Related Experiment Video
Updated: Aug 5, 2026
04:25
Manual Segmentation of the Human Choroid Plexus Using Brain MRI
Published on: December 15, 2023
Deep Learning-Based Detection, Classification, and Segmentation of Cerebral Microbleeds
Habip Eser Akkaya1, Önder Polat2, Hasan Çolakoğlu3
1Department of Radiology, Ankara Training and Research Hospital, Ankara, Türkiye.
The Eurasian Journal of Medicine
|July 30, 2026
Summary
This study explored using artificial intelligence (AI) to detect cerebral microhemorrhages on MRI scans. AI models show potential for aiding diagnosis, but require further development for clinical use.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in medicine
- Cerebrovascular disease research
Background:
- Cerebral microhemorrhages are linked to cerebrovascular events.
- Artificial intelligence (AI) integration in medicine requires scientific validation.
- Assessing AI's role in detecting microhemorrhages is crucial.
Purpose of the Study:
- To investigate the feasibility of using AI for detecting and assessing cerebral microhemorrhages.
- To evaluate deep learning models for microhemorrhage detection and segmentation in MRI data.
Main Methods:
- Retrospective analysis of 108 patients with microhemorrhages and 108 controls.
- Utilized cranial magnetic resonance imaging (MRI) with susceptibility-weighted sequences.
- Trained customized DenseNet for classification and nnU-Net for segmentation.
Main Results:
- Classification model achieved 80% AUC and 75% accuracy, with 68% specificity.
- Segmentation model showed 89.4% sensitivity and a 62.05% Dice similarity coefficient.
- AI models demonstrated capability in detecting and localizing microhemorrhages.
Conclusions:
- AI models show promise for detecting and localizing microhemorrhages in 3D MRI.
- The classification model's specificity needs improvement for clinical application.
- AI may serve as a supportive tool for microhemorrhage detection with further refinement.
