Quantitative diagnosis of rotator cuff tears based on sonographic pattern recognition

Area of Science:

• Orthopedics
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Shoulder pain affects nearly 70% of people, often due to subacromial disorders.
• Rotator cuff tears, particularly in the supraspinatus tendon, are a severe form of these disorders.
• Shoulder ultrasound is a key diagnostic tool for supraspinatus tears.

Purpose of the Study:

• To develop a computer-aided tear classification (CTC) system for identifying supraspinatus tears.
• To enhance the accuracy and reduce inter-operator variability in ultrasound diagnosis of rotator cuff tears.
• To evaluate the clinical utility of an AI-driven diagnostic aid for supraspinatus tendinopathy and tears.

Main Methods:

• Utilized 191 ultrasound images (89 tendinopathy, 102 tears) from 136 patients.
• Extracted intensity and texture features from supraspinatus lesions.
• Developed a binary logistic regression classifier for lesion classification using extracted features.

Main Results:

• The CTC system achieved a high accuracy rate of 92% (176/191) in classifying supraspinatus lesions.
• Demonstrated a strong diagnostic performance with an area under the receiver operating characteristic curve (Az) of 0.9694.
• The system effectively differentiated between supraspinatus tendinopathy and tears.

Conclusions:

• The developed CTC system shows significant potential for clinical application in diagnosing supraspinatus tears.
• AI-assisted ultrasound analysis can improve diagnostic accuracy and consistency for rotator cuff injuries.
• This technology may aid clinicians in better managing shoulder pain and related disorders.