Kidney segmentation in CT sequences using graph cuts based active contours model and contextual continuity.

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Renal Pathology

Background:

• Accurate segmentation of renal tissues in computed tomography (CT) images is crucial for computer-aided diagnosis and pathology detection.
• Clinical applications require reliable tools for managing renal diagnosis and treatments.

Purpose of the Study:

• To develop a novel radiology tool for extracting renal tissues from CT image sequences.
• To enhance the accuracy and efficiency of kidney segmentation in abdominal CT scans.

Main Methods:

• A graph-cuts-based active contours model with an adaptive narrow band was proposed for kidney extraction.
• Segmentation is performed slice-by-slice using contextual continuity and interactive initialization on adjacent slices.
• An energy function combining boundary and regional information is optimized, with an adaptive search range determined by contextual continuity and object size.

Main Results:

• The method was evaluated on 30 CT image sequences containing normal and pathological renal tissues.
• Average Dice similarity coefficients ranged from 92.37% to 95.71%, with standard deviations between 2.18% and 3.87%.
• The average automatic segmentation time per kidney per slice was approximately 0.36 seconds.

Conclusions:

• The proposed algorithm integrates graph cuts, active contours, and contextual continuity for effective kidney segmentation in CT sequences.
• The method leverages energy minimization and image sequence characteristics for robust performance.
• The developed tool achieves accurate and efficient renal tissue extraction, supporting clinical diagnosis and treatment.