Localization of prostatic tumor's infection based on normalized mutual information MRI image segmentation

Area of Science:

• Medical Imaging
• Radiotherapy
• Computational Anatomy

Background:

• Accurate prostate cancer localization is crucial for effective radiotherapy planning.
• Existing segmentation methods face challenges with image variability and processing time.

Purpose of the Study:

• To evaluate normalized mutual information (MRI) for prostate cancer segmentation and localization.
• To assess the role of this segmentation in guiding prostate cancer treatment and minimizing damage to surrounding organs.

Main Methods:

• Utilized normalized mutual information to measure image similarity and select relevant atlases.
• Applied global weighted voting and normalized mutual information for label image fusion.
• Performed multi-atlas segmentation leveraging structural similarities in medical images.

Main Results:

• Successfully segmented prostate MRI images, with basic localization achieved in 10 test cases.
• Demonstrated high accuracy in delineating the target area for radiotherapy.
• Identified potential for reducing damage to organs like the rectum and bladder.

Conclusions:

• Normalized mutual information-based multi-atlas segmentation offers high accuracy for prostate cancer localization.
• The method shows promise in improving radiotherapy precision but requires further optimization for accuracy and speed to achieve clinical practicality.