Biomedical engineering advances最新文献

{"title":"Classification of PSQA outcomes in prostate VMAT treatments: a comparative study of machine learning models","authors":"Francis C. Djoumessi Zamo ,&nbsp;Alexandre Ngwa Ebongue ,&nbsp;Daniel Bongue ,&nbsp;Maurice Moyo Ndontchueng ,&nbsp;Christopher F. Njeh","doi":"10.1016/j.bea.2026.100206","DOIUrl":"10.1016/j.bea.2026.100206","url":null,"abstract":"<div><h3>Objective</h3><div>The present study aimed to construct and compare machine learning (ML) models for classifying patient-specific quality assurance (PSQA) outcomes in prostate volumetric modulated arc therapy (VMAT) treatment.</div></div><div><h3>Methods</h3><div>A total of 1247 prostate VMAT plans were retrospectively analyzed and several metrics and anatomical information extracted from the RT-plans files and contours were used as predictive variables. The following machine learning (ML) models: Logistic Regression (LR), Decision Trees (DT), Random Forest (RF), Gradient Boosting (GB), AdaBoost (AB), k-Nearest Neighbors (KNN), Naïve Bayes (NB) and Neural Network (NN) were developed to classify the PSQA outcomes, and their performances compared with different metrics.</div></div><div><h3>Results</h3><div>The results demonstrated that Random Forest and Gradient Boosting achieved the highest classification accuracy with area under the curve (AUC) values of 0.95 and 0.96, respectively and Accuracy of 0.91 for both classifiers. These models effectively balanced precision and accuracy while minimizing false negative and false positives rates which is critical for identifying potentially unsafe treatment plans.</div></div><div><h3>Conclusion</h3><div>ML-based PSQA classification (Random Forest and Gradient Boosting) demonstrates strong potential for optimizing quality assurance in prostate VMAT treatments. By integrating predictive analytics into clinical workflows, radiation oncology departments can improve efficiency, reduce resource demands, and enhance patient safety, paving the way for more adaptive and automated QA protocols.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100206"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978018","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Audio-based deep learning classification of laryngeal pathologies with detection of precancerous and cancerous lesions using Gammatone Cepstral coefficients","authors":"Julia Zofia Tomaszewska ,&nbsp;Wojciech Kukwa ,&nbsp;Apostolos Georgakis","doi":"10.1016/j.bea.2026.100211","DOIUrl":"10.1016/j.bea.2026.100211","url":null,"abstract":"<div><h3>Introduction</h3><div>Despite extensive research on audio-based voice pathology detection, current literature lacks clear and consistent evidence identifying acoustic features capable of reliably discriminating precancerous and cancerous laryngeal lesions, particularly when analysed using continuous speech signals.</div></div><div><h3>Problem statement</h3><div>The performance of audio-based laryngeal pathology classification systems on continuous speech remains significantly underreported, and commonly used Mel-Frequency Cepstral Coefficients (MFCCs) may be suboptimal for capturing pathology-related acoustic characteristics.</div></div><div><h3>Objectives</h3><div>This study investigates the hypothesis that continuous speech audio signals analysed with Gammatone Cepstral Coefficients (GTCCs) enable the accurate and precise detection of laryngeal pathologies, with the specific focus on precancerous and cancerous lesions.</div></div><div><h3>Methods</h3><div>An audio-based classification system employing GTCCs for feature extraction and a one-dimensional Convolutional Neural Network (CNN) for classification is proposed. The system considers three classes: precancerous and cancerous lesions, neuromuscular disorders, and healthy cases. Performance was evaluated using two datasets: a custom speech dataset collected for this research and the Saarbruecken Voice Database (SVD).</div></div><div><h3>Results</h3><div>GTCCs derived from speech signals delivered superior classification accuracy compared to the widely used Mel-Frequency Cepstral Coefficients (MFCCs). On the custom dataset, the proposed method achieved an average classification accuracy of 85.04% ±1.23 compared to 63.22% ± 1.62 using MFCCs. On SVD, GTCCs achieved 73.93% ±1.42, compared to 60.36% ±2.44 for MFCCs. The statistical significance of the obtained results was evidenced using <em>t</em>-test with the significance level set at 1%.</div></div><div><h3>Conclusions</h3><div>The results demonstrate that GTCCs extracted from continuous speech signals provide a robust and effective representation for audio-based laryngeal pathology classification, highlighting their potential for use in automated pre-screening systems targeting precancerous and cancerous voice disorders.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100211"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multimodal EMG–IMU sensor fusion with dual-output LSTM for fatigue estimation during neonatal chest compressions","authors":"Prashant Purohit ,&nbsp;John R. LaCourse","doi":"10.1016/j.bea.2026.100209","DOIUrl":"10.1016/j.bea.2026.100209","url":null,"abstract":"<div><h3>Background</h3><div>During neonatal cardiopulmonary resuscitation (NCPR), rescuer fatigue develops rapidly and compromises compression quality. Conventional feedback systems infer fatigue indirectly from mechanics (depth/rate) and may miss early neuromuscular changes.</div></div><div><h3>Objective</h3><div>To develop and evaluate a multimodal framework that fuses surface EMG (physiology) and IMU (biomechanics) to improve the accuracy of (i) fatigue level classification and (ii) prediction of fatigue-onset time during neonatal chest compressions (NCPR).</div></div><div><h3>Methods</h3><div>Twenty trained providers performed simulated neonatal compressions on a manikin while synchronized EMG (deltoid, triceps, upper trapezius) and 3-axis IMU signals were recorded and windowed (2 s, 50% overlap). Features included EMG RMS, MAV, median frequency (MF), and IMU depth dynamics. A dual-output Long Short-Term Memory (LSTM) jointly produced 3-class fatigue labels and onset-time regression.</div></div><div><h3>Results</h3><div>Fusion outperformed unimodal models: 98.3% accuracy, macro-F1 0.982, AUC 0.99; onset prediction RMSE 38.3 s, R² 0.68. EMG-only: 69.4% accuracy; IMU-only: 96.7%. EMG provided early physiological fatigue signatures, complementing IMU mechanical degradation.</div></div><div><h3>Conclusion</h3><div>EMG–IMU fusion with temporal deep learning improves fatigue estimation during NCPR and is suitable for real-time feedback to support optimal rescuer rotation. Earlier, physiology-aware fatigue detection enables proactive team management before compression quality declines. Lightweight LSTM fusion runs in real time and generalizes across rescuers.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100209"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Influence of stance width and foot rotation on muscle activity and ground reaction forces during squats in female lacrosse players","authors":"Anuradhi Bandara,&nbsp;Shinichi Kawamoto,&nbsp;Mona Makita,&nbsp;Momoko Nagai-Tanima,&nbsp;Tomoki Aoyama","doi":"10.1016/j.bea.2026.100210","DOIUrl":"10.1016/j.bea.2026.100210","url":null,"abstract":"<div><div>Female lacrosse players experience a high burden of non-contact lower-extremity injuries, and squat-based neuromuscular training is commonly used to develop lower-limb strength, coordination, and load tolerance. How common technique modifications such as stance width and foot rotation affect joint excursion, muscle activation, ground reaction forces (GRFs), and inter-limb loading symmetry during squats in this population remains unclear. This study aimed to examine the effects of stance width and foot rotation on lower-limb joint excursion, surface electromyography (sEMG), GRFs, and inter-limb GRF asymmetry during bodyweight squats in female lacrosse players. Ten Japanese university-level female lacrosse players performed squats under six stance conditions (narrow, shoulder-width, wide × parallel or external rotation). sEMG was recorded from ten lower-limb muscles, synchronized with 3D kinematics and bilateral force plates. Friedman tests with false discovery rate (FDR) correction evaluated stance-related differences, with effect sizes estimated using Kendall’s W. Stance significantly influenced hip (χ² = 33.31, <em>p</em> &lt; 0.001, <em>W</em> = 0.67), knee (χ² = 19.94, <em>p</em> = 0.001, <em>W</em> = 0.40), and ankle (χ² = 24.23, <em>p</em> &lt; 0.001, <em>W</em> = 0.49) joint excursion. Wide external rotation (WidER) yielded the greatest hip joint flexion–extension excursion (116.9° ± 7.9°), whereas narrow parallel stance (NarPar) produced the smallest (98.2° ± 4.8°). During the descending phase, gluteus maximus activation was significantly higher in wide stances compared with narrow and shoulder-width conditions (<em>q</em> &lt; 0.013). GRFs showed consistent vertical peaks across stances (∼56–62 % body weight), but mediolateral peaks were substantially higher in WidER (∼17 % body weight) than in narrow stances (∼5–7 % body weight). In female lacrosse players, squat stance meaningfully modulates mechanics even under bodyweight loading. WidER squats preferentially increase hip excursion, gluteus maximus activation, and global mediolateral GRFs, whereas narrow parallel squats increase ankle dorsiflexion demands and are associated with greater vertical loading asymmetry. These findings support tailoring stance width and foot rotation to target hip-dominant strength and frontal-plane control versus ankle mobility demands within lacrosse-oriented neuromuscular training.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100210"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146078567","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Suture-less vascular anastomosis – State of the art, challenges and perspectives: A review","authors":"John Ser Pheng Loh ,&nbsp;Yijia Yuan ,&nbsp;Kuan-Che Feng ,&nbsp;Robert Heymann ,&nbsp;Justin Kok Soon Tan ,&nbsp;Lars Rasmusson ,&nbsp;Hwa Liang Leo ,&nbsp;Reinhilde Jacobs","doi":"10.1016/j.bea.2026.100207","DOIUrl":"10.1016/j.bea.2026.100207","url":null,"abstract":"<div><div>Vascular microanastomosis is a cornerstone of reconstructive, transplantation, and cardiothoracic surgery, enabling the reconnection of tiny blood vessels to restore circulation to transplanted tissues or donor organs. Traditionally, this delicate process relies on hand suturing under microscopic magnification, a demanding and time-consuming technique that requires extensive training and precision. Despite decades of refinement, conventional suturing remains limited by human factors. Technical errors can lead to thrombosis, tissue necrosis, hematoma, or flap loss, with serious implications for patient outcomes and healthcare costs. In response, suture-less anastomosis methods have emerged as promising alternatives. These devices use mechanical couplers, rings, clips, magnets, or bioengineered scaffolds to connect vessel ends rapidly and reproducibly, aiming to reduce operative time, minimize ischemia, and improve procedural consistency across surgical teams. Recent innovations have introduced biodegradable and intraluminal designs that reduce foreign-body reactions, lower palpability, and accommodate vessel growth, offering distinct advantages in paediatric and long-term reconstructive settings.</div><div>Despite these advances, the widespread clinical adoption of suture-less technologies remains constrained by unresolved challenges. Key considerations include ensuring mechanical stability under physiological pulsation, optimizing biocompatibility to prevent thrombosis at the junction, and adapting device geometries to the diversity of vessel sizes and wall structures encountered in clinical practice. Continued translational research is needed to refine materials, simplify deployment mechanisms, and integrate these systems seamlessly into microsurgical workflows. This review synthesizes current developments in suture-less vascular anastomosis, critically evaluating their benefits and limitations across experimental and clinical studies. It also identifies future research priorities at the intersection of materials science, additive manufacturing, and surgical engineering. As these disciplines converge, next-generation suture-less devices hold the potential to redefine vascular repair by making micro-anastomosis faster, safer, and more accessible, thus transforming reconstructive and transplant surgery for patients who depend on these life-saving procedures.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100207"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978019","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A predictive model for birth delivery mode utilizing shape characteristics of uterine contractions from electrohysterographic signals and obstetric parameters","authors":"Rubana H. Chowdhury ,&nbsp;Quazi D. Hossain ,&nbsp;Mohiuddin Ahmad","doi":"10.1016/j.bea.2026.100212","DOIUrl":"10.1016/j.bea.2026.100212","url":null,"abstract":"<div><div>Birth delivery outcomes have a substantial impact on maternal and neonatal health. The objective of this work is to develop a predictive model that can identify the method of delivery during active labor, specifically by identifying contraction bursts, so that physicians can make decisions more quickly. We trained and assessed our model using the Icelandic 16-electrode EHG Database, comprising 122 abdominal EHG recordings from 45 pregnant women (collected in Iceland between 2008 and 2010), including both third-trimester prenatal and labor recordings. Contractions were detected using the examination of zero-crossing rate (ZCR) and root mean square (RMS) of uterine electrohysterographic (EHG) data. By applying the discrete Fourier transform (DFT) to each contraction burst, we obtained geometric features. These, in conjunction with seven standardized obstetric parameters, served as inputs to a Random Forest (RF) classifier. 4 channels on the upper left of the uterus from the 16-channel EHG database showed the best consistency index (average CCI&gt; 90%). The proposed model attained an accuracy of 99% (95% CI: 0.93–1.00) in distinguishing between cesarean and vaginal deliveries, and an accuracy of 94% (95% CI: 0.91–1.00) in differentiating spontaneous from induced vaginal deliveries. Analysis of feature importance indicated that shape-based UC features exhibited greater predictive power than certain conventional obstetric variables. The combination of UC’s geometry with obstetric data creates a robust, interpretable, and non-invasive framework for predicting delivery mode. The results suggest that this method may serve as an effective clinical decision-support tool to reduce unnecessary cesarean sections and improve delivery planning.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100212"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146188990","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A fused attention-based hybrid model for semi-supervised medical image segmentation","authors":"Masum Shah Junayed,&nbsp;Sheida Nabavi","doi":"10.1016/j.bea.2025.100203","DOIUrl":"10.1016/j.bea.2025.100203","url":null,"abstract":"<div><div>Addressing the significant challenge of extensive data annotation in medical image segmentation, semi-supervised techniques have become an effective alternative for leveraging both labeled and unlabeled data. However, accurately capturing complex local structures and global contextual information remains difficult, often leading to inconsistent segmentation results. To mitigate these limitations, we propose a fused transformer-based hybrid architecture for semi-supervised medical image segmentation. The model integrates a parallel backbone comprising Deformation Convolution Blocks (DCB) and Fused Transformer Blocks (FTB), followed by a segmentation head for precise mask prediction. The DCB enhances spatial adaptability for irregular and artifact-rich regions, while the FTB – with its fused attention mechanism – captures long-range dependencies efficiently. A Ghost Layer Perceptron (GLP) embedded within the transformer further improves computational efficiency without compromising representation quality. In addition, the incorporation of a consistency loss and unsupervised contrastive learning facilitates robust feature discrimination on unlabeled data, improving generalization across modalities. Extensive experiments on four publicly available medical imaging datasets demonstrate that the proposed model achieves comparable or better accuracy than recent state-of-the-art methods, while requiring substantially fewer parameters and lower computational cost, underscoring its practicality for real-world clinical applications.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100203"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146188992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel pipeline for converting surface electromyography signals into muscle activations","authors":"P. Tsakonas ,&nbsp;N.D. Evans ,&nbsp;J. Hardwicke ,&nbsp;M.J. Chappell","doi":"10.1016/j.bea.2025.100204","DOIUrl":"10.1016/j.bea.2025.100204","url":null,"abstract":"<div><div>This study introduces a novel pipeline for converting surface electromyography (sEMG) signals into muscle activations using the Hilbert-Huang Transform. Traditional approaches in this context often apply low-pass filters that suppress high-frequency components, potentially discarding physiologically relevant signal information. In contrast, the proposed method leverages Empirical Mode Decomposition and Hilbert spectral analysis to preserve the nonstationary and multi-frequency nature of sEMG data. Activation outputs are then mapped through physiologically inspired dynamics, yielding time-resolved muscle activations. Comparative analyses were conducted across three muscles (EDC, FDS, FDP) using data from 10 subjects each performing 5 cylindrical grasps. Intra-subject comparisons using Wilcoxon signed-rank tests revealed statistically significant improvements (p &lt; 0.001) in nearly all trials. Linear mixed-effects analysis of log-transformed activations showed that the new pipeline yields significantly higher muscle activations within each muscle: EDC GMR = 1.31 (95% CI: 1.255–1.359), FDS GMR = 1.35 (95% CI: 1.296–1.396), and FDP GMR = 1.29 (95% CI: 1.248–1.342), all p &lt; 0.001. These results suggest that the choice of sEMG processing pipeline can meaningfully alter activation estimates and potentially influence musculoskeletal model estimation. The method presented provides a robust and physiologically consistent alternative for applications in biomechanics, prosthetic control, and neuromuscular modelling.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100204"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145760797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Use of the Geometric Pattern Transformation to detect diabetes mellitus type 2 from blood glucose signals","authors":"Marcos Maillot ,&nbsp;Ariel Amadio ,&nbsp;Cristian Bonini ,&nbsp;Leandro Robles Dávila ,&nbsp;Walter Legnani ,&nbsp;Dino Otero","doi":"10.1016/j.bea.2025.100205","DOIUrl":"10.1016/j.bea.2025.100205","url":null,"abstract":"<div><h3>Purpose</h3><div>Type 2 Diabetes Mellitus (DMT2) is a disease with a high incidence worldwide, and various estimates project an increase in the near future. This research introduces a new methodology based on blood glucose concentration recordings obtained from Continuous Glucose Monitors (CGM) for the detection of the disease. The main research question is whether it is possible to enhance the detection ability of DMT2 by applying Geometric Pattern Transformation (GPT) to the glucose records, compared to basic statistical tools.</div></div><div><h3>Methods</h3><div>The standard deviation of the glucose signal from continuous glucose monitoring (CGM) was evaluated as a parameter to distinguish between individuals with and without diabetes. The GPT technique was then applied to the glucose data to assess its effectiveness in enhancing disease detection compared to traditional statistical methods.</div></div><div><h3>Results</h3><div>The findings indicate that traditional statistical tools are insufficient to achieve the same performance as the proposed method for detecting DMT2. Applying GPT significantly improved detection accuracy, showing a clear advantage in differentiating between subjects with and without DMT2. Moreover, glucose monitoring over a period of at least 3 days proved to be as effective as longer periods for detection purposes.</div></div><div><h3>Conclusions</h3><div>The proposed methodology, using basic mathematical operations on glucose data, effectively distinguishes individuals with DMT2 from those without. The parameters used to assess detection quality demonstrate a marked improvement due to GPT. Additionally, a 3-day monitoring period is sufficient for reliable detection, potentially streamlining the diagnostic process.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100205"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939593","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Lung-Mamba: Lung nodule segmentation model optimized by Mamba’s selective state spaces","authors":"Hadrien T. Gayap,&nbsp;Moulay A. Akhloufi","doi":"10.1016/j.bea.2026.100214","DOIUrl":"10.1016/j.bea.2026.100214","url":null,"abstract":"<div><div>The low five-year survival rate for lung cancer underscores the importance of early detection. A key component of this process is the accurate segmentation of pulmonary nodules from CT scans to quantify their characteristics. While deep learning models have advanced this field, Transformer-based architectures face limitations due to their high computational complexity with high-resolution medical images. This paper introduces Lung-Mamba, a deep learning model for lung nodule segmentation that combines a U-Net framework with the recently proposed Mamba architecture. Mamba utilizes Selective State Spaces to model long-range dependencies with linear complexity, offering an efficient alternative to Transformers. The proposed architecture integrates Mamba layers into a U-Net to capture both local features and global context. Evaluated on the LIDC-IDRI dataset, using 12,465 nodules for training and 3117 for testing, Lung-Mamba achieves a Dice score of 96.48%. This result positions the model as an effective and computationally efficient method for medical image segmentation, demonstrating the benefit of integrating state-space models into established convolutional frameworks.</div></div>","PeriodicalId":72384,"journal":{"name":"Biomedical engineering advances","volume":"11 ","pages":"Article 100214"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146188991","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}