Carotid artery segmentation using level set method with double adaptive threshold (DATLS) on TOF-MRA images.

Area of Science:

• Medical Imaging
• Image Segmentation
• Computational Anatomy

Background:

• Accurate segmentation of the carotid artery lumen is crucial for diagnosing atherosclerosis and stroke risk.
• Time-of-flight Magnetic Resonance Angiography (TOF-MRA) is a key imaging modality for visualizing carotid arteries.
• Segmentation of TOF-MRA images can be challenging due to weak vessel boundaries.

Purpose of the Study:

• To develop and evaluate a semi-automatic method for segmenting the carotid artery lumen on TOF-MRA images.
• To improve segmentation accuracy, particularly in cases with ill-defined vessel borders.
• To compare the proposed method against traditional level set techniques.

Main Methods:

• Modified a traditional level set method by incorporating a Double Adaptive Threshold (DATLS).
• Two thresholds were derived from the probability density function of an initial region of interest to guide contour generation and evolution.
• The DATLS method was validated on TOF-MRA datasets, comparing its performance against manual segmentation and a traditional level set approach using Dice Similarity Coefficient (DSC), Mean Contour Distance (MCD), and Hausdorff Distance (HD).

Main Results:

• The DATLS method demonstrated superior performance compared to the traditional level set, achieving a higher DSC (0.88 ± 0.07 vs. 0.75 ± 0.15) and lower MCD and HD.
• Across 283 cases, DATLS achieved an average DSC of 0.87 ± 0.09, MCD of 0.64 ± 0.87 mm, and HD of 1.76 ± 2.23 mm.
• DATLS exhibited better intra- and inter-operator reproducibility than manual delineation for DSC, MCD, and HD.

Conclusions:

• The proposed DATLS method significantly outperforms the traditional level set in terms of accuracy, robustness, and computational efficiency for carotid artery lumen segmentation on TOF-MRA.
• DATLS shows promise as a valuable tool for semi-automatic segmentation of carotid arteries in TOF-MRA, especially when dealing with challenging images.
• The method's improved reproducibility suggests its potential for reliable clinical application.