Automatic bone segmentation and bone-cartilage interface extraction for the shoulder joint from magnetic resonance...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Accurate segmentation of shoulder bones and bone-cartilage interfaces (BCI) from 3D MR images is crucial for diagnosing glenohumeral joint conditions.
• Manual segmentation is time-consuming and subject to inter-observer variability.
• Automated methods are needed to improve efficiency and consistency in shoulder imaging analysis.

Purpose of the Study:

• To develop and validate a statistical shape model for automated segmentation of the proximal humerus and scapula.
• To enable subsequent automated extraction of the bone-cartilage interface (BCI) from 3D MR images of the shoulder.
• To quantitatively assess the accuracy of the automated segmentation and BCI extraction using the Dice Similarity Coefficient (DSC).

Main Methods:

• A statistical shape model approach was employed for automated segmentation of the proximal humerus and scapula.
• 3D MR images of the shoulder from 25 healthy subjects were used.
• Automated segmentations were compared against manual segmentations using the Dice Similarity Coefficient (DSC).

Main Results:

• High mean DSC scores were achieved for automated bone segmentation: 0.926 ± 0.050 for the humerus and 0.837 ± 0.059 for the scapula.
• Accurate mean DSC values were obtained for BCI extraction: 0.806 ± 0.133 for the humerus and 0.795 ± 0.117 for the scapula.
• The model-based approach demonstrated successful automated bone segmentation and BCI extraction.

Conclusions:

• The presented statistical shape model approach effectively automates the segmentation of proximal humerus and scapula and BCI extraction from shoulder MR images.
• The high DSC scores indicate the reliability and accuracy of the automated method.
• This framework holds promise for future automated quantitative analysis of cartilage in the glenohumeral joint.