Related Experiment Video
Updated: Jul 31, 2025
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
2.5K
Fully automated imaging protocol independent system for pituitary adenoma segmentation: a convolutional neural
Martin Černý1,2, Jan Kybic3, Martin Májovský4
1Department of Neurosurgery and Neurooncology, 1st Faculty of Medicine, Charles University, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic. dr.martin.cerny@gmail.com.
Neurosurgical Review
|May 10, 2023
Summary
This study introduces an automated system for pituitary adenoma segmentation in MRI scans, achieving high accuracy across diverse imaging protocols. The system shows promise for clinical applications, though further development is needed.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuroscience
Background:
- Pituitary adenoma segmentation from MRI is crucial for diagnosis and treatment planning.
- Current segmentation methods often require manual input and are sensitive to imaging protocol variations.
- Developing an automated, protocol-independent system is essential for clinical workflow efficiency.
Purpose of the Study:
- To develop and evaluate a fully automated, imaging protocol-independent system for pituitary adenoma segmentation using artificial neural networks.
- To assess the accuracy and clinical utility of the automated segmentation system.
- To compare the system's performance against human expert evaluation.
Main Methods:
- Two independent artificial neural networks were trained on 394 patient MRI scans acquired across various protocols and field strengths (1.5T and 3T).
- A segmentation model assigned pixel-wise labels (pituitary adenoma, internal carotid artery, normal pituitary gland, background).
- A slice selection model identified clinically relevant slices, and performance was evaluated on 99 patients during training and 28 prospectively.
Main Results:
- The segmentation model achieved Dice coefficients of 0.910 for tumour, 0.719 for internal carotid artery, and 0.240 for normal gland.
- The slice selection model demonstrated 82.5% accuracy, 88.7% sensitivity, 76.7% specificity, and an AUC of 0.904.
- Human expert review rated 71.4% of segmentations as accurate, indicating good clinical comparability.
Conclusions:
- The developed automated system demonstrates robust performance and generalization across diverse MRI protocols for pituitary adenoma segmentation.
- The system's accuracy is comparable to recent studies and shows potential for clinical application, pending further development of clinical frameworks.
- This protocol-independent approach represents a significant advancement in automated medical image analysis for neuro-oncology.
