Gait Patterns as Biomarkers: A Video-Based Approach for Classifying Scoliosis

TL;DR

This work addresses the need for non-invasive, scalable scoliosis screening in adolescents by introducing a gait-based video approach. It presents Scoliosis1K, the first large-scale dataset of over 1,050 adolescents with 447,900 silhouettes across 1,493 sequences labeled by Cobb angle thresholds, and develops ScoNet and ScoNet-MT to classify scoliosis from gait patterns using privacy-preserving silhouettes. ScoNet-MT, which adds a gait-recognition task via multi-task learning and triplet loss, achieves significantly higher accuracy ($82.0\%$) and specificity ($76.5\%$) than the baseline ScoNet and traditional screening methods, indicating strong potential for practical deployment. The approach offers a scalable, privacy-preserving alternative to radiography, with implications for improving screening in resource-limited settings and guiding future non-invasive diagnostic methods.

Abstract

Scoliosis presents significant diagnostic challenges, particularly in adolescents, where early detection is crucial for effective treatment. Traditional diagnostic and follow-up methods, which rely on physical examinations and radiography, face limitations due to the need for clinical expertise and the risk of radiation exposure, thus restricting their use for widespread early screening. In response, we introduce a novel video-based, non-invasive method for scoliosis classification using gait analysis, effectively circumventing these limitations. This study presents Scoliosis1K, the first large-scale dataset specifically designed for video-based scoliosis classification, encompassing over one thousand adolescents. Leveraging this dataset, we developed ScoNet, an initial model that faced challenges in handling the complexities of real-world data. This led to the development of ScoNet-MT, an enhanced model incorporating multi-task learning, which demonstrates promising diagnostic accuracy for practical applications. Our findings demonstrate that gait can serve as a non-invasive biomarker for scoliosis, revolutionizing screening practices through deep learning and setting a precedent for non-invasive diagnostic methodologies. The dataset and code are publicly available at https://zhouzi180.github.io/Scoliosis1K/.

Figures (4)

• Figure 1: Comparative overview of scoliosis diagnosis methods: (a) Traditional X-ray examination, the clinical gold standard Thuaimer2024; (b) Non-invasive analysis of bareback photos zhang2023deep; (c) Our proposed gait analysis approach, enabling efficient, large-scale early adolescent screening to identify cases requiring further radiographic investigation, highlighting its non-invasive and privacy-preserving characteristics.
• Figure 2: Silhouettes from the Scoliosis1K dataset: (a) positive, (b) neutral, and (c) negative samples.
• Figure 3: The Proposed Pipeline: The participant is tracked throughout the video, excluding non-participant entities like clinicians. The participant's silhouette is then segmented, followed by scoliosis classification using ScoNet-MT based on gait analysis.
• Figure 4: Visualization results of our method.