations without lesions. Magnetic resonance imaging examinations triaged to radiological reading were counted and subsequently processed by CAD. The number of benign lesions correctly classified by CAD was recorded. The false-positive fraction of the CAD was compared with that of unassisted radiological reading in the second screening round. Receiver operating characteristics (ROC) analysis was performed and the generalizability of CAT and CAD was assessed by comparing results from first and second screening rounds. Computer-aided triaging dismissed 950 of 2901 (32.7%) examinations with 49 lesions in total; none were malignant. Subsequent CAD classified 132 of 285 (46.3%) lesions as benign without misclassifying any malignant lesion. Together, CAT and CAD yielded significantly fewer false-positive lesions, 53 of 109 (48.6%) and 89 of 109 (78.9%), respectively ( P = 0.001), than radiological reading alone. Computer-aided triaging had a smaller area under the ROC curve in the second screening round compared with the first, 0.83 versus 0.76 ( P = 0.001), but this did not affect the negative predictive value at the 100% sensitivity operating threshold. Computer-aided diagnosis was not associated with significant differences in area under the ROC curve (0.857 vs 0.753, P = 0.08). At the operating thresholds, the specificities of CAT (39.7% vs 41.0%, P = 0.70) and CAD (41.0% vs 38.2%, P = 0.62) were successfully reproduced in the second round. The combined application of CAT and CAD showed potential to reduce workload of radiologists and to reduce number of biopsies on benign lesions. Computer-aided triaging (CAT) correctly dismissed 950 of 2901 (32.7%) examinations with 49 lesions in total; none were malignant. Subsequent computer-aided diagnosis (CAD) classified 132 of 285 (46.3%) lesions as benign without misclassifying any malignant lesion. Together, CAT and CAD yielded significantly fewer false-positive lesions, 53 of 109 (48.6%) and 89 of 109 (78.9%), respectively ( P = 0.001), than radiological reading alone.

Decoupling multi-task causality for improved skin lesion segmentation and classification

Multi-task learning has been used widely in many computer aided diagnosis applications recently, while the trade-off between different tasks remains challenging. Also, the inherent causality is less studied. In this paper, we focus on skin lesion analysis, including lesion classification, detection and segmentation. By defining the chain relationship (i.e., lesion detection boosts contour segmentation, and segmentation boosts lesion classification in turn), and further decoupling each pair-wise causality (e.g., detection to segmentation) from the Pareto efficiency view, we can solve the common trade-off issue between multi-task. On this basis, we propose a novel paradigm to improve the skin lesion segmentation and classification separately, and favourable feature fusion ways for each task are explored. Moreover, to address the huge model size problem, we design an effective model compression scheme (MCS). Extensive experiments on the ISIC2017 and PH2 datasets are conducted to evaluate the proposed paradigm. The results demonstrate that the popular models such as ResNet, DenseNet and UNet for lesion analysis can be boosted by applying the proposed paradigm, and the designed MCS reduces the amount of model parameters efficiently. We achieve performance improvements on skin lesion segmentation and classification without strenuous network design and soaring model complexity. This proposed approach is promising for the multi-task diagnosis setting in other medical applications.