期刊
CLINICAL AND TRANSLATIONAL IMAGING
卷 11, 期 3, 页码 271-283出版社
SPRINGER-VERLAG ITALIA SRL
DOI: 10.1007/s40336-023-00539-7
关键词
Deep learning; Classification; Bone scan; Systematic review; Bone metastasis
This systematic review evaluates the use of deep learning models in bone scan classification and concludes that they have strong potential for clinical diagnosis assistance.
IntroductionBone scintigraphy, a nuclear medicine technique, is widely used for the detection of bone metastasis. Deep learning has also been used in bone scan classification. Thus, we performed this systematic review to draw the most up-to-date conclusions on this topic.Materials/methodsDatabases (PubMed, Cochrane, Embase, and IEEE Xplore) were searched for articles on neural-network-based bone scan classification (determination of whether a patient has bone metastases) from inception to April 4, 2022. The study quality was evaluated using QUADAS-2.ResultsWe collected a total of 616 articles. After article review, 24 articles were selected for inclusion in the final systematic review. 14 studies adopted convolutional neural networks (CNN) methods to extract initial image features. Other ten studies used either bone scan index (BSI) or region-specific thresholding methods. Most of the included studies exhibited high quality.ConclusionDeep-learning-based models have shown strong potential for incorporation into the clinical scenario of diagnosis assistance.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据