Validation of phalanx bone three-dimensional surface segmentation from computed tomography images using laser...

Area of Science:

• Medical imaging
• Anatomical modeling
• Biomechanical analysis

Background:

• Accurate anatomical modeling is crucial for biomechanical studies and surgical planning.
• Manual segmentation of bone structures from medical imaging data, such as computed tomography (CT), is a common practice.
• Validating the accuracy of manual segmentation against precise measurement techniques is essential.

Purpose of the Study:

• To assess the accuracy of manually defined bony regions of interest from CT scans.
• To evaluate the impact of smoothing algorithms on the accuracy of bone surface representation.
• To compare CT-based segmentation with laser surface scanning for validation.

Main Methods:

• Coronal reformatted CT images of index finger phalanges from five cadavers were used.
• Two smoothing algorithms (image-based and Laplacian surface-based) were applied to segmented surfaces.
• Segmented surfaces were compared to laser surface scans of the same specimens.

Main Results:

• Manual segmentation demonstrated high overlap (0.91 average) between tracers.
• The mean difference between manual, unsmoothed surfaces and laser scans was 0.20 mm.
• Both image-based (0.21 mm) and Laplacian (0.20 mm) smoothing showed minimal impact on accuracy.

Conclusions:

• Manual segmentation of CT data provides accurate bone surface geometry.
• Smoothing techniques do not significantly alter the accuracy of CT-based bone surface representations.
• The validation method is applicable to other bones and segmentation/filtering techniques.