Automated model-based rib cage segmentation and labeling in CT images

Area of Science:

• Medical Imaging
• Computer Vision
• Anatomy

Background:

• Automated analysis of medical images is crucial for efficient diagnosis.
• Accurate segmentation of anatomical structures like the rib cage aids in various clinical applications.
• Current methods for rib cage segmentation can be time-consuming and prone to error.

Purpose of the Study:

• To develop and evaluate a novel model-based approach for automated rib cage labeling and segmentation in chest CT scans.
• To assess the accuracy and robustness of the proposed method on unseen data.

Main Methods:

• Creation of a mean rib cage and vertebral column model from 29 CT datasets.
• Development of a ray search-based procedure for initial rib cage detection and model pose estimation.
• Adaptation of the model to 18 unseen CT datasets for segmentation.

Main Results:

• Successful detection, labeling, and segmentation of the rib cage in 16 out of 18 (89%) unseen CT datasets.
• Achieved a mean segmentation error of less than 1.3 mm between the true and detected object surfaces.
• Rib cage detection failed in one dataset, and automated labeling failed in another.

Conclusions:

• The proposed model-based approach demonstrates high accuracy and potential for automated rib cage segmentation in chest CT scans.
• The method shows promise for improving efficiency and consistency in radiological analysis.
• Further refinement may be needed to address failure cases in detection and labeling.