An effective and open source interactive 3D medical image segmentation solution

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Image Processing

Background:

• 3D medical image segmentation is crucial for clinical applications but suffers from variability.
• Existing automatic methods often lack user-centric control and flexibility.
• Interactive annotation tools are valuable for addressing segmentation ambiguities.

Purpose of the Study:

• To develop an effective and user-oriented interactive 3D medical image segmentation method.
• To create an open-source software solution for efficient and customizable annotation.
• To address the limitations of variability in automated segmentation outputs.

Main Methods:

• An adaptive dynamic programming approach is employed to integrate user interactions.
• Initial segmentation is achieved through feature-based geodesic computation.
• Refinement utilizes an efficient local updating scheme for real-time adjustments.

Main Results:

• The method successfully refines 3D segmentations using user input.
• The adaptive dynamic programming approach efficiently incorporates interactions.
• The software module, integrated into 3D Slicer, offers a user-friendly GUI.

Conclusions:

• The proposed interactive segmentation method is effective and efficient for 3D medical images.
• The open-source software provides a valuable tool for researchers and clinicians.
• The approach enhances segmentation accuracy and allows for user-specific customization.