Computers in biology and medicine最新文献

{"title":"Mechano-electric feedback reduces the occurrence of delayed-afterdepolarization-driven focal activity","authors":"Navneet Roshan,&nbsp;Rahul Pandit","doi":"10.1016/j.compbiomed.2025.111436","DOIUrl":"10.1016/j.compbiomed.2025.111436","url":null,"abstract":"<div><div>Delayed afterdepolarizations (DADs), which occur during the diastolic phase of a cardiomyocyte action potential (AP), are frequently observed under specific pathophysiological conditions. The synchronization of DAD-capable myocytes can effectively overcome the inherent source-sink mismatch with adjacent normal myocytes, so it is an important mechanism for the genesis of premature ventricular contractions (PVCs). Our study elucidates the role of <em>mechano-electrical feedback</em> in modulating this critical source-sink requirement and its interplay with diffusional anisotropy in cardiac tissue. We combine the ten Tusscher-Panfilov 06 (TP06) electrophysiological model for human ventricular myocytes with cardiac-tissue-mechanics models and account for spontaneous calcium releases (SCRs) and the randomness associated with these and with the disordered arrangement of DAD myocytes in a clump. Our work leads to a quantification of the time dependence of the intracellular cytosolic calcium transient <span><math><mi>C</mi><msub><mi>a</mi><mtext>i</mtext></msub></math></span> and the active-contraction ratio <span><math><msub><mi>λ</mi><mtext>o</mtext></msub></math></span>, without and with SCRs and shows explicitly how the parameter <span><math><msub><mi>g</mi><mrow><mtext>spon</mtext></mrow></msub></math></span>, which controls the strength of the SCRs, affects DADs and <span><math><msub><mi>λ</mi><mtext>o</mtext></msub></math></span>. We then illustrate the spatiotemporal evolution of the PVCs in the presence of domain deformation and a clump of DAD-capable myocytes. In a cable-type domain we demonstrate how contraction and elongation of the domain, the coupling interval of the DADs with the previous AP, and the width of the probability distribution function (PDF) of the coupling interval influence the source-size requirement for the formation of PVCs. We then extend our cable results to two dimensions (2D) by calculating the effect of the mechano-electric feedback on the origin and evolution of DAD-induced PVCs in 2D tissue. Our work leads to the following insights on the interplay of mechanical deformation and electrophysiology in DAD-induced PVCs: (1): mechano-electrical feedback markedly reduces the source requirement, which we quantify by the size of a clump of DAD-myocytes; (2): an extended coupling interval of DADs, relative to the preceding AP, reduces the source requirement; (3): sparse distribution of DAD-myocytes in a clump and the distribution in their coupling intervals, along with mechano-electric feedback, reduces the chances of the DAD-induced PVCs; (4): this suppression of PVCs is more pronounced in 2D than in 1D domains.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111436"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145916856","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":"Accurate prediction of anticancer peptides using a stacking ensemble of convolutional and transformer models with conjoint sequence representations","authors":"Huynh Anh Duy ,&nbsp;Phurinut Khampasri ,&nbsp;Pimmada Janthanet ,&nbsp;Patlissa Pattiyamongkhonkul ,&nbsp;Tarapong Srisongkram","doi":"10.1016/j.compbiomed.2026.111463","DOIUrl":"10.1016/j.compbiomed.2026.111463","url":null,"abstract":"<div><div>We present a high-performance predictive framework for anticancer peptide (ACP) identification, based on a stacking ensemble learning approach that synergistically combines convolutional neural networks and transformer models using a random forest as a meta-classifier. This architecture is driven by conjoint sequence representations that integrate both one-hot encoding and pre-trained evolutionary scale modeling embeddings, enabling the extraction of complementary local and global features from peptide sequences. Our proposed model achieved a robust accuracy of 88.9% on the primary ACP data set, while maintaining competitive or superior performance across multiple external benchmark data sets, with accuracies ranging from 83.2% to 95.2%, highlighting its strong generalization capability on par with the state-of-the-art models. To demonstrate translational relevance, the model was applied to a curated set of clinically approved and candidate ACPs, producing probabilistic scores to support experimental prioritization. To further enhance model interpretability, SHapley Additive exPlanations analysis was employed, revealing lysine as a consistently influential residue, alongside other positively charged and hydrophobic amino acids. These findings not only corroborate known mechanistic insights into ACP-membrane interactions but also highlight the utility of model-derived feature importance in guiding peptide design. Taken together, this work introduces a robust, interpretable, and generalizable approach for computational ACP prediction, offering valuable implications for peptide-based anticancer drug discovery. To enhance the accessibility and translational potential of our model, we developed an interactive web-based prediction tool, named <em>ACPredictor</em>, for the identification of ACPs. This platform is freely available at <span><span>https://acpredictor.streamlit.app/</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111463"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974291","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":"Unravelling the structural impact of progesterone receptor mutations in myoma and progesterone intolerance through computational modeling","authors":"F. Saritha ,&nbsp;R. Aswath Kumar ,&nbsp;K.V. Dileep","doi":"10.1016/j.compbiomed.2026.111476","DOIUrl":"10.1016/j.compbiomed.2026.111476","url":null,"abstract":"<div><div>Progesterone (P4) is a steroid hormone involved in the regulation of female reproductive functions. The endogenous progesterone receptor (PR), a member of the nuclear receptor family of ligand-dependent transcription regulators responsible for P4 action in the body through the ‘ligand binding domain’ (LBD). PR isoforms, PR-A and PR-B, are encoded by a single gene, PGR and variations in this gene can disrupt cellular signaling. In the current study, putative disease-causing mutations on PR has been identified through computationally and its mechanistic effects were explored using structural bioinformatics tools. Studies suggested that 11 of 66 missense variants (within the LBD) induce structural destabilization and were identified as potentially deleterious. Our ensemble docking suggested that these variations have a limited impact on P4 binding, however they significantly disrupt the binding of co-activators as evident by the protein-peptide docking. The binding of co-activators to the PR is the determining factor for the P4 signaling. Finally, based on the free energy of binding, we proposed two variations such as R869H and C798Y could cause myoma and progesterone tolerance conditions respectively. These findings were further validated through the use of allostery predictions. Our results reveal distinct mechanisms by which PR mutations modulate receptor function, laying the framework for future mechanistic studies and therapeutic development for PR-associated reproductive disorders.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111476"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974290","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":"Reliable leukemia detection via transfer-enhanced Bayesian CNNs","authors":"Xhesina Hita ,&nbsp;Farrukh Javed ,&nbsp;Stefano Lodi","doi":"10.1016/j.compbiomed.2025.111419","DOIUrl":"10.1016/j.compbiomed.2025.111419","url":null,"abstract":"<div><div>Accurate and early detection of Acute Lymphoblastic Leukemia (ALL) is critical for timely intervention and improved patient outcomes. However, the development of reliable deep learning models for hematological image analysis is challenged by limited data availability, dataset bias, and the need for trustworthy predictions in clinical settings. In this study, we propose a Bayesian deep learning framework that integrates transfer learning, data augmentation, and uncertainty quantification for robust classification of leukemic and healthy lymphocytes from peripheral blood smear images. Three widely used convolutional neural network architectures, InceptionV3, VGG16, and ResNet50, pretrained on ImageNet are fine-tuned on the ALL-IDB2 dataset and extended with Monte Carlo dropout to enable Bayesian inference. Model performance is evaluated using 10-fold cross-validation on both original and augmented datasets, with accuracy, sensitivity, specificity, Youden’s index, and Brier score used as evaluation metrics. Among the evaluated models, VGG16 demonstrates the most consistent improvements under data augmentation, achieving the highest accuracy (<span><math><mn>98.65</mn><mi>%</mi><mo>±</mo><mn>0.09</mn></math></span>), Youden’s index (<span><math><mn>0.97</mn><mo>±</mo><mn>0.001</mn></math></span>) and Brier score (<span><math><mn>0.035</mn><mo>±</mo><mn>0.010</mn></math></span>), while ResNet50 shows strong but more moderate gains. In contrast, InceptionV3 exhibits limited sensitivity to augmentation and comparatively lower robustness. Beyond average predictive performance, Bayesian uncertainty analysis reveals that misclassifications and borderline predictions are consistently associated with elevated predictive entropy and mutual information. Saliency map inspection further indicates that high-uncertainty cases correspond to diffuse or non-localized attention patterns, suggesting reliance on spurious contextual features rather than stable morphological cues. These findings highlight the importance of uncertainty-aware predictions for identifying cases that may require expert pathological review. Overall, the proposed framework combines strong diagnostic performance with interpretable uncertainty estimates, supporting its role as a transparent and clinically trustworthy tool for AI-assisted leukemia screening.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111419"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145883525","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":"In-silico and In-vitro role of chimeric peptide on the impact of nsSNPs in human Cannabinoid receptors 1 and 2","authors":"Jerine Peter Simon ,&nbsp;Huiming Bao ,&nbsp;Zhanyu Niu ,&nbsp;Hongyang Man ,&nbsp;Shouliang Dong","doi":"10.1016/j.compbiomed.2026.111451","DOIUrl":"10.1016/j.compbiomed.2026.111451","url":null,"abstract":"<div><div>Cannabinoid receptors 1 and 2 (CNR1 and CNR2) play an important role in the endocannabinoid system and intracellular pathways. Their beneficial effects became destroyed in the event of some mutation. Identifying protective drugs for mutation is highly necessary to eliminate its destruction. The research aims to understand the role of chimeric peptides in CNR1 and CNR2 nsSNPs. The research objectives are to find destructive nsSNPs through <em>in-silico</em> prediction and protein-only state molecular dynamics simulation. To understand the role of chimeric peptides on interaction with nsSNPs through docking analysis and protein-ligand state molecular dynamics simulation. To validate <em>in-silico</em> findings through <em>in-vitro</em> cell expression experiments. The research methods involved 36 <em>in-silico</em> online prediction tools, simulation for 100 ns, 29 trajectory analysis modules, three docking programs, and western blotting techniques. Four CNR1 and six CNR2 nsSNPs were shortlisted as destructive nsSNPs using <em>in-silico</em> tools. Further, CNR1 N134T and CNR2 I298N nsSNPs were shortlisted as highly destructive nsSNPs using simulation and docking analysis. Beneficial peptides were shortlisted using docking analysis and simulation with wild type. This research found that the chimeric peptide MP-13 has the best binding affinity and dynamics properties with wild-type and nsSNPs. It also almost restores dynamics properties and improves binding affinity in nsSNPs. It is most effective in influencing CNR1 and CNR2 signaling responses despite nsSNPs. Based on <em>in-silico</em> and <em>in-vitro</em> analysis, the research concludes that MP-13 has the strongest effect on CNR1 and CNR2 nsSNPs. The workflow of the present research is represented in a graphical illustration.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111451"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923032","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":"Automating prostate biopsy guidance: A robust CNN approach for non-rigid 3D/3D MR-TRUS image registration","authors":"Thi Thao Ho ,&nbsp;Clément Beitone ,&nbsp;Jocelyne Troccaz ,&nbsp;Sandrine Voros","doi":"10.1016/j.compbiomed.2025.111423","DOIUrl":"10.1016/j.compbiomed.2025.111423","url":null,"abstract":"<div><div>Fusing transrectal ultrasound (TRUS) and magnetic resonance (MR) images has significantly improved the accuracy of prostate cancer detection during targeted biopsies. However, automatic MR-TRUS registration remains challenging due to the large anatomical and appearance differences between the two imaging modalities. This study introduces a fully automatic, weakly supervised deep learning (DL) approach for predicting dense displacement fields (DDFs) between 3D MR and 3D TRUS images without requiring segmentation images. The proposed method consists of two main stages: (1) a <span>dl</span>-based preprocessing step that aligns the prostate centroids of the MR and TRUS images to improve initialization, and (2) a UNet-inspired registration network (RegResUNet) combined with a spatial transformer layer (STL) that directly predicts voxel-level DDFs from the aligned 3D input images. The network is trained using weak supervision based on anatomical masks, enabling accurate registration without ground truth DDFs. Experimental results demonstrate that the proposed method achieves substantial improvements over rigid baseline registration, with an average surface registration error (SRE) of 0.97 <span><math><mo>±</mo></math></span> 0.85 mm and an average Dice similarity coefficient (DSC) of 0.93 <span><math><mo>±</mo></math></span> 0.02. Notably, the proposed network outperforms several state-of-the-art non-rigid registration models while maintaining computational efficiency. The whole pipeline eliminates the need for intermediate manual and time-consuming segmentation steps. The automatic and robust capabilities of the proposed approach, combined with its short inference time, highlight its potential for clinical use in prostate cancer interventions, reducing human factors, ensuring consistent results, and contributing to improved patient comfort by minimizing procedure time.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111423"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923037","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":"Molecular characterization of palmitoylation in prostate cancer reveals KIFC2 as a prognostic biomarker and potential therapeutic target","authors":"Liang Huang ,&nbsp;Shusuan Jiang ,&nbsp;Fuhua Zeng ,&nbsp;Gongqian Zeng ,&nbsp;Hong Shan","doi":"10.1016/j.compbiomed.2026.111450","DOIUrl":"10.1016/j.compbiomed.2026.111450","url":null,"abstract":"<div><div>Protein palmitoylation, a reversible post-translational lipid modification, has been implicated in regulating cancer cell signaling and progression; however, its role in prostate cancer (PCa) remains unclear. In this study, we comprehensively analyzed palmitoylation-related genes (PRGs) in PCa by integrating bulk transcriptomic, single-cell RNA sequencing, and spatial transcriptomic datasets. Unsupervised consensus clustering based on PRG expression identified two molecular subtypes with distinct prognoses, immune infiltration profiles, and pathway activities. Differential expression and weighted gene co-expression network analyses revealed five key feature genes, among which KIFC2 was highly expressed in tumor cells and correlated with poor clinical outcomes. The KIFC2 was predominantly enriched in high-grade adenocarcinoma regions. Functional experiments demonstrated that silencing KIFC2 significantly inhibited proliferation and promoted apoptosis in PC3 and DU145 prostate cancer cell lines. Additionally, high KIFC2 expression was associated with increased cell cycle progression and oncogenic signaling pathways, including KRAS and PI3K-AKT. Collectively, these results suggest that palmitoylation and KIFC2 play critical roles in PCa progression and may serve as promising biomarkers and therapeutic targets.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111450"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145951441","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":"Photoplethysmography imaging reveals frequency-specific microvascular responses to histamine in human skin","authors":"Stefan Borik ,&nbsp;Miriam Zemanikova ,&nbsp;Patrik Procka ,&nbsp;Jan Seleng ,&nbsp;Michal Labuda ,&nbsp;Aymen A. Alian","doi":"10.1016/j.compbiomed.2026.111464","DOIUrl":"10.1016/j.compbiomed.2026.111464","url":null,"abstract":"<div><h3>Background</h3><div>Skin prick testing (SPT) remains the gold standard for allergy diagnostics, yet it relies on subjective assessment of wheal size and provides limited insight into the underlying vascular mechanisms. In this study, the histamine positive control within the SPT framework was used as a standardized model to characterize frequency-specific microvascular responses using photoplethysmography imaging (PPGI).</div></div><div><h3>Methods</h3><div>PPGI data were analyzed from 16 healthy volunteers who underwent SPT with histamine and saline control sites. Vascular oscillations were analyzed in the cardiac (0.6–2.5 Hz), respiratory (0.15–0.6 Hz), and myogenic (0.05–0.15 Hz) frequency bands using 1-min segments to characterize time-resolved changes across the post-application stages.</div></div><div><h3>Results</h3><div>A significant suppression of myogenic oscillations was detected at the histamine site, with an 80 % reduction observed 10–15 min post-application and a 60 % reduction at 15–20 min relative to baseline. Respiratory-band energy showed no significant changes during the experiment. No significant changes were observed at the negative control sites. Skin blood-volume-oscillation maps supported these observations, showing localized reduction of myogenic vasomotion in the affected areas.</div></div><div><h3>Conclusion</h3><div>Histamine-induced allergic responses produce localized, frequency-specific alterations in cutaneous blood flow, characterized by a sustained suppression of myogenic oscillations accompanied with flare-related hyperemia. These results support the hypothesis of “vascular locking,” wherein vasodilation attenuates smooth muscle contractility, and highlight the potential of PPGI to enhance the diagnostic resolution of allergy testing by capturing vascular dynamics across multiple frequency bands.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111464"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145951492","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":"Designing of a multi-epitope vaccine targeting enterovirus D68: An integrated immunoinformatic and reverse vaccinology approach","authors":"Hailah M. Almohaimeed ,&nbsp;Amany I. Almars ,&nbsp;Nada Alkhorayef ,&nbsp;Ahmed M. Basri ,&nbsp;Fayez Alsulaimani ,&nbsp;Muhammad Shahbaz ,&nbsp;Fahad M. Alshabrmi ,&nbsp;Sarfaraz Alam ,&nbsp;Tahir Muhammad ,&nbsp;Muhammad Shahab","doi":"10.1016/j.compbiomed.2025.111433","DOIUrl":"10.1016/j.compbiomed.2025.111433","url":null,"abstract":"<div><div>Enterovirus D68 (EV-D68) is an enterovirus known for causing respiratory infections, as well as flaccid myelitis, meningitis and encephalitis. Despite the efforts, no licensed vaccine against EV-D68 is currently available. Vaccine development efforts are ongoing; however, the process is complex and requires extensive clinical validation. In contrast, immunoinformatic is a rapidly expanding area with the potential to significantly influence the therapeutic interventions and vaccine development for infectious diseases. Herein, immunoinformatic and reverse vaccinology strategies were utilized to design a multi-epitope vaccine construct targeting EV-D68 virus. In this connection, three virulent proteins were selected for analysis based on their immunogenic characteristics. Further, B-cells and T-cells epitopes were predicted and connected through suitable linkers and adjuvant. The predicted T-cell epitopes within the vaccine construct exhibited a significant worldwide population coverage. Moreover, Robetta was utilized to predict the 3D structure of the vaccine construct. Subsequently the molecular docking simulation of construct was employed to study the molecular interactions by using Toll-like receptors as target proteins and further subjected to MD simulation. The results reveal the stability of the vaccine-receptor complex throughout the simulation. Finally, <em>in silico</em> cloning showed potential for the predicted vaccine within the <em>Escherichia coli</em> expression system. These findings provide valuable insights that may guide subsequent experimental studies and contribute meaningfully to the early phases of EV-D68 vaccine research and development. By streamlining candidate selection and optimizing design parameters, our findings holds promise for accelerating the transition from computational predictions to effective vaccine formulations.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111433"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145882871","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":"Deep Laplacian Coordinates: End-to-end deeply guided anisotropic diffusion for COVID-19 pulmonary lesion segmentation","authors":"Aldimir Bruzadin,&nbsp;Marilaine Colnago,&nbsp;Lucas C. Ribas,&nbsp;Wallace Casaca","doi":"10.1016/j.compbiomed.2025.111431","DOIUrl":"10.1016/j.compbiomed.2025.111431","url":null,"abstract":"<div><div>Despite notable advances in deep learning, accurately segmenting lung lesions in computed tomography remains a significant challenge due to the scarcity of annotated data and the high diversity in lesion appearance. To address these issues, seeded image segmentation stands out as a flexible and accurate approach, adapting to diverse image contexts and target definitions. Building on this perspective, we introduce the <em>Deep Laplacian Coordinates Neural Network</em> (DLCNN): a novel framework that integrates deep boundary detection, anisotropic diffusion and seed-driven labeling to segment lung lesions caused by COVID-19. DLCNN employs a semantically enriched deep contour network that predicts edge weights for a graph-based image representation. These weights are then incorporated into our label propagation model, which is built upon the Laplacian Coordinates diffuser, leveraging many attractive properties such as global optimality, robust boundary delineation and directionally adaptive diffusion. By combining the representational power of deep boundary learning with the generalizability of a seed-driven anisotropic diffusion model, the proposed framework accurately captures lung lesions, even when boundaries are poorly defined. DLCNN consistently outperforms both recent and state-of-the-art marker-based segmentation methods, as confirmed by extensive quantitative and qualitative analyses, particularly in complex scenarios involving low contrast and irregular lesion shapes.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"202 ","pages":"Article 111431"},"PeriodicalIF":6.3,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145882874","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}