Automatic segmentation of medical images using a novel Harris Hawk optimization method and an active contour model

Area of Science:

• Medical Imaging
• Computational Biology
• Artificial Intelligence

Background:

• Accurate medical image segmentation is crucial for disease diagnosis and treatment planning.
• Challenges in medical imaging include intensity inhomogeneity, noise, low contrast, and ill-defined boundaries, hindering automated segmentation.
• Existing methods often struggle with diverse imaging modalities and complex artifacts.

Purpose of the Study:

• To develop a fully automated medical image segmentation method.
• To address common challenges in medical image segmentation using a hybrid approach.
• To improve segmentation accuracy for skin and cardiac imaging.

Main Methods:

• A novel method integrating thresholding and an active contour model was proposed.
• Harris Hawks optimizer was employed to determine optimal thresholding values for initial contour generation.
• A spatially varying Gaussian kernel refined the active contour model for improved segmentation.

Main Results:

• The method was validated on standard skin (ISBI 2016) and cardiac (ACDC, MICCAI 2017) datasets.
• Superior segmentation results were achieved, with an overall Dice Score of 0.90 for the skin dataset.
• An overall Dice Score of 0.93 was obtained for the cardiac dataset, outperforming state-of-the-art algorithms.

Conclusions:

• The proposed automated method effectively segments regions of interest in medical images.
• The hybrid approach demonstrates superior performance compared to existing algorithms for both skin and cardiac datasets.
• This technique offers a promising solution for accurate and automated medical image segmentation.