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 <a class="link-external link-https" href="https://zhouzi180.github.io/Scoliosis1K/" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the early detection and classification of adolescent scoliosis. Specifically, traditional scoliosis diagnosis methods rely on physical examinations and X - ray imaging, which have limitations such as requiring clinical expertise, having a high radiation risk, and being difficult to conduct extensive early - screening. Therefore, this paper proposes a non - invasive video - based method to classify scoliosis through gait analysis, thereby overcoming the limitations of traditional methods. ### Problem Background - **The Impact of Scoliosis**: Scoliosis is a complex spinal deformity. Especially in adolescents, it can have a significant impact on physical health and quality of life. - **Limitations of Traditional Diagnosis Methods**: - Physical examinations and X - ray imaging require professional clinical knowledge. - X - ray imaging has a radiation risk, which limits its application in early - wide - screening. - These methods are usually only effective when symptoms are obvious, while scoliosis in the early stage is often asymptomatic. ### Proposed Solutions To address the above challenges, the author proposes a video - based gait analysis method for non - invasively classifying scoliosis. The main contributions include: 1. **Creation of the Scoliosis1K Dataset**: This is the first large - scale video dataset specifically used for video - based scoliosis classification, containing data of more than 1,000 adolescents. 2. **Development of the ScoNet Model**: This is the first baseline model used for classifying scoliosis through gait analysis. 3. **Improvement to the ScoNet - MT Model**: Introduce a multi - task learning mechanism to better handle the complexity of real - world data and improve classification accuracy. ### Method Overview - **Data Collection and Pre - processing**: Use cameras to record videos and convert the videos into binary contour sequences to protect privacy and focus on the human body area. - **Gait Feature Extraction**: Convert participant contours into 3D feature maps through the ResNet architecture, and further extract key features using temporal pooling and horizontal pooling. - **Classification Models**: ScoNet and ScoNet - MT are trained using cross - entropy loss and triplet loss respectively to improve classification accuracy. ### Experimental Results The experimental results show that ScoNet - MT is significantly superior to the initial ScoNet model in terms of accuracy and specificity, especially when dealing with class - imbalanced data. This indicates that gait can be used as a non - invasive biomarker for the early screening of scoliosis. ### Conclusion This study demonstrates the effectiveness of using gait as a non - invasive biomarker for scoliosis classification, providing new tools for early and accurate scoliosis classification. This method is scalable and privacy - protecting, especially suitable for areas with limited resources, and is expected to significantly improve the screening of adolescent scoliosis in the future.

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

Gait characteristics and clinical relevance of hereditary spinocerebellar ataxia on deep learning

Development and validation of deep learning algorithms for scoliosis screening using back images

A New Method for Scoliosis Screening Incorporating Deep Learning With Back Images

Explainable Deep-Learning-Based Gait Analysis of Hip-Knee Cyclogram for the Prediction of Adolescent Idiopathic Scoliosis Progression

Intelligent Scoliosis Screening and Diagnosis: A Survey

A Multi-Scale Keypoint Estimation Network with Self-Supervision for Spinal Curvature Assessment of Idiopathic Scoliosis from the Imperfect Dataset

Intelligent Evaluation Method for Scoliosis at Home Using Back Photos Captured by Mobile Phones

Spinal Curve Assessment of Idiopathic Scoliosis with a Small Dataset Via a Multi-Scale Keypoint Estimation Approach

Symmetric Perception and Ordinal Regression for Detecting Scoliosis Natural Image

A multi-stage ensemble network system to diagnose adolescent idiopathic scoliosis

Remote Pathological Gait Classification System

The relation between work-induced neuroendocrine reactivity and recovery, subjective need for recovery, and health status.

Wearable accelerometers reveal objective assessment of walking symmetry and regularity in idiopathic scoliosis patients

Automating Cobb Angle Measurement for Adolescent Idiopathic Scoliosis using Instance Segmentation

Classifying Simulated Gait Impairments using Privacy-preserving Explainable Artificial Intelligence and Mobile Phone Videos

Diagnosis of generalized joint hypermobility with gait patterns using a deep neural network

Lenke Classification of Scoliosis Based on Segmentation Network and Adaptive Shape Descriptor

Accounting for Biomechanical Measures from Musculoskeletal Simulation of Upright Posture Does Not Enhance the Prediction of Curve Progression in Adolescent Idiopathic Scoliosis

Design, reliability, and validity of a portable electronic device based on ergonomics for early screening of adolescent scoliosis

Copper-dependent depolymerization of lignin in the presence of fungal metabolite, pyridine.