Related Experiment Video
Updated: Sep 24, 2025
12:30
A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures
Published on: July 2, 2014
20.5K
Beyond automatic medical image segmentation-the spectrum between fully manual and fully automatic delineation
Michael J Trimpl1,2,3, Sergey Primakov4, Philippe Lambin4
1Mirada Medical Ltd, Oxford, United Kingdom.
Physics in Medicine and Biology
|May 6, 2022
Summary
This review explores AI-driven medical image segmentation, focusing on balancing user interaction and data needs for accurate contouring. It highlights interactive AI as a promising alternative to fully automatic or manual methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Manual segmentation is time-consuming and inconsistent.
- Deep learning has advanced fully automatic segmentation but requires extensive data and clinician review.
- Current methods face challenges in balancing automation, user interaction, and data availability.
Purpose of the Study:
- To review segmentation methods between manual and fully automatic approaches.
- To explore effective user interaction strategies in AI-assisted segmentation.
- To present emerging avenues for improving medical image segmentation.
Main Methods:
- Review of existing literature on semi-automatic and fully automatic segmentation.
- Analysis of deep learning applications in medical image segmentation.
- Exploration of human-AI interaction models for contouring.
Main Results:
- Fully automatic methods show promise but often need manual correction.
- Interactive AI methods offer a spectrum of user involvement.
- The optimal balance between automation and interaction is key.
Conclusions:
- AI-driven interactive segmentation is an emerging field with significant potential.
- Effective clinician-AI collaboration can enhance segmentation accuracy and efficiency.
- Further research is needed to optimize these hybrid approaches for clinical practice.
