Automatic spinal disease diagnoses assisted by 3D unaligned transverse CT slices

Area of Science:

• Medical Imaging
• Computational Anatomy
• Spinal Diagnostics

Background:

• Accurate 3D reconstruction of spinal anatomy from medical imaging is crucial for diagnosing diseases.
• Existing methods often struggle with unaligned or arbitrarily spaced imaging slices.
• Spinal conditions like disc herniation and deformities require precise quantitative assessment.

Purpose of the Study:

• To develop a robust 3D reconstruction method for unaligned transverse spinal slices.
• To create an automatic diagnostic tool for identifying spinal diseases from these reconstructions.
• To enable qualitative and quantitative analysis of spinal pathologies.

Main Methods:

• Extended the Marching Cubes algorithm for generating triangulated isosurfaces from unaligned slices.
• Developed an automatic analysis of disc and vertebral body boundaries on transverse slices.
• Implemented algorithms to estimate disc herniation, canal compression, and spinal curve deformities.

Main Results:

• Successfully generated 3D isosurfaces from arbitrarily angled and spaced unaligned transverse slices.
• The system accurately estimated the presence and extent of disc herniation and canal compression.
• Identified and quantified deformities in the spinal curve.
• Demonstrated utility as a qualitative and quantitative diagnostic tool.

Conclusions:

• The proposed 3D reconstruction and automatic diagnosis methods are effective for spinal diseases.
• This approach overcomes limitations of dealing with unaligned transverse slices.
• The system offers a valuable tool for enhanced spinal disease diagnosis and management.