Bladder segmentation in MRI images using active region growing model.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computer Vision

Background:

• Prostate segmentation in MRI is challenging due to poor contrast at the bladder interface.
• Simultaneous segmentation of prostate and bladder using coupled models with non-overlapping constraints is a viable solution.
• Accurate pre-segmentation of organs is necessary for coupled segmentation approaches.

Purpose of the Study:

• To propose a fast deformable model for accurate bladder segmentation in MRI.
• To address segmentation difficulties at the prostate-bladder interface caused by poor contrast.
• To provide a reliable pre-segmentation method for coupled segmentation models.

Main Methods:

• A fast deformable model utilizing inflation and internal forces was developed.
• Forces were locally adapted based on gray-level information to enhance boundary adherence.
• The model was designed to overcome leakage issues at weak image edges.

Main Results:

• The algorithm demonstrated good performance across 33 MRI volumes from 5 different devices.
• The deformable model successfully produced smooth and accurate bladder surfaces.
• The method effectively handled poor contrast regions at the prostate-bladder interface.

Conclusions:

• The proposed fast deformable model offers an effective solution for bladder segmentation in MRI.
• This method enhances accuracy, particularly in challenging regions with poor contrast.
• The algorithm provides a robust pre-segmentation tool for coupled prostate and bladder analysis.