Related Experiment Video
Updated: Sep 2, 2026
Clinical Efficacy of an Innovative Multidimensional Traction Therapy in Moderate Adolescent Idiopathic Scoliosis
Published on: February 10, 2026
Surface Topography and Machine Learning: Strides Towards Radiation-Free Scoliosis Assessment: A Systematic Review
Paolo Brigato1,2, Davide Palombi3, Justin L Reyes4
1Departments of Medicine and Surgery, Research Unit of Orthopaedic and Trauma Surgery, Università Campus Bio-Medico di Roma.
Background:
Assessment of idiopathic scoliosis (IS) relies on radiographs and measurement of coronal curve magnitude, but repeated radiation exposure raises safety concerns in children. Noninvasive surface topography (ST) combined with machine learning (ML) has emerged as a potential radiation-free alternative for estimating curve severity, predicting coronal curve magnitude (Cobb angle), and monitoring longitudinally. This review aims to systematically review current evidence on the application of ML techniques to ST-based assessment of IS.
Methods:
This systematic review was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. PubMed, Embase, Scopus, Cochrane Library, and Web of Science were searched from database inception through August 2025 for studies applying ML algorithms to noninvasive ST methods in IS. Extracted data included study design, patient characteristics, surface acquisition techniques, ML model architecture, validation strategies, and performance metrics. Methodological quality and risk of bias were assessed using the QUADAS-2 tool.
Results:
Twelve studies met the inclusion criteria, encompassing 5015 patients and more than 6000 trunk surface scans. ST techniques included laser scanning, Moiré topography, raster stereography, depth sensors, and markerless 3D imaging. ML models demonstrated high accuracy in estimating curve severity and predicting coronal curve magnitude, with mean absolute errors of ∼3 to 6 degrees and sensitivities exceeding 0.9, approaching the variability of radiographic measurements. Curve-type classification showed lower and more heterogeneous performance, whereas binary classification models demonstrated strong potential for radiation-free screening and follow-up.
Conclusions:
ML applied to ST represents a promising noninvasive and radiation-free approach for IS assessment. Current evidence supports clinically meaningful performance in estimating curve severity and coronal curve magnitude, although limitations persist in classifying curve type. Larger prospective and multicenter studies are needed to validate these methods and define their role in clinical practice.
Level Of Evidence:
Level III.
