Hao Zhang, Genji Yuan, Ziyue Zhang, Xiang Guo, Ruixiang Xu, Tongshuai Xu, Xin Zhong, Meng Kong, Kai Zhu, Xuexiao Ma
{"title":"A multi-scene deep learning model for automated segmentation of acute vertebral compression fractures from radiographs: a multicenter cohort study.","authors":"Hao Zhang, Genji Yuan, Ziyue Zhang, Xiang Guo, Ruixiang Xu, Tongshuai Xu, Xin Zhong, Meng Kong, Kai Zhu, Xuexiao Ma","doi":"10.1186/s13244-024-01861-y","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>To develop a multi-scene model that can automatically segment acute vertebral compression fractures (VCFs) from spine radiographs.</p><p><strong>Methods: </strong>In this multicenter study, we collected radiographs from five hospitals (Hospitals A-E) between November 2016 and October 2019. The study included participants with acute VCFs, as well as healthy controls. For the development of the Positioning and Focus Network (PFNet), we used a training dataset consisting of 1071 participants from Hospitals A and B. The validation dataset included 458 participants from Hospitals A and B, whereas external test datasets 1-3 included 301 participants from Hospital C, 223 from Hospital D, and 261 from Hospital E, respectively. We evaluated the segmentation performance of the PFNet model and compared it with previously described approaches. Additionally, we used qualitative comparison and gradient-weighted class activation mapping (Grad-CAM) to explain the feature learning and segmentation results of the PFNet model.</p><p><strong>Results: </strong>The PFNet model achieved accuracies of 99.93%, 98.53%, 99.21%, and 100% for the segmentation of acute VCFs in the validation dataset and external test datasets 1-3, respectively. The receiver operating characteristic curves comparing the four models across the validation and external test datasets consistently showed that the PFNet model outperformed other approaches, achieving the highest values for all measures. The qualitative comparison and Grad-CAM provided an intuitive view of the interpretability and effectiveness of our PFNet model.</p><p><strong>Conclusion: </strong>In this study, we successfully developed a multi-scene model based on spine radiographs for precise preoperative and intraoperative segmentation of acute VCFs.</p><p><strong>Critical relevance statement: </strong>Our PFNet model demonstrated high accuracy in multi-scene segmentation in clinical settings, making it a significant advancement in this field.</p><p><strong>Key points: </strong>This study developed the first multi-scene deep learning model capable of segmenting acute VCFs from spine radiographs. The model's architecture consists of two crucial modules: an attention-guided module and a supervised decoding module. The exceptional generalization and consistently superior performance of our model were validated using multicenter external test datasets.</p>","PeriodicalId":13639,"journal":{"name":"Insights into Imaging","volume":"15 1","pages":"290"},"PeriodicalIF":4.1000,"publicationDate":"2024-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11612108/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Insights into Imaging","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1186/s13244-024-01861-y","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0
Abstract
Objective: To develop a multi-scene model that can automatically segment acute vertebral compression fractures (VCFs) from spine radiographs.
Methods: In this multicenter study, we collected radiographs from five hospitals (Hospitals A-E) between November 2016 and October 2019. The study included participants with acute VCFs, as well as healthy controls. For the development of the Positioning and Focus Network (PFNet), we used a training dataset consisting of 1071 participants from Hospitals A and B. The validation dataset included 458 participants from Hospitals A and B, whereas external test datasets 1-3 included 301 participants from Hospital C, 223 from Hospital D, and 261 from Hospital E, respectively. We evaluated the segmentation performance of the PFNet model and compared it with previously described approaches. Additionally, we used qualitative comparison and gradient-weighted class activation mapping (Grad-CAM) to explain the feature learning and segmentation results of the PFNet model.
Results: The PFNet model achieved accuracies of 99.93%, 98.53%, 99.21%, and 100% for the segmentation of acute VCFs in the validation dataset and external test datasets 1-3, respectively. The receiver operating characteristic curves comparing the four models across the validation and external test datasets consistently showed that the PFNet model outperformed other approaches, achieving the highest values for all measures. The qualitative comparison and Grad-CAM provided an intuitive view of the interpretability and effectiveness of our PFNet model.
Conclusion: In this study, we successfully developed a multi-scene model based on spine radiographs for precise preoperative and intraoperative segmentation of acute VCFs.
Critical relevance statement: Our PFNet model demonstrated high accuracy in multi-scene segmentation in clinical settings, making it a significant advancement in this field.
Key points: This study developed the first multi-scene deep learning model capable of segmenting acute VCFs from spine radiographs. The model's architecture consists of two crucial modules: an attention-guided module and a supervised decoding module. The exceptional generalization and consistently superior performance of our model were validated using multicenter external test datasets.
期刊介绍:
Insights into Imaging (I³) is a peer-reviewed open access journal published under the brand SpringerOpen. All content published in the journal is freely available online to anyone, anywhere!
I³ continuously updates scientific knowledge and progress in best-practice standards in radiology through the publication of original articles and state-of-the-art reviews and opinions, along with recommendations and statements from the leading radiological societies in Europe.
Founded by the European Society of Radiology (ESR), I³ creates a platform for educational material, guidelines and recommendations, and a forum for topics of controversy.
A balanced combination of review articles, original papers, short communications from European radiological congresses and information on society matters makes I³ an indispensable source for current information in this field.
I³ is owned by the ESR, however authors retain copyright to their article according to the Creative Commons Attribution License (see Copyright and License Agreement). All articles can be read, redistributed and reused for free, as long as the author of the original work is cited properly.
The open access fees (article-processing charges) for this journal are kindly sponsored by ESR for all Members.
The journal went open access in 2012, which means that all articles published since then are freely available online.