A few-shot learning framework for the diagnosis of osteopenia and osteoporosis using knee X-ray images.

IF 1.4 4区 医学 Q4 MEDICINE, RESEARCH & EXPERIMENTAL
Hua Xie, Chenqi Gu, Wenchao Zhang, Jiacheng Zhu, Jin He, Zhou Huang, Jinzhou Zhu, Zhonghua Xu
{"title":"A few-shot learning framework for the diagnosis of osteopenia and osteoporosis using knee X-ray images.","authors":"Hua Xie, Chenqi Gu, Wenchao Zhang, Jiacheng Zhu, Jin He, Zhou Huang, Jinzhou Zhu, Zhonghua Xu","doi":"10.1177/03000605241274576","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>We developed a few-shot learning (FSL) framework for the diagnosis of osteopenia and osteoporosis in knee X-ray images.</p><p><strong>Methods: </strong>Computer vision models containing deep convolutional neural networks were fine-tuned to enable generalization from natural images (ImageNet) to chest X-ray images (normal vs. pneumonia, base images). Then, a series of automated machine learning classifiers based on the Euclidean distances of base images were developed to make predictions for novel images (normal vs. osteopenia vs. osteoporosis). The performance of the FSL framework was compared with that of junior and senior radiologists. In addition, the gradient-weighted class activation mapping algorithm was used for visual interpretation.</p><p><strong>Results: </strong>In Cohort #1, the mean accuracy (0.728) and sensitivity (0.774) of the FSL models were higher than those of the radiologists (0.512 and 0.448). A diagnostic pipeline of FSL model (first)-radiologists (second) achieved better performance (0.653 accuracy, 0.582 sensitivity, and 0.816 specificity) than radiologists alone. In Cohort #2, the diagnostic pipeline also showed improved performance.</p><p><strong>Conclusions: </strong>The FSL framework yielded practical performance with respect to the diagnosis of osteopenia and osteoporosis in comparison with radiologists. This retrospective study supports the use of promising FSL methods in computer-aided diagnosis tasks involving limited samples.</p>","PeriodicalId":16129,"journal":{"name":"Journal of International Medical Research","volume":null,"pages":null},"PeriodicalIF":1.4000,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11375658/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of International Medical Research","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1177/03000605241274576","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"MEDICINE, RESEARCH & EXPERIMENTAL","Score":null,"Total":0}
引用次数: 0

Abstract

Objective: We developed a few-shot learning (FSL) framework for the diagnosis of osteopenia and osteoporosis in knee X-ray images.

Methods: Computer vision models containing deep convolutional neural networks were fine-tuned to enable generalization from natural images (ImageNet) to chest X-ray images (normal vs. pneumonia, base images). Then, a series of automated machine learning classifiers based on the Euclidean distances of base images were developed to make predictions for novel images (normal vs. osteopenia vs. osteoporosis). The performance of the FSL framework was compared with that of junior and senior radiologists. In addition, the gradient-weighted class activation mapping algorithm was used for visual interpretation.

Results: In Cohort #1, the mean accuracy (0.728) and sensitivity (0.774) of the FSL models were higher than those of the radiologists (0.512 and 0.448). A diagnostic pipeline of FSL model (first)-radiologists (second) achieved better performance (0.653 accuracy, 0.582 sensitivity, and 0.816 specificity) than radiologists alone. In Cohort #2, the diagnostic pipeline also showed improved performance.

Conclusions: The FSL framework yielded practical performance with respect to the diagnosis of osteopenia and osteoporosis in comparison with radiologists. This retrospective study supports the use of promising FSL methods in computer-aided diagnosis tasks involving limited samples.

利用膝关节 X 光图像诊断骨质增生和骨质疏松症的少量学习框架。
目的我们开发了一种用于诊断膝关节 X 光图像中骨质增生和骨质疏松症的少量学习(FSL)框架:我们对包含深度卷积神经网络的计算机视觉模型进行了微调,以实现从自然图像(ImageNet)到胸部 X 光图像(正常与肺炎、基础图像)的泛化。然后,基于基础图像的欧氏距离开发了一系列自动机器学习分类器,以便对新图像(正常与骨质疏松症与骨质疏松症)进行预测。FSL 框架的性能与初级和高级放射科医生的性能进行了比较。此外,梯度加权类激活映射算法也被用于视觉判读:在队列 1 中,FSL 模型的平均准确度(0.728)和灵敏度(0.774)均高于放射科医生(0.512 和 0.448)。由 FSL 模型(第一位)和放射科医生(第二位)组成的诊断流水线比单纯由放射科医生组成的诊断流水线取得了更好的效果(准确性为 0.653,灵敏度为 0.582,特异性为 0.816)。在 2 号队列中,诊断管道的性能也有所提高:与放射科医生相比,FSL 框架在诊断骨质疏松症和骨质疏松症方面具有实用性。这项回顾性研究支持在涉及有限样本的计算机辅助诊断任务中使用前景广阔的 FSL 方法。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
3.20
自引率
0.00%
发文量
555
审稿时长
1 months
期刊介绍: _Journal of International Medical Research_ is a leading international journal for rapid publication of original medical, pre-clinical and clinical research, reviews, preliminary and pilot studies on a page charge basis. As a service to authors, every article accepted by peer review will be given a full technical edit to make papers as accessible and readable to the international medical community as rapidly as possible. Once the technical edit queries have been answered to the satisfaction of the journal, the paper will be published and made available freely to everyone under a creative commons licence. Symposium proceedings, summaries of presentations or collections of medical, pre-clinical or clinical data on a specific topic are welcome for publication as supplements. Print ISSN: 0300-0605
×
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:481959085
Book学术官方微信