Evaluation of Operator Variability and Validation of an AI-Assisted α-Angle Measurement System for DDH Using a...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pediatric Orthopedics

Background:

• Ultrasound using the Graf method is standard for developmental dysplasia of the hip (DDH) detection.
• Significant intra- and inter-operator variability limits the reliability of manual hip ultrasound assessments.
• Automated measurement systems are needed to enhance reproducibility in DDH screening.

Purpose of the Study:

• To quantify operator variability in hip ultrasound assessments for DDH.
• To validate an AI-assisted system for automated alpha-angle measurement.
• To improve the reproducibility and reliability of hip ultrasound in DDH screening.

Main Methods:

• Thirty participants (clinicians, residents, students) performed hip ultrasound scans on a standardized infant hip phantom.
• Analysis included examination time, iliac margin inclination, and alpha-angle measurements to assess variability.
• An AI system was developed for automated landmark detection and alpha-angle calculation from static images and dynamic videos.

Main Results:

• Clinicians showed shorter examination times and better reproducibility than less experienced operators.
• Manual alpha-angle measurements consistently underestimated the reference value.
• AI-assisted dynamic ultrasound analysis achieved the highest accuracy (mean 69.2°) and consistency, outperforming manual measurements.

Conclusions:

• Substantial operator variability exists in manual hip ultrasound assessments for DDH.
• AI-assisted dynamic ultrasound analysis offers improved accuracy and consistency over manual methods.
• This technology holds promise for enhancing the reliability of routine DDH screening.