Reliability and validity of adolescent idiopathic scoliosis diagnosis based on mobile phone AI scoliosis screening algorithm and infrared 3D spine assessment system.

BACKGROUND

Scoliosis affects 2–3% of the population, with adolescent idiopathic scoliosis (AIS) being the most common form. Early detection is crucial, but traditional Cobb angle measurement using X-rays has limitations, including radiation risks. A non-invasive, low-cost Infrared 3D spine system with AI was developed for radiation-free scoliosis screening and follow-up.

METHODS

From August 2023 to February 2024, 69 AIS-suspected patients (ages 6–18) were enrolled at West China Hospital. The diagnostic validity of a mobile AI screening algorithm and infrared 3D spine system was evaluated against X-ray and Electronic Scoliosis Scale, measuring Cobb and Angle of Trunk Rotation (ATR). Statistical analyses used SPSS, including paired t-tests, Wilcoxon Signed-Rank, Bland-Altman, and Pearson’s correlation to assess accuracy and reliability, with ICC > 0.80 indicating high reliability.

RESULTS

The study included 69 AIS cases (mean age 14.66). The diagnostic tool I3SAS showed 98.53% sensitivity, 100% specificity, and an AUC of 0.98. Intraobserver and interrater reliability for Cobb and ATR measurements were excellent, with ICC values of 0.997 to 0.998 for X-rays and 0.992 to 0.998 for the infrared system. Accuracy analysis showed minimal differences in Cobb angle (0.43) and ATR angle (-0.21), with strong Pearson’s correlations (0.964 to 0.989).

CONCLUSIONS

The mobile AI scoliosis screening algorithm and infrared 3D spine system are reliable, valid, and feasible tools for AIS diagnosis, with strong potential for clinical screening use.