Improvement of medical images with multi-objective genetic algorithm and adaptive morphological transformations

Area of Science:

• Medical Imaging
• Image Processing
• Computer Vision

Background:

• Medical image corruption (blurring, weakening) necessitates enhancement techniques.
• Previous methods often used fixed-size structuring elements in morphological operations.
• Constant coefficients in image enhancement formulas limit adaptability.

Purpose of the Study:

• To develop an adaptive medical image enhancement algorithm.
• To improve upon traditional morphological transformations by incorporating variable parameters.
• To achieve optimal image enhancement tailored to individual medical images.

Main Methods:

• Utilized top-hat and bottom-hat morphological transformations with variable-sized disk structuring elements.
• Incorporated variable coefficients into the image enhancement formula.
• Employed a genetic algorithm with a multi-objective fitting function to optimize transformation parameters (disk sizes, coefficients).

Main Results:

• Demonstrated successful enhancement of medical images through visual inspection and quantitative metrics (Entropy, SSIM, NIQE, BRISQE, AMBE, PSNR).
• The proposed method effectively preserves image information.
• Achieved noise amplification suppression and prevented significant brightness increases.

Conclusions:

• The novel approach using variable structuring elements and coefficients significantly enhances medical image quality.
• The adaptive nature of the algorithm ensures optimal results for diverse image conditions.
• This technique offers a robust solution for improving diagnostic accuracy through clearer medical images.