Comparative analysis of a manual and an automated 3D landmark digitization method of the torso in adolescents with idiopathic scoliosis

对青少年特发性脊柱侧弯患者躯干的手动和自动三维标志点数字化方法进行比较分析

阅读:1

Abstract

Adolescent idiopathic scoliosis (AIS) often presents with significant 3D asymmetry of the torso, posing challenges for both patients and clinicians. Surface topography, based on the identification of anatomical landmarks, offers a non-invasive alternative to x-rays for monitoring shape changes over time, thereby reducing radiation exposure. However, the current gold standard, a manual landmarking process, is labor intensive and prone to error. Here, we present an automated 3D landmark digitization method designed to address these limitations. We performed the validation comparing the automated 3D landmark digitization method against the manual gQ1old standard across three phases: preliminary error assessment, geometric morphometrics (GMM) shape/size evaluation, and practical allometry application. Our results show that the automated method effectively quantifies torso shape, achieving a nonsignificant measurement error in both groups (23.1 mm in patients with p-value = 0.33; and 20.3 mm in controls with p-value = 0.30). It has also captured variance patterns comparable to the manual approach, showing high agreement for PC1 (0.94; CI95%: 0.91-0.96) and good agreement for PC2 (0.85; CI95%: 0.78-0.90), and performs similarly in assessing allometry, without significant differences in capturing the allometric signal (p-value = 0.09). However, the automated method exhibited reduced ability to capture shape variability, highlighting potential areas for improvement. These results suggest that automated, non-radiographic techniques hold promise for clinical application in tracking AIS progression. Future refinements could further improve accuracy, paving the way for safer and more efficient scoliosis management strategies.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。