Computers in biology and medicine最新文献

{"title":"hERGBoost: A gradient boosting model for quantitative IC50 prediction of hERG channel blockers","authors":"Myeong-Sang Yu ,&nbsp;Jingyu Lee ,&nbsp;Yunhyeok Lee ,&nbsp;Daeahn Cho ,&nbsp;Kwang-Seok Oh ,&nbsp;Jidon Jang ,&nbsp;Nuong Thi Nong ,&nbsp;Hyang-Mi Lee ,&nbsp;Dokyun Na","doi":"10.1016/j.compbiomed.2024.109416","DOIUrl":"10.1016/j.compbiomed.2024.109416","url":null,"abstract":"<div><div>The human ether-a-go-go-related gene (hERG) potassium channel is pivotal in drug discovery due to its susceptibility to blockage by drug candidate molecules, which can cause severe cardiotoxic effects. Consequently, identifying and excluding potential hERG channel blockers at the earliest stages of drug development is crucial. Most traditional machine learning models predict a molecule's cardiotoxicity or non-cardiotoxicity typically at 10 μM, which doesn't account for compounds with low IC₅₀ values that are non-toxic at therapeutic levels due to their high effectiveness at lower concentrations. To address the need for more precise, quantitative predictions, we developed hERGBoost, a cutting-edge machine learning model employing a gradient-boosting algorithm. This model demonstrates superior accuracy in predicting the IC₅₀ of drug candidates. Trained on a specially curated dataset for this study, hERGBoost not only exhibited excellent performance in external validation, achieving an <em>R</em>² score of 0.394 and a low root mean square error of 0.616 but also significantly outstripped previous models in both qualitative and quantitative assessments. Representing a notable leap forward in the prediction of hERG channel blockers, the hERGBoost model and its datasets are freely available to the drug discovery community on our web server at. <span><span>http://ssbio.cau.ac.kr/software/hergboost</span><svg><path></path></svg></span> This resource promises to be invaluable in advancing safer pharmaceutical development.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109416"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647093","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":"Automatic volumetric temperature regulation during in vivo MRI-guided laser-induced thermotherapy (MRg-LITT) with multiple laser probes","authors":"Manon Desclides ,&nbsp;Valéry Ozenne ,&nbsp;Pierre Bour ,&nbsp;Thibaut Faller ,&nbsp;Guillaume Machinet ,&nbsp;Christophe Pierre ,&nbsp;Julie Carcreff ,&nbsp;Stéphane Chemouny ,&nbsp;Bruno Quesson","doi":"10.1016/j.compbiomed.2024.109445","DOIUrl":"10.1016/j.compbiomed.2024.109445","url":null,"abstract":"<div><h3>Background</h3><div>Clinical Laser-Induced Thermotherapy (LITT) currently lacks precise control of tissue temperature increase during the procedure. This study presents a new method to automatically regulate the maximum temperature increase in vivo at different positions by adjusting LITT power delivered by multiple laser probes using real-time volumetric MR-thermometry.</div></div><div><h3>Methods</h3><div>The regulation algorithm was evaluated in vivo on a pig leg muscle. Temperature regulation was performed in volumes surrounding each laser probe tip. The power delivered to each laser probe was automatically adjusted every second using a feedback control algorithm by processing on-the-fly MR-thermometry images (10 slices/second) on a 1.5 T clinical scanner (1.56 mm × 1.56 mm x 3 mm resolution), using the proton-resonance frequency (PRF) shift technique. Several experimental conditions were tested with predefined temperature-time profiles corresponding to conditions of thermal ablation (+30 °C above body temperature) or moderate hyperthermia (+10 and + 15 °C). Control images were acquired after injection of Gadolinium at the end of experiment and were compared with the thermal dose images calculated from the thermometry images.</div></div><div><h3>Results</h3><div>The mean difference and root mean squared error between target temperatures and measured ones remained below 0.5 °C and 2 °C respectively, for 5 min duration. Lesion sizes observed on thermal dose and on images acquired after gadolinium injection were in good agreement.</div></div><div><h3>Conclusion</h3><div>Automatic regulation of in vivo temperature increase during LITT procedures with multiple laser emitters control is feasible. The method provides an adaptative solution to improve the safety and efficacity of such clinical procedures.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109445"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647136","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":"Hemodynamic microenvironment of coronary stent strut malapposition","authors":"Wei Wu ,&nbsp;Sartaj Tanweer ,&nbsp;Ruben K.A. Tapia-Orihuela ,&nbsp;Parth Munjal ,&nbsp;Yash Vardhan Trivedi ,&nbsp;Shijia Zhao ,&nbsp;Hammad Zafar ,&nbsp;Haritha Darapaneni ,&nbsp;Vineeth S. Dasari ,&nbsp;Changkye Lee ,&nbsp;Rakshita Ramesh Bhat ,&nbsp;Ghassan S. Kassab ,&nbsp;Yiannis S. Chatzizisis","doi":"10.1016/j.compbiomed.2024.109378","DOIUrl":"10.1016/j.compbiomed.2024.109378","url":null,"abstract":"<div><h3>Objective</h3><div>This study aims to investigate the micro-hemodynamic effects of strut malapposition in patient-specific stented coronary bifurcations.</div></div><div><h3>Methods</h3><div>Using the mapping-back technique, three-dimensional reconstructions of clinical post-stenting artery bifurcations with strut malapposition were accurately generated from optical coherence tomography scans of 9 patients. Computational fluid dynamics (CFD) simulations were then conducted with these models to examine the impact of strut malapposition on various fluid dynamic parameters, including flow patterns, vorticity, strain rates, viscosity, and wall shear stress (WSS). For statistical analysis, virtually apposed models were created to evaluate WSS metrics. Additionally, follow-up data for 5 out of the 9 patients were reviewed to assess evidence of late thrombosis and restenosis.</div></div><div><h3>Results</h3><div>Malapposed struts induce significant alterations in flow dynamics, including the formation of recirculation regions and the transition from laminar to disturbed flow. The local curvature of the lumen also affects the development of these recirculation regions. Our study demonstrates, for the first time, that the vorticity on the abluminal side of malapposed struts exhibits an opposite sign compared to the surrounding region. The strain rate around these struts shows a distinct transition, with high values at the stent surface that rapidly diminish within the strut-lumen gap. This transition is accompanied by an increase in viscosity within these regions. Furthermore, as the malapposition distance increases, strain rates on the malapposed struts increase while viscosity decreases. Significant differences in WSS metrics were observed between clinically malapposed and virtually apposed scenarios. In clinical follow-up cases, no evidence of thrombosis was found despite the complex micro-hemodynamics in these patients.</div></div><div><h3>Conclusion</h3><div>There is complex interplay between stent malapposition and hemodynamics within a patient-specific bifurcation. The significant impact of local lumen curvature on flow dynamics underscores the limitations of idealized artery models. Moreover, the absence of thrombosis in subsequent clinical follow-up cases suggests that additional factors, such as antiplatelet medication, may play a significant role in mitigating these risks.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109378"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142643404","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Data augmentation with generative models improves detection of Non-B DNA structures","authors":"Oleksandr Cherednichenko,&nbsp;Maria Poptsova","doi":"10.1016/j.compbiomed.2024.109440","DOIUrl":"10.1016/j.compbiomed.2024.109440","url":null,"abstract":"<div><div>Non-B DNA structures, or flipons, are important functional elements that regulate a large spectrum of cellular programs. Experimental technologies for flipon detection are limited to the subsets that are active at the time of an experiment and cannot capture whole-genome functional set. Thus, the task of generating reliable whole-genome annotations of non-B DNA structures is put on deep learning models, however their quality depends on the available experimental data for training. The data augmentation approach as the combination of synthetic and real data is widely used in various fields. Deep generative models demonstrated promising results in data augmentation improving classifiers’ performance. Here we aimed at testing performance of diffusion models in comparison to other generative models in generating synthetic non-B DNA structures for data augmentation approach. We tested denoising diffusion probabilistic and implicit models (DDPM and DDIM), Wasserstein generative adversarial network (WGAN), vector quantised variational autoencoder (VQ-VAE) and showed that data augmentation improves the quality of classifiers. Diffusion models overall show the best results, but when considering three criteria of generative trilemma - quality of generated samples, diversity and sampling speed, we conclude that trade-off is possible between generative diffusion model and other architectures such as WGAN and VQ-VAE.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109440"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647139","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":"Interactive mining of neural pathways to preoperative neurosurgical planning","authors":"Shin-Ting Wu ,&nbsp;Raphael Voltoline ,&nbsp;Rodrigo Lacerda Benites ,&nbsp;Brunno Machado de Campos ,&nbsp;João Paulo Sant’Ana Santos de Souza ,&nbsp;Enrico Ghizoni","doi":"10.1016/j.compbiomed.2024.109334","DOIUrl":"10.1016/j.compbiomed.2024.109334","url":null,"abstract":"<div><h3>Background and objective:</h3><div>Preoperative understanding of white matter anatomy, including its spatial relationship with pathology and superficial landmarks, is vital for effective surgical planning. The ability to interactively synthesize neural pathways from diffusion data and dynamically discern neuroanatomy-referenced fiber patterns enables neurosurgeons to construct detailed mental models of the patient’s brain and assess surgical risks. We present a novel interactive software designed for real-time mining of neural pathways from diffusion-weighted magnetic resonance imaging (DW-MRI) data. This software leverages a user-guided approach, integrating curvilinear reformatting and surgeon expertise with diffusion tensor imaging (DTI) data, and employs a finite-state machine interaction model to facilitate intuitive use through a windows, icons, menus, and pointers (WIMP) interface.</div></div><div><h3>Methods:</h3><div>The proposed system merges user analytical skills with neuroanatomy-referenced DTI data, including scalar maps, tensor glyphs, and streamlines, within a visually interactive environment. Key features of the system include optimized GPU-based rendering for enhanced graphical representation and the proposed finite-state machine model that enables seamless interaction through intuitive controls. This approach allows for real-time manipulation of DTI data and dynamic generation of depth maps for each frame, facilitating practical exploration and analysis.</div></div><div><h3>Results:</h3><div>After testing seven control volumes, our system demonstrates tract reconstruction capabilities comparable to MRTrix software’s. The evaluation of GPU-based fiber tracking and rendering performance, using NVIDIA Nsight Visual Studio Edition, confirms the system’s interactive responsiveness. Preliminary results indicate that the environment effectively extracts critical fibers and evaluates their spatial relationships with surgical targets and landmarks. This functionality provides valuable insights for refining preoperative planning, optimizing surgical approaches, and minimizing potential functional damage.</div></div><div><h3>Conclusion:</h3><div>Our WIMP-based interactive environment empowers surgeons with enhanced capabilities for real-time manipulation of neuroanatomy-referenced DTI data. Integrating curvilinear reformatting and finite-state machine interaction enhances user experience significantly, making it a valuable tool for improving surgical safety and precision. This low-cost, accessible approach has the potential to facilitate minimally invasive procedures, accurate landmark identification, and reduced functional damage, particularly in resource-limited settings.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109334"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142643356","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":"Leveraging AI technology for distinguishing Eucommiae Cortex processing levels and evaluating anti-fatigue potential","authors":"Yijing Pan ,&nbsp;Shunshun Wang ,&nbsp;Kehong Ming ,&nbsp;Xinyue Liu ,&nbsp;Huiming Yu ,&nbsp;Qianqian Du ,&nbsp;Chenxi Deng ,&nbsp;Qingjia Chi ,&nbsp;Xianqiong Liu ,&nbsp;Chunli Wang ,&nbsp;Kang Xu","doi":"10.1016/j.compbiomed.2024.109408","DOIUrl":"10.1016/j.compbiomed.2024.109408","url":null,"abstract":"<div><div>Eucommiae Cortex (ECO) is a traditional medicinal and edible plant endemic to China, highly prized for its numerous health benefits. It typically undergoes special processing before application. The efficacy of ECO is influenced by processing techniques, necessitating the assurance of stability and consistency in its effects. However, existing methods for identifying ECO are cumbersome, thus, there is an urgent need to develop an accurate, rapid, and non-invasive assessment method. Deep learning techniques employing ResNet and Vision Transformer (ViT) models were employed to classify ECO images at various processing levels. Concurrently, the anti-fatigue properties of ECO were assessed through swimming time, pole climbing experiments, and biochemical analyses including SDH, LDH, ATP content, Na⁺-K⁺-ATPase, and Ca²⁺-Mg²⁺-ATPase indices. We demonstrated the efficacy of using image analysis to automatically classify ECO with a high degree of accuracy. The results indicated that the Vision Transformer model performed exceptionally well, achieving an accuracy rate exceeding 95 % in grading ECO images. Additionally, our study revealed that mice treated with moderately processed ECO displayed enhanced fatigue mitigation compared to other processing levels, as evidenced by multiple assessments.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109408"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647094","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":"Enhanced cross-dataset electroencephalogram-based emotion recognition using unsupervised domain adaptation","authors":"Md Niaz Imtiaz,&nbsp;Naimul Khan","doi":"10.1016/j.compbiomed.2024.109394","DOIUrl":"10.1016/j.compbiomed.2024.109394","url":null,"abstract":"<div><div>Emotion recognition holds great promise in healthcare and in the development of affect-sensitive systems such as brain–computer interfaces (BCIs). However, the high cost of labeled data and significant differences in electroencephalogram (EEG) signals among individuals limit the cross-domain application of EEG-based emotion recognition models. Addressing cross-dataset scenarios poses greater challenges due to changes in subject demographics, recording devices, and stimuli presented. To tackle these challenges, we propose an improved method for classifying EEG-based emotions across domains with different distributions. We propose a Gradual Proximity-guided Target Data Selection (GPTDS) technique, which gradually selects reliable target domain samples for training based on their proximity to the source clusters and the model’s confidence in predicting them. This approach avoids negative transfer caused by diverse and unreliable samples. Additionally, we introduce a cost-effective test-time augmentation (TTA) technique named Prediction Confidence-aware Test-Time Augmentation (PC-TTA). Traditional TTA methods often face substantial computational burden, limiting their practical utility. By applying TTA only when necessary, based on the model’s predictive confidence, our approach improves the model’s performance during inference while minimizing computational costs compared to traditional TTA approaches. Experiments on the DEAP and SEED datasets demonstrate that our method outperforms state-of-the-art approaches, achieving accuracies of 67.44% when trained on DEAP and tested on SEED, and 59.68% vice versa, with improvements of 7.09% and 6.07% over the baseline. It excels in detecting both positive and negative emotions, highlighting its effectiveness for practical emotion recognition in healthcare applications. Moreover, our proposed PC-TTA technique reduces computational time by a factor of 15 compared to traditional full TTA approaches.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109394"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142643394","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable machine learning versus known nomogram for predicting non-sentinel lymph node metastases in breast cancer patients: A comparative study","authors":"Asieh Sadat Fattahi ,&nbsp;Maryam Hoseini ,&nbsp;Toktam Dehghani ,&nbsp;Raheleh Ghouchan Nezhad Noor Nia ,&nbsp;Zeinab Naseri ,&nbsp;Amirali Ebrahimzadeh ,&nbsp;Ali Mahri ,&nbsp;Saeid Eslami","doi":"10.1016/j.compbiomed.2024.109412","DOIUrl":"10.1016/j.compbiomed.2024.109412","url":null,"abstract":"<div><h3>Introduction</h3><div>Axillary lymph node dissection (ALND) is the standard of care for breast cancer patients with positive sentinel lymph nodes (SLN), which are the first lymph nodes that drain the breast. However, many patients with positive SLNs may not have additional positive nodes, making the prediction of non-sentinel lymph node (NSLN) metastasis challenging. Reliable prognostic tools are essential for accurately assessing NSLN metastasis. The Memorial Sloan Kettering Cancer Center (MSKCC) nomogram has demonstrated effectiveness in this context, but it requires further evaluation within the Iranian breast cancer population. While ALND remains the gold standard, its unnecessary application in patients without evidence of additional positive nodes raises concerns due to potential complications such as lymphedema, nerve injury, and shoulder joint dysfunction. Furthermore, integrating Artificial Intelligence (AI) and Machine Learning (ML) techniques presents an opportunity to enhance the precision of NSLN metastasis predictions.</div></div><div><h3>Method</h3><div>This study conducts an extensive comparative analysis between the MSKCC nomogram and various ML models to predict NSLN metastasis, utilizing a dataset of Iranian breast cancer patients. Employing eXplainable Artificial Intelligence (XAI) methodologies, we analyzed 16 clinical features across a cohort of 183 patients. Our methodology includes rigorous statistical evaluations and the training and validation of ML models to assess the precision and robustness of these models compared to the MSKCC nomogram.</div></div><div><h3>Results</h3><div>Our analysis revealed that the Random Forest (RF) model outperformed the MSKCC nomogram, achieving an accuracy of 72.2 % and an AUC of 0.77, compared to the nomogram's AUC of 0.73. Logistic Regression (LR) also demonstrated competitive performance with an accuracy of 65 % and an AUC of 0.73. The RF model exhibited high sensitivity (75 %) and precision (73 %), effectively identifying critical predictors of NSLN metastasis, including the presence of ductal carcinoma in situ (DCIS) and tumor characteristics such as type and grade. Explainable AI techniques, particularly SHAP values, provided insights into feature importance, enhancing model interpretability.</div></div><div><h3>Conclusion</h3><div>Our study offers a comprehensive comparison between ML models and the MSKCC nomogram for predicting NSLN metastasis among Iranian breast cancer patients. These findings contribute valuable insights to the discourse on personalized treatment approaches, emphasizing the need for tailored prognostic tools across diverse populations. The implications of this research extend to clinical decision-making, potentially improving the accuracy and efficacy of breast cancer management within the Iranian healthcare landscape.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109412"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647141","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":"Identification of significant hub genes and pathways associated with metastatic breast cancer and tolerogenic dendritic cell via bioinformatics analysis","authors":"Kirstie Wong Chee Ching ,&nbsp;Noor Fatmawati Mokhtar ,&nbsp;Gee Jun Tye","doi":"10.1016/j.compbiomed.2024.109396","DOIUrl":"10.1016/j.compbiomed.2024.109396","url":null,"abstract":"<div><div>Metastatic breast cancer (MBC) is an advanced-stage breast cancer associated with more than 90 % of cancer-related deaths. Immunosuppressive properties of tolerogenic dendritic cells (tolDCs) in tumour immune microenvironment (TIME) may be a risk factor for the rapid progression to MBC. However, the exact connections between the two are unknown. The aim of the current study is to uncover gene signatures and key pathways associated with MBC and tolDCs via an integrated bioinformatics approach. Gene expression profiles of MBC and tolDCs were retrieved from Gene Expression Omnibus (GEO) to identify common differentially expressed genes (DEGs). From DGE analysis, 529 upregulated common DEGs and 367 downregulated common DEGs had been identified. In enrichment analysis, common DEGs enriched in GO terms of defense response to virus and KEGG pathway of transcriptional misregulation in cancer were reported to be significantly associated with MBC and tolDCs. From the constructed PPI networks, 23 hub genes were identified, although only 5 genes were significant; 3 upregulated (<em>ISG15</em>, <em>OAS2</em> and <em>RSAD2</em>) and 2 downregulated (<em>eEF2</em> and <em>PPARG</em>) as they were found to be significantly correlated and had the same expression trend as predicted in validation analysis of overall survival (OS) analysis, expression levels, immune infiltration analysis and immunohistochemistry (IHC) analysis. These 5 hub genes can now be exploited in developing novel therapeutic interventions and as diagnostic biomarkers for enhancing the clinical outcomes of MBC patients.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109396"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142643410","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":"Augmenting a spine CT scans dataset using VAEs, GANs, and transfer learning for improved detection of vertebral compression fractures","authors":"Zeina El Kojok,&nbsp;Hadi Al Khansa,&nbsp;Fouad Trad,&nbsp;Ali Chehab","doi":"10.1016/j.compbiomed.2024.109446","DOIUrl":"10.1016/j.compbiomed.2024.109446","url":null,"abstract":"<div><div>In recent years, deep learning has become a popular tool to analyze and classify medical images. However, challenges such as limited data availability, high labeling costs, and privacy concerns remain significant obstacles. As such, generative models have been extensively explored as a solution to generate new images and overcome the stated challenges. In this paper, we augment a dataset of chest CT scans for Vertebral Compression Fractures (VCFs) collected from the American University of Beirut Medical Center (AUBMC), specifically targeting the detection of incidental fractures that are often overlooked in routine chest CTs, as these scans are not typically focused on spinal analysis. Our goal is to enhance AI systems to enable automated early detection of such incidental fractures, addressing a critical healthcare gap and leading to improved patient outcomes by catching fractures that might otherwise go undiagnosed. We first generate a synthetic dataset based on the segmented CTSpine1K dataset to simulate real grayscale data that aligns with our specific scenario. Then, we use this generated data to evaluate the generative capabilities of Deep Convolutional Generative Adverserial Networks (DCGANs), variational autoencoders (VAEs), and VAE-GAN models. The VAE-GAN model demonstrated the highest performance, achieving a Fréchet Inception Distance (FID) five times lower than the other architectures. To adapt this model to real-image scenarios, we perform transfer learning on the GAN, training it with the real dataset collected from AUBMC and generating additional samples. Finally, we train a CNN using augmented datasets that include both real and generated synthetic data and compare its performance to training on real data alone. We then evaluate the model exclusively on a test set composed of real images to assess the effect of the generated data on real-world performance. We find that training on augmented datasets significantly improves the classification accuracy on a test set composed of real images by 16 %, increasing it from 73 % to 89 %. This improvement demonstrates that the generated data is of high quality and enhances the model's ability to perform well against unseen, real data.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"184 ","pages":"Article 109446"},"PeriodicalIF":7.0,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142647130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}