Automated segmentation method for spinal column based on a dual elliptic column model and its application for virtual...

Area of Science:

• Medical Imaging
• Radiology
• Computational Anatomy

Background:

• Accurate segmentation of vertebral bones in computed tomographic (CT) data is crucial for image-based radiological tasks.
• Segmenting affected spines, particularly those with conditions like scoliosis or bone metastases, presents significant challenges.

Purpose of the Study:

• To propose and evaluate a new method for segmenting thoracic and lumbar vertebral bodies from thin-slice CT images.
• To demonstrate the feasibility of a deformable model-based segmentation scheme on clinical datasets with various bone diseases.
• To apply the segmentation algorithm for virtual straightening of the thoracolumbar spine.

Main Methods:

• A deformable model-based segmentation scheme was developed for thoracic and lumbar vertebral bodies.
• The method was applied to thin-slice computed tomographic images.
• The algorithm was utilized for virtual straightening of the thoracolumbar spine.

Main Results:

• The proposed segmentation method demonstrated applicability to spines affected by scoliosis and multiple bone metastases.
• Results were validated on a database comprising 16 patients.
• The technique proved effective in handling complex spinal pathologies.

Conclusions:

• The developed deformable model-based segmentation method is effective for analyzing vertebral bones in CT data.
• This approach is applicable to clinical datasets featuring spinal deformities and metastatic diseases.
• The method shows potential for improving image-based radiological assessments and interventions.