{"title":"在脊柱侧弯评估中,中心点人工智能模型在自动估计钴角方面的性能优于 VFLDNet。","authors":"Qingqing Lu, Lixin Ni, Zhehao Zhang, Lulin Zou, Lijun Guo, Yuning Pan","doi":"10.1007/s00586-024-08538-6","DOIUrl":null,"url":null,"abstract":"<p><strong>Purpose: </strong>Aims to establish the superiority of our proposed model over the state-of-the-art vertebra-focused landmark detection network (VFLDNet) in automating Cobb angle estimation from spinal radiographs.</p><p><strong>Methods: </strong>Utilizing a private dataset for external validation, we compared the performance of our center-point detection-based vertebra localization and tilt estimation network (VLTENet) with the key-point detection-based VFLDNet. Both models' Cobb angle predictions were rigorously evaluated against manual consensus score using metrics such as mean absolute error (MAE), correlation coefficient, intraclass correlation coefficient (ICC), Fleiss' kappa, Bland-Altman analysis, and classification metrics [sensitivity (SN), specificity, accuracy] focusing on major curve estimation and scoliosis severity classification.</p><p><strong>Results: </strong>A retrospective analysis of 118 cases with 342 Cobb angle measurements revealed that our model achieved a MAE of 2.15° for total Cobb angles and 1.89° for the major curve, significantly outperforming VFLDNet's MAE of 2.80°and 2.57°, respectively. Both models demonstrated robust correlation and ICC, but our model excelled in classification consistency, particularly in predicting major curve magnitude (ours: kappa = 0.83; VFLDNet: kappa = 0.67). In subgroup analyses by scoliosis severity, our model consistently surpassed VFLDNet, displaying superior mean (SD) differences, narrower limits of agreement, and higher SN, specificity, and accuracy, most notably in moderate (ours: SN = 86.84%; VFLDNet: SN = 83.16%) to severe (ours: SN = 92.86%; VFLDNet: SN = 85.71%) scoliosis.</p><p><strong>Conclusion: </strong>Our model emerges as the superior choice for automated Cobb angle estimation, particularly in assessing major curve and moderate to severe scoliosis, underscoring its potential to revolutionize clinical workflows and enhance patient care.</p>","PeriodicalId":12323,"journal":{"name":"European Spine Journal","volume":" ","pages":"4710-4719"},"PeriodicalIF":2.6000,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Superior performance of a center-point AI model over VFLDNet in automated cobb angle estimation for scoliosis assessment.\",\"authors\":\"Qingqing Lu, Lixin Ni, Zhehao Zhang, Lulin Zou, Lijun Guo, Yuning Pan\",\"doi\":\"10.1007/s00586-024-08538-6\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Purpose: </strong>Aims to establish the superiority of our proposed model over the state-of-the-art vertebra-focused landmark detection network (VFLDNet) in automating Cobb angle estimation from spinal radiographs.</p><p><strong>Methods: </strong>Utilizing a private dataset for external validation, we compared the performance of our center-point detection-based vertebra localization and tilt estimation network (VLTENet) with the key-point detection-based VFLDNet. Both models' Cobb angle predictions were rigorously evaluated against manual consensus score using metrics such as mean absolute error (MAE), correlation coefficient, intraclass correlation coefficient (ICC), Fleiss' kappa, Bland-Altman analysis, and classification metrics [sensitivity (SN), specificity, accuracy] focusing on major curve estimation and scoliosis severity classification.</p><p><strong>Results: </strong>A retrospective analysis of 118 cases with 342 Cobb angle measurements revealed that our model achieved a MAE of 2.15° for total Cobb angles and 1.89° for the major curve, significantly outperforming VFLDNet's MAE of 2.80°and 2.57°, respectively. Both models demonstrated robust correlation and ICC, but our model excelled in classification consistency, particularly in predicting major curve magnitude (ours: kappa = 0.83; VFLDNet: kappa = 0.67). In subgroup analyses by scoliosis severity, our model consistently surpassed VFLDNet, displaying superior mean (SD) differences, narrower limits of agreement, and higher SN, specificity, and accuracy, most notably in moderate (ours: SN = 86.84%; VFLDNet: SN = 83.16%) to severe (ours: SN = 92.86%; VFLDNet: SN = 85.71%) scoliosis.</p><p><strong>Conclusion: </strong>Our model emerges as the superior choice for automated Cobb angle estimation, particularly in assessing major curve and moderate to severe scoliosis, underscoring its potential to revolutionize clinical workflows and enhance patient care.</p>\",\"PeriodicalId\":12323,\"journal\":{\"name\":\"European Spine Journal\",\"volume\":\" \",\"pages\":\"4710-4719\"},\"PeriodicalIF\":2.6000,\"publicationDate\":\"2024-12-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"European Spine Journal\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1007/s00586-024-08538-6\",\"RegionNum\":3,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2024/10/29 0:00:00\",\"PubModel\":\"Epub\",\"JCR\":\"Q2\",\"JCRName\":\"CLINICAL NEUROLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"European Spine Journal","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s00586-024-08538-6","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/10/29 0:00:00","PubModel":"Epub","JCR":"Q2","JCRName":"CLINICAL NEUROLOGY","Score":null,"Total":0}
Superior performance of a center-point AI model over VFLDNet in automated cobb angle estimation for scoliosis assessment.
Purpose: Aims to establish the superiority of our proposed model over the state-of-the-art vertebra-focused landmark detection network (VFLDNet) in automating Cobb angle estimation from spinal radiographs.
Methods: Utilizing a private dataset for external validation, we compared the performance of our center-point detection-based vertebra localization and tilt estimation network (VLTENet) with the key-point detection-based VFLDNet. Both models' Cobb angle predictions were rigorously evaluated against manual consensus score using metrics such as mean absolute error (MAE), correlation coefficient, intraclass correlation coefficient (ICC), Fleiss' kappa, Bland-Altman analysis, and classification metrics [sensitivity (SN), specificity, accuracy] focusing on major curve estimation and scoliosis severity classification.
Results: A retrospective analysis of 118 cases with 342 Cobb angle measurements revealed that our model achieved a MAE of 2.15° for total Cobb angles and 1.89° for the major curve, significantly outperforming VFLDNet's MAE of 2.80°and 2.57°, respectively. Both models demonstrated robust correlation and ICC, but our model excelled in classification consistency, particularly in predicting major curve magnitude (ours: kappa = 0.83; VFLDNet: kappa = 0.67). In subgroup analyses by scoliosis severity, our model consistently surpassed VFLDNet, displaying superior mean (SD) differences, narrower limits of agreement, and higher SN, specificity, and accuracy, most notably in moderate (ours: SN = 86.84%; VFLDNet: SN = 83.16%) to severe (ours: SN = 92.86%; VFLDNet: SN = 85.71%) scoliosis.
Conclusion: Our model emerges as the superior choice for automated Cobb angle estimation, particularly in assessing major curve and moderate to severe scoliosis, underscoring its potential to revolutionize clinical workflows and enhance patient care.
期刊介绍:
"European Spine Journal" is a publication founded in response to the increasing trend toward specialization in spinal surgery and spinal pathology in general. The Journal is devoted to all spine related disciplines, including functional and surgical anatomy of the spine, biomechanics and pathophysiology, diagnostic procedures, and neurology, surgery and outcomes. The aim of "European Spine Journal" is to support the further development of highly innovative spine treatments including but not restricted to surgery and to provide an integrated and balanced view of diagnostic, research and treatment procedures as well as outcomes that will enhance effective collaboration among specialists worldwide. The “European Spine Journal” also participates in education by means of videos, interactive meetings and the endorsement of educative efforts.
Official publication of EUROSPINE, The Spine Society of Europe