Pamela Franco , Cristian Montalba , Raúl Caulier-Cisterna , Carlos Milovic , Alfonso González , Juan Pablo Ramirez-Mahaluf , Juan Undurraga , Rodrigo Salas , Nicolás Crossley , Cristian Tejos , Sergio Uribe
{"title":"Interpretable machine learning model for characterizing magnetic susceptibility-based biomarkers in first episode psychosis","authors":"Pamela Franco , Cristian Montalba , Raúl Caulier-Cisterna , Carlos Milovic , Alfonso González , Juan Pablo Ramirez-Mahaluf , Juan Undurraga , Rodrigo Salas , Nicolás Crossley , Cristian Tejos , Sergio Uribe","doi":"10.1016/j.cmpb.2025.109067","DOIUrl":"10.1016/j.cmpb.2025.109067","url":null,"abstract":"<div><h3>Background and Purpose</h3><div>Several studies have shown changes in neurochemicals within the deep-brain nuclei of patients with psychosis. These alterations indicate a dysfunction in dopamine within subcortical regions affected by fluctuations in iron concentrations. Quantitative Susceptibility Mapping (QSM) is a method employed to measure iron concentration, offering a potential means to identify dopamine dysfunction in these subcortical areas. This study employed a random forest algorithm to predict susceptibility features of the First-Episode Psychosis (FEP) and the response to antipsychotics using Shapley Additionality Explanation (SHAP) values.</div></div><div><h3>Methods</h3><div>3D multi-echo Gradient Echo (GRE) and T1-weighted GRE were obtained in 61 healthy-volunteers (HV) and 76 FEP patients (32 % Treatment-Resistant Schizophrenia (TRS) and 68 % treatment-Responsive Schizophrenia (RS)) using a 3T Philips Ingenia MRI scanner. QSM and R2* were reconstructed and averaged in twenty-two segmented regions of interest. We used a Sequential Forward Selection as a feature selection algorithm and a Random Forest as a model to predict FEP patients and their response to antipsychotics. We further applied the SHAP framework to identify informative features and their interpretations. Finally, multiple correlation patterns from magnetic susceptibility parameters were extracted using hierarchical clustering.</div></div><div><h3>Results</h3><div>Our approach accurately classifies HV and FEP patients with 76.48 ± 10.73 % accuracy (using four features) and TRS vs RS patients with 76.43 ± 12.57 % accuracy (using four features), using 10-fold stratified cross-validation. The SHAP analyses indicated the top four nonlinear relationships between the selected features. Hierarchical clustering revealed two groups of correlated features for each study.</div></div><div><h3>Conclusions</h3><div>Early prediction of treatment response enables tailored strategies for FEP patients with treatment resistance, ensuring timely and effective interventions.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109067"},"PeriodicalIF":4.8,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145019203","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zhenwei Wang , Qiule Sun , Bingbing Zhang , Pengfei Wang , Jianxin Zhang , Qiang Zhang
{"title":"PM2: A new prompting multi-modal model paradigm for few-shot medical image classification","authors":"Zhenwei Wang , Qiule Sun , Bingbing Zhang , Pengfei Wang , Jianxin Zhang , Qiang Zhang","doi":"10.1016/j.cmpb.2025.109042","DOIUrl":"10.1016/j.cmpb.2025.109042","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Few-shot learning has emerged as a key technological solution to address challenges such as limited data and the difficulty of acquiring annotations in medical image classification. However, relying solely on a single image modality is insufficient to capture conceptual categories. Therefore, medical image classification requires a comprehensive approach to capture conceptual category information that aids in the interpretation of image content.</div></div><div><h3>Methods:</h3><div>This study proposes a novel medical image classification paradigm based on a multi-modal foundation model, called PM<sup>2</sup>. In addition to the image modality, PM<sup>2</sup> introduces supplementary text input (prompt) to further describe images or conceptual categories and facilitate cross-modal few-shot learning. We empirically studied five different prompting schemes under this new paradigm. Furthermore, linear probing in multi-modal models only takes class token as input, ignoring the rich statistical data contained in high-level visual tokens. Therefore, we alternately perform linear classification on the feature distributions of visual tokens and class token. To effectively extract statistical information, we use global covariance pool with efficient matrix power normalization to aggregate the visual tokens. We then combine two classification heads: one for handling image class token and prompt representations encoded by the text encoder, and the other for classifying the feature distributions of visual tokens.</div></div><div><h3>Results:</h3><div>Experimental results on three datasets: breast cancer, brain tumor, and diabetic retinopathy demonstrate that PM<sup>2</sup> effectively improves the performance of medical image classification. Compared to existing multi-modal models, PM<sup>2</sup> achieves state-of-the-art performance.</div></div><div><h3>Conclusions:</h3><div>Integrating text prompts as supplementary samples effectively enhances the model’s performance. Additionally, by leveraging second-order features of visual tokens to enrich the category feature space and combining them with class token, the model’s representational capacity is significantly strengthened.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109042"},"PeriodicalIF":4.8,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145019202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mingyuan Qin , Lei Feng , Jing Lu , Ziyan Sun , Zhengyu Yu , Lianyi Han
{"title":"ZeroTuneBio NER: A three-stage framework for zero-shot and zero-tuning biomedical entity extraction using large language models and prompt engineering","authors":"Mingyuan Qin , Lei Feng , Jing Lu , Ziyan Sun , Zhengyu Yu , Lianyi Han","doi":"10.1016/j.cmpb.2025.109070","DOIUrl":"10.1016/j.cmpb.2025.109070","url":null,"abstract":"<div><h3>Objective</h3><div>This study aims to (1) enhance the performance of large language models (LLMs) in biomedical entity extraction, (2) investigate zero-shot named entity recognition (NER) capabilities without fine-tuning, and (3) compare the proposed framework with existing models and human annotation methods. Additionally, we analyze discrepancies between human and LLM-generated annotations to refine manual labeling processes for specialized datasets.</div></div><div><h3>Materials and Methods</h3><div>We propose <strong>ZeroTuneBio NER</strong>, a three-stage NER framework integrating chain-of-thought reasoning and prompt engineering. Evaluated on three public datasets (disease, chemistry, and gene), the method requires no task-specific examples or LLM fine-tuning, addressing challenges in complex concept interpretation.</div></div><div><h3>Results</h3><div>ZeroTuneBio NER excels in tasks without strict matching, achieving an average F1-score improvement of <strong>0.28</strong> over direct LLM queries and a partial-matching F1-score of <strong>∼88</strong> <strong>%</strong>. It rivals the performance of a fine-tuned LLaMA model trained on <strong>11,240 examples</strong> and surpasses BioBERT trained on <strong>22,480 examples</strong> when strict-matching errors are excluded. Notably, LLMs significantly optimize manual annotation, accelerating speed and reducing costs.</div></div><div><h3>Conclusion</h3><div>ZeroTuneBio NER demonstrates that LLMs can perform high-quality NER without fine-tuning, reducing reliance on manual annotation. The framework broadens LLM applications in biomedical NER, while our analysis highlights its scalability and future research directions.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109070"},"PeriodicalIF":4.8,"publicationDate":"2025-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145046605","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jiahong Jin , Tianshuai Li , Hongda Liu , Marion S. Greene , Xingying Yao , Heng Hong , Xin-Hua Hu
{"title":"Reconstruction of three human lymphocyte subtypes for benchmarking in 3D morphology and modeling","authors":"Jiahong Jin , Tianshuai Li , Hongda Liu , Marion S. Greene , Xingying Yao , Heng Hong , Xin-Hua Hu","doi":"10.1016/j.cmpb.2025.109069","DOIUrl":"10.1016/j.cmpb.2025.109069","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Lymphocytes play critical roles in human immune response. Reconstruction of human primary cells from confocal image stacks provides important benchmark data for phenotype comparison and enables optical modeling to understand, for example, label-free classification.</div></div><div><h3>Methods</h3><div>We present a novel method of section-for-clustering (SFC) to automate organelle segmentation in all slices of a fluorescence confocal image stack by taking the advantage of spatial correlation among slices for reconstruction of live primary cells.</div></div><div><h3>Results</h3><div>A total of 217 live CD4+ T, CD8+ T and CD19+ B cells have been isolated from human spleen tissues for staining and confocal imaging. The SFC method has been applied to determine 24 cellular, nuclear and mitochondrial parameters for comparison of 3D morphology and all lymphocytes have been found to possess large nucleus-to-cell volume ratios. Although CD4+ T and CD8+ T cells exhibit high morphological similarity as expected, multiple parameters reveal statistically significant differences between CD4+ T and CD19+ B cells. The subtypes were classified by morphological parameters using a support vector machine method with accuracies much less than those by diffraction images. To illustrate the difference, we derived realistic optical cell models from the reconstructed lymphocytes to demonstrate that varied refractive index within organelles can supply intriguing features for accurate classification.</div></div><div><h3>Conclusions</h3><div>The presented method provides an accurate, efficient and robust approach to automate organelle segmentation of fluorescence confocal image stacks and yields one of the largest morphological databases on primary human lymphocytes for quantitative 3D assay and optical modeling.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109069"},"PeriodicalIF":4.8,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145027099","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Neighboring tissues as diagnostic windows: Neighborhood effects in radiomic detection of pancreatic ductal adenocarcinoma","authors":"Seyed Masoud Rezaeijo , Alireza Eftekhar , Saleh Rouhi , Behnaz Keshavarzi , Zahra Mohammadi , Leila Alipour Firouzabad , Maryam Abidi , Salimeh Ghafeli","doi":"10.1016/j.cmpb.2025.109056","DOIUrl":"10.1016/j.cmpb.2025.109056","url":null,"abstract":"<div><h3>Objective</h3><div>This study introduces a pioneering, tumor-independent diagnostic approach for Pancreatic Ductal Adenocarcinoma (PDAC), utilizing radiomic and deep features from non-tumorous CT regions to detect subtle, systemic tissue alterations beyond conventional imaging limits.</div></div><div><h3>Materials and Methods</h3><div>A retrospective cohort of 1263 patients was analyzed, including both PDAC and non-PDAC cases, with anatomical segmentation masks encompassing veins, arteries, pancreatic parenchyma, pancreatic duct, and common bile duct. Radiomic features (n = 107 per region) were extracted using the PyRadiomics library, while deep features were derived from a 3D convolutional autoencoder trained on cropped and normalized anatomical volumes. Three analytical approaches were implemented: (1) healthy tissue analysis using features exclusively from non-tumorous structures, (2) row-wise label combination treating each anatomical label as a separate instance, and (3) column-wise patient-level fusion aggregating multi-tissue features. Each dataset underwent multiple feature selection methods and was classified using ensemble and neural machine learning models. SHAP and t-SNE analyses were conducted for model interpretability and visualization.</div></div><div><h3>Results</h3><div>Radiomic analysis of non-tumorous anatomical regions demonstrated high diagnostic performance for PDAC detection, particularly in the pancreatic duct and parenchyma. Among the three approaches, patient-level feature aggregation (Approach 3) achieved the best results, with an F1-score of 88.33 % and AUC of 97.98 %. In contrast, deep features exhibited limited discriminative power when used in isolation but improved moderately in fusion strategies. SHAP and t-SNE analyses confirmed that tissue-wide radiomic signatures serve as robust biomarkers, supporting the hypothesis that PDAC induces detectable changes beyond the tumor region. These findings validate a novel, tumor-independent diagnostic framework for early PDAC classification.</div></div><div><h3>Conclusions</h3><div>Non-tumorous anatomical structures encode valuable diagnostic information for PDAC. Systemic feature integration provides a robust, interpretable framework for early detection, particularly in radiologically occult or ambiguous cases.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109056"},"PeriodicalIF":4.8,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145046510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jing Ning , Ming Ding , Zejiang Li , Minrui Cai , Xiuyan Liu , Yan Cai
{"title":"Quantitative correlation between small airway morphology with respiratory function during disease progression in COPD: CFD analysis of human airways based on CT and OCT imaging","authors":"Jing Ning , Ming Ding , Zejiang Li , Minrui Cai , Xiuyan Liu , Yan Cai","doi":"10.1016/j.cmpb.2025.109066","DOIUrl":"10.1016/j.cmpb.2025.109066","url":null,"abstract":"<div><h3>Background and Objective</h3><div>The quantitative knowledge of the influence of the small airway disease on the functional changes in chronic obstructive pulmonary disease (COPD) patients has been severely limited.</div></div><div><h3>Methods</h3><div>This study presents an innovative patient-specific computational framework that integrates CT and OCT imaging data with multiscale computational fluid dynamics (CFD) analysis. A three-dimensional tracheobronchial tree is reconstructed from CT scans of a mild COPD patient, spanning from the central airway to the 4th generation bronchial bifurcations. OCT imaging is subsequently conducted on upper, middle, and lower lobe bronchi of the right lung to quantify airway radius and wall thickness at 5th-9th generation bifurcations. These morphological parameters, hypothesized to correlate with small airway resistance and compliance, are implemented as impedance boundary conditions at the 3D model outlets.</div></div><div><h3>Results</h3><div>The simulation results demonstrate significant alterations in pressure gradients and velocity profiles under varying impedance conditions. The structure-function analysis quantify the morphological changes in small airways and their influences on the global respiratory function during disease progression. It is found that the relative residual volume (RV/TV) in the lung grows by up to 20 % from the early stage to the current stage of the disease. Additionally, the value of RV/TV may increase by up to 60 % if the radius of the 5th generation airway is halved.</div></div><div><h3>Conclusions</h3><div>By synergizing patient-specific geometry with impedance-adaptive boundary conditions derived from multimodal imaging, the framework facilitates accurate quantification of the structure-function relationships between small airway morphology and lung function, and enables patient-specific assessments for COPD patients.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109066"},"PeriodicalIF":4.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145010938","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xiaowei Qin , Zhibin Bi , Wenbin Li , Huipeng Zhang , Ming Han , Kongxi Zhang , Jian Wu , Lei Huang
{"title":"Machine learning-based plasma-derived extracellular vesicle signatures for digestive system cancers prediction","authors":"Xiaowei Qin , Zhibin Bi , Wenbin Li , Huipeng Zhang , Ming Han , Kongxi Zhang , Jian Wu , Lei Huang","doi":"10.1016/j.cmpb.2025.109064","DOIUrl":"10.1016/j.cmpb.2025.109064","url":null,"abstract":"<div><h3>Background</h3><div>Digestive system cancers (DSCs) represent a heterogeneous group of malignancies characterized by a poor prognosis and a lack of accurate early diagnostic methods. While traditional serological biomarkers and non-coding RNA continue to be commonly diagnostic marker for these cancers, their sensitivity and specificity in detection are often limited. RNA in plasma-derived extracellular vesicles (PDEV) has emerged as a promising diagnostic tool for a variety of cancers, but its application in the detection of various DSCs has not yet been fully explored.</div></div><div><h3>Methods</h3><div>By integrating PDEV sequencing data from the exoRBase 2.0 database, a total of 444 participants were included in the study, including 326 patients of DSCs, and 118 healthy individuals. The dataset was divided into training and test sets. The PDEV-diagnostic model was constructed using various machine learning algorithms and underwent 5-fold cross-validation in the training sets. The model's performance metrics were further evaluated in the test set. Additionally, the features were assessed using bulk RNA-seq and single RNA-seq datasets for different DSCs.</div></div><div><h3>Results</h3><div>Based on various feature selection methods and a comparison of 10 machine learning algorithms using seven metrics, the XGBoost model was selected as the PDEV-diagnostic model, with an AUC of 0.83 and 0.94 in the training and test sets, respectively, and 9 exosome predictors, including BANK1, MALAT1, FGA, UBR4, ILR-7,FGB, PLPP5,PCAT19, and CIITA for DSCs prediction.</div></div><div><h3>Conclusions</h3><div>The machine learning-based PDEV diagnostic models exhibit remarkable accuracy in identifying patients of DSCs. These nine exosomal mRNAs/lncRNAs consequently showed promise as non-invasive biomarkers for DSCs diagnosis.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109064"},"PeriodicalIF":4.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145061185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Evi M.C. Huijben , Sina Amirrajab , Josien P.W. Pluim
{"title":"Enhancing reconstruction-based out-of-distribution detection in brain MRI with model and metric ensembles","authors":"Evi M.C. Huijben , Sina Amirrajab , Josien P.W. Pluim","doi":"10.1016/j.cmpb.2025.109045","DOIUrl":"10.1016/j.cmpb.2025.109045","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Out-of-distribution (OOD) detection is crucial for safely deploying automated medical image analysis systems, as abnormal patterns in images could hamper their performance. However, OOD detection in medical imaging remains an open challenge. In this study, we aim to optimize a reconstruction-based autoencoder specifically for OOD detection. We address three gaps: the underexplored potential of a simple OOD detection model, the lack of optimization of deep learning strategies specifically for OOD detection, and the selection of appropriate reconstruction metrics.</div></div><div><h3>Methods:</h3><div>We investigated the effectiveness of a reconstruction-based autoencoder for unsupervised detection of synthetic local and global artifacts in brain MRI. We evaluated the general reconstruction capability of the model, analyzed the impact of the selected training epoch and reconstruction metrics, assessed the potential of model and/or metric ensembles, and tested the model on a dataset containing a diverse range of artifacts.</div></div><div><h3>Results:</h3><div>Among the metrics assessed, the learned perceptual image patch similarity (LPIPS) and the contrast component of structural similarity index measure (SSIM) consistently outperformed others in detecting homogeneous circular anomalies. By combining two well-converged models and using LPIPS and contrast as reconstruction metrics, we achieved a pixel-level area under the Precision–Recall curve of 0.66. Furthermore, with the more realistic OOD dataset, we observed that the detection performance varied between artifact types; local artifacts were more difficult to detect, while global artifacts showed better detection results.</div></div><div><h3>Conclusions:</h3><div>Our study shows that a reconstruction-based autoencoder, when combined with appropriate metrics, enhances OOD detection in brain MRI. These findings emphasize the importance of carefully selecting metrics and model configurations and highlight the need for tailored approaches, as standard deep learning approaches do not always align with the unique challenges of OOD detection. Improving OOD detection can increase the reliability of automated medical image analysis.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109045"},"PeriodicalIF":4.8,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145010937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Oluwatunmise Akinniyi , Jose Dixon , Joseph Aina , Francesca Weaks , Gehad A. Saleh , Md Mahmudur Rahman , Timothy Meeker , Hari Trivedi , Judy Wawira Gichoya , Fahmi Khalifa
{"title":"The role of AI for improved management of breast cancer: Enhanced diagnosis and health disparity mitigation","authors":"Oluwatunmise Akinniyi , Jose Dixon , Joseph Aina , Francesca Weaks , Gehad A. Saleh , Md Mahmudur Rahman , Timothy Meeker , Hari Trivedi , Judy Wawira Gichoya , Fahmi Khalifa","doi":"10.1016/j.cmpb.2025.109036","DOIUrl":"10.1016/j.cmpb.2025.109036","url":null,"abstract":"<div><div>Breast Cancer (BC) remains a leading cause of morbidity and mortality among women globally, accounting for 30% of all new cancer cases (with approximately 44,000 women dying), according to recent American Cancer Society reports. Therefore, accurate BC screening, diagnosis, and classification are crucial for timely interventions and improved patient outcomes. The main goal of this paper is to provide a comprehensive review of the latest advancements in BC detection, focusing on diagnostic BC imaging, Artificial Intelligence (AI) driven analysis, and health disparity considerations. We first examine diverse imaging techniques such as Mammography, Ultrasound, and Dynamic Contrast-Enhanced Magnetic Resonance Imaging, and provide an overview of their pros and cons. Then, we provided an intensive review of the State-of-the-Art (SOTA) literature on the role of AI in BC classification and segmentation. Lastly, we examined the role of AI in BC health disparities. A key contribution of this work lies in its integrative approach, consolidating insights from multiple research areas, imaging methods, AI-driven methodologies, and health disparities in a single resource. This paper evaluates the effectiveness of modern AI-based tools in enhancing diagnostic accuracy and discusses their potential to address biases in BC diagnosis, thus promoting equitable healthcare access. By integrating clinical, technical, and equity perspectives, this review aims to inform real-world decision-making, supporting the development of bias-aware AI tools, guiding equitable screening policy, and enhancing clinical practice in breast cancer care. Additionally, our critical analysis and discussion of recent SOTA highlights the strengths, limitations, and knowledge gaps for future directions of AI roles in BC. In total, these findings and future venue suggestions serve as a practical reference for researchers, clinicians, and policymakers, underscoring the need for interdisciplinary collaboration to harness AI’s full potential in BC diagnosis and reduce global health disparities.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109036"},"PeriodicalIF":4.8,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145019204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Shuang Leng , Jianguo Chen , Xulei Yang , Ru-San Tan , Ping Chai , Lynette Teo , James Yip , Ju Le Tan , Liang Zhong
{"title":"MPTN: A video-based multi-point tracking network for atrioventricular junction detection and tracking in cardiovascular magnetic resonance imaging","authors":"Shuang Leng , Jianguo Chen , Xulei Yang , Ru-San Tan , Ping Chai , Lynette Teo , James Yip , Ju Le Tan , Liang Zhong","doi":"10.1016/j.cmpb.2025.109048","DOIUrl":"10.1016/j.cmpb.2025.109048","url":null,"abstract":"<div><h3>Background and Objective</h3><div>To develop an end-to-end artificial intelligence solution—video-based Multi-Point Tracking Network (MPTN), for detecting and tracking atrioventricular junction (AVJ) points from cardiovascular magnetic resonance and deriving AVJ motion parameters.</div></div><div><h3>Methods</h3><div>The MPTN model consists of two modules: AVJ point detection and AVJ motion tracking. The detection module utilizes convolutional-based feature extraction and elastic regression to detect all candidate AVJ points. The tracking module employs the optimized DeepSORT algorithm to dynamically capture spatio-temporal continuity between cardiac frames. The model was trained and evaluated on datasets from 286 subjects, including normal controls and patients with heart failure, acute myocardial infarction, pulmonary arterial hypertension, and repaired tetralogy of Fallot. AVJ motion parameters, including systolic velocity S’, early diastolic velocity E’, late diastolic velocity A’, and displacements, were derived from tracked trajectories.</div></div><div><h3>Results</h3><div>Our MPTN model demonstrated promising performance compared to ground truth, with correlations of 0.92 for S’, 0.93 for E’, 0.89 for A’ in mitral annular motion velocities, and 0.93 for mitral annular plane systolic excursion. For tricuspid annular motion, the correlations were 0.91 for S’, 0.90 for E’, 0.87 for A’, and 0.86 for tricuspid annular plane systolic excursion. The MPTN-derived AVJ motion parameters exhibited strong diagnostic capabilities in detecting echocardiography-derived ventricular systolic and diastolic dysfunction, with an area under the curve ranging from 0.83 to 0.88 and accuracies ranging from 78 % to 85 %.</div></div><div><h3>Conclusions</h3><div>Our work provides an initial framework for cardiac motion tracking and function evaluation, which may support future advances in diagnosis of heart diseases.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109048"},"PeriodicalIF":4.8,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145004987","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}