Computer-Aided Cobb Measurement Based on Automatic Detection of Vertebral Slopes Using Deep Neural Network

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Scoliosis assessment relies on Cobb angle measurement, which suffers from high variability.
• Accurate and objective scoliosis assessment is critical for effective treatment planning.

Purpose of the Study:

• To develop a computer-aided method to decrease Cobb angle measurement variability.
• To assess the performance of a deep neural network (DNN) for automatic Cobb angle calculation.

Main Methods:

• A DNN was trained using vertebral patches from spinal radiographs.
• The DNN predicted vertebral slopes to automatically calculate Cobb angles.
• Measurements were compared between the DNN system and manual methods by an experienced surgeon and two examiners on model and in vivo radiographs.

Main Results:

• The DNN system demonstrated high repeatability on model radiographs (ICC > 0.98, MAE < 3°).
• Reliability was lower for in vivo radiographs, with differences exceeding 5° compared to manual measurements.
• DNN performance is dependent on the quantity and type of training data.

Conclusions:

• Sufficient training data, including in vivo radiographs, is essential for improving DNN performance in Cobb angle measurement.
• This computer-aided system offers potential for reliable and objective scoliosis assessments through automatic Cobb angle measurement.