基于深度学习的肿瘤患者脑MR图像自动多类组织分割。

IF 1 4区 医学 Q4 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING
Ankit Kandpal, Puneet Kumar, Rakesh Kumar Gupta, Anup Singh
{"title":"基于深度学习的肿瘤患者脑MR图像自动多类组织分割。","authors":"Ankit Kandpal, Puneet Kumar, Rakesh Kumar Gupta, Anup Singh","doi":"10.1097/RCT.0000000000001750","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>Precise delineation of brain tissues, including lesions, in MR images is crucial for data analysis and objectively assessing conditions like neurological disorders and brain tumors. Existing methods for tissue segmentation often fall short in addressing patients with lesions, particularly those with brain tumors. This study aimed to develop and evaluate a robust pipeline utilizing convolutional neural networks for rapid and automatic segmentation of whole brain tissues, including tumor lesions.</p><p><strong>Materials and methods: </strong>The proposed pipeline was developed using BraTS'21 data (1251 patients) and tested on local hospital data (100 patients). Ground truth masks for lesions as well as brain tissues were generated. Two convolutional neural networks based on deep residual U-Net framework were trained for segmenting brain tissues and tumor lesions. The performance of the pipeline was evaluated on independent test data using dice similarity coefficient (DSC) and volume similarity (VS).</p><p><strong>Results: </strong>The proposed pipeline achieved a mean DSC of 0.84 and a mean VS of 0.93 on the BraTS'21 test data set. On the local hospital test data set, it attained a mean DSC of 0.78 and a mean VS of 0.91. The proposed pipeline also generated satisfactory masks in cases where the SPM12 software performed inadequately.</p><p><strong>Conclusions: </strong>The proposed pipeline offers a reliable and automatic solution for segmenting brain tissues and tumor lesions in MR images. Its adaptability makes it a valuable tool for both research and clinical applications, potentially streamlining workflows and enhancing the precision of analyses in neurological and oncological studies.</p>","PeriodicalId":15402,"journal":{"name":"Journal of Computer Assisted Tomography","volume":" ","pages":""},"PeriodicalIF":1.0000,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Automatic Multiclass Tissue Segmentation Using Deep Learning in Brain MR Images of Tumor Patients.\",\"authors\":\"Ankit Kandpal, Puneet Kumar, Rakesh Kumar Gupta, Anup Singh\",\"doi\":\"10.1097/RCT.0000000000001750\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Objective: </strong>Precise delineation of brain tissues, including lesions, in MR images is crucial for data analysis and objectively assessing conditions like neurological disorders and brain tumors. Existing methods for tissue segmentation often fall short in addressing patients with lesions, particularly those with brain tumors. This study aimed to develop and evaluate a robust pipeline utilizing convolutional neural networks for rapid and automatic segmentation of whole brain tissues, including tumor lesions.</p><p><strong>Materials and methods: </strong>The proposed pipeline was developed using BraTS'21 data (1251 patients) and tested on local hospital data (100 patients). Ground truth masks for lesions as well as brain tissues were generated. Two convolutional neural networks based on deep residual U-Net framework were trained for segmenting brain tissues and tumor lesions. The performance of the pipeline was evaluated on independent test data using dice similarity coefficient (DSC) and volume similarity (VS).</p><p><strong>Results: </strong>The proposed pipeline achieved a mean DSC of 0.84 and a mean VS of 0.93 on the BraTS'21 test data set. On the local hospital test data set, it attained a mean DSC of 0.78 and a mean VS of 0.91. The proposed pipeline also generated satisfactory masks in cases where the SPM12 software performed inadequately.</p><p><strong>Conclusions: </strong>The proposed pipeline offers a reliable and automatic solution for segmenting brain tissues and tumor lesions in MR images. Its adaptability makes it a valuable tool for both research and clinical applications, potentially streamlining workflows and enhancing the precision of analyses in neurological and oncological studies.</p>\",\"PeriodicalId\":15402,\"journal\":{\"name\":\"Journal of Computer Assisted Tomography\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":1.0000,\"publicationDate\":\"2025-06-30\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Computer Assisted Tomography\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1097/RCT.0000000000001750\",\"RegionNum\":4,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Computer Assisted Tomography","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1097/RCT.0000000000001750","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0

摘要

目的:在MR图像中精确描绘脑组织,包括病变,对于数据分析和客观评估神经系统疾病和脑肿瘤等疾病至关重要。现有的组织分割方法在处理病变患者,特别是脑肿瘤患者时往往存在不足。本研究旨在开发和评估利用卷积神经网络快速自动分割全脑组织(包括肿瘤病变)的强大管道。材料和方法:拟议的管道使用BraTS的21个数据(1251名患者)开发,并在当地医院数据(100名患者)上进行测试。生成了病灶和脑组织的真相面具。基于深度残差U-Net框架训练了两个卷积神经网络用于脑组织和肿瘤病灶的分割。采用骰子相似系数(DSC)和体积相似系数(VS)对独立测试数据进行管道性能评价。结果:在BraTS'21测试数据集上,拟议管道的平均DSC为0.84,平均VS为0.93。在当地医院测试数据集上,平均DSC为0.78,平均VS为0.91。在SPM12软件性能不佳的情况下,拟议的管道也产生了令人满意的掩码。结论:该管道为MR图像中脑组织和肿瘤病变的分割提供了可靠、自动化的解决方案。它的适应性使其成为研究和临床应用的宝贵工具,有可能简化工作流程并提高神经学和肿瘤学研究的分析精度。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Automatic Multiclass Tissue Segmentation Using Deep Learning in Brain MR Images of Tumor Patients.

Objective: Precise delineation of brain tissues, including lesions, in MR images is crucial for data analysis and objectively assessing conditions like neurological disorders and brain tumors. Existing methods for tissue segmentation often fall short in addressing patients with lesions, particularly those with brain tumors. This study aimed to develop and evaluate a robust pipeline utilizing convolutional neural networks for rapid and automatic segmentation of whole brain tissues, including tumor lesions.

Materials and methods: The proposed pipeline was developed using BraTS'21 data (1251 patients) and tested on local hospital data (100 patients). Ground truth masks for lesions as well as brain tissues were generated. Two convolutional neural networks based on deep residual U-Net framework were trained for segmenting brain tissues and tumor lesions. The performance of the pipeline was evaluated on independent test data using dice similarity coefficient (DSC) and volume similarity (VS).

Results: The proposed pipeline achieved a mean DSC of 0.84 and a mean VS of 0.93 on the BraTS'21 test data set. On the local hospital test data set, it attained a mean DSC of 0.78 and a mean VS of 0.91. The proposed pipeline also generated satisfactory masks in cases where the SPM12 software performed inadequately.

Conclusions: The proposed pipeline offers a reliable and automatic solution for segmenting brain tissues and tumor lesions in MR images. Its adaptability makes it a valuable tool for both research and clinical applications, potentially streamlining workflows and enhancing the precision of analyses in neurological and oncological studies.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
CiteScore
2.50
自引率
0.00%
发文量
230
审稿时长
4-8 weeks
期刊介绍: The mission of Journal of Computer Assisted Tomography is to showcase the latest clinical and research developments in CT, MR, and closely related diagnostic techniques. We encourage submission of both original research and review articles that have immediate or promissory clinical applications. Topics of special interest include: 1) functional MR and CT of the brain and body; 2) advanced/innovative MRI techniques (diffusion, perfusion, rapid scanning); and 3) advanced/innovative CT techniques (perfusion, multi-energy, dose-reduction, and processing).
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信