Dynamic Infrared Thermography Reveals Bilateral Thermal Recovery Asymmetry in Scoliosis.

Background: Repeated radiographic monitoring exposes scoliosis patients to cumulative radiation. Dynamic infrared thermography (DIT) may provide a non-invasive alternative by characterizing thermal recovery behavior following controlled cooling. This study investigated whether reheating-based thermographic analysis could reveal scoliosis-related asymmetry. Methods: A total of 13 healthy controls and 10 individuals with radiographically confirmed scoliosis were enrolled. Following controlled thoracolumbar cooling, DIT images were acquired. Bilateral paraspinal cold regions were segmented during a 5-min rewarming period. Dynamic thermal asymmetry features were extracted; Sustained Asymmetry Index (SAI) and Dynamic Thermal Asymmetry Index (DTAI) were calculated to quantify asymmetry persistence over time. Results: Mean Asymmetry demonstrated significant group-level separation (0.027 ± 0.021 vs. 0.079 ± 0.070, p = 0.038, AUC = 0.762). The SAI demonstrated the strongest discriminative performance (0.299 ± 0.114 vs. 0.521 ± 0.143, p = 0.0014, AUC = 0.900; sensitivity 0.70, specificity 1.00), indicating persistent bilateral thermal divergence throughout reheating. The DTAI was significantly higher in the scoliosis group (-0.405 ± 0.407 vs. 0.527 ± 0.878, p = 0.0084, AUC = 0.831). Inspection of recovery curves revealed distinct thermal recovery phenotypes, including persistent asymmetry, delayed asymmetry peaks, and near-normal thermal recovery patterns. No significant correlation was observed between DTAI and Cobb angle (ρ = 0.183, p = 0.611). Conclusions: DIT combined with quantitative thermal recovery analysis can reveal bilateral thermo-physiological asymmetry associated with scoliosis. Persistence-based thermal recovery metrics, particularly the SAI, demonstrated strong discriminative performance. These findings support the potential utility of DIT recovery analysis as a non-invasive physiological biomarker framework for scoliosis assessment.