Abdomen disease diagnosis in CT images using flexiscale curvelet transform and improved genetic algorithm

Area of Science:

• Medical Imaging
• Computer Science
• Artificial Intelligence

Background:

• Computed tomography (CT) imaging is crucial for diagnosing abdomen diseases.
• Feature extraction from CT images is key for accurate diagnosis.
• Conventional methods may face limitations in optimizing feature extraction parameters.

Purpose of the Study:

• To develop an optimized feature extraction method for abdomen disease diagnosis using CT images.
• To improve the scale optimization of the flexi-scale curvelet transform.
• To enhance the performance of genetic algorithms in medical image analysis.

Main Methods:

• Flexi-scale curvelet transform for feature extraction from CT images.
• An improved genetic algorithm (GA) with modified elitism and novel chromosome combination strategy.
• Application of the proposed method to 120 abdominal CT images (normal, cysts, tumors, stones).

Main Results:

• The improved GA effectively optimized the scale parameters for the flexi-scale curvelet transform.
• Features extracted using the optimized transform were more discriminative than conventional methods.
• The system demonstrated potential for accurate abdomen disease diagnosis.

Conclusions:

• The proposed flexi-scale curvelet transform with an improved GA offers a promising approach for abdomen disease diagnosis.
• Enhanced feature extraction leads to better diagnostic accuracy in medical imaging.
• This method shows potential as a valuable diagnostic tool for various abdomen pathologies.