Evaluation of a computer-aided method for measuring the Cobb angle on chest X-rays.

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Scoliosis diagnosis relies on Cobb angle measurement from X-rays.
• Manual measurement is subjective and time-consuming.
• Automating this process can improve efficiency and consistency.

Purpose of the Study:

• To develop and evaluate a computer-aided method for automatic Cobb angle measurement on chest X-rays.
• To assess the reliability and accuracy of this method compared to manual measurements.
• To determine the effectiveness of the computer-aided method in diagnosing scoliosis.

Main Methods:

• Utilized two Mask R-CNN models to detect and segment the spine and vertebrae.
• Measured Cobb angles automatically from the segmented outputs.
• Compared automated measurements with manual measurements from two radiologists on 248 chest X-rays.

Main Results:

• The computer-aided method demonstrated good reliability (ICC=0.854) with a mean absolute difference of 3.32° compared to manual methods.
• Intra- and inter-observer reliability for manual measurements were high (ICC=0.941 and 0.887).
• The method achieved 89.59% sensitivity and 70.37% specificity for scoliosis diagnosis.

Conclusions:

• The computer-aided method shows significant potential for automated Cobb angle measurement.
• This technology can aid in the reliable diagnosis of scoliosis from chest X-rays.
• Further development could enhance specificity for clinical application.