Related Experiment Video
Updated: Sep 17, 2025
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
2.9K
Focal cortical dysplasia (type II) detection with multi-modal MRI and a deep-learning framework
Anand Shankar1, Manob Jyoti Saikia2, Samarendra Dandapat3
1Department of Electronics and Communication Engineering, Indian Institute of Information Technology Guwahati, Guwahati, Assam, 781015, India.
Npj Imaging
|July 2, 2025
Summary
Focal cortical dysplasia type II (FCD-II), a brain malformation causing epilepsy, can be effectively analyzed using deep learning (DL) MRI techniques. The DenseNet201 model demonstrated superior performance in identifying FCD-II, aiding diagnosis and treatment planning.
Area of Science:
- Neuroimaging
- Medical Artificial Intelligence
- Developmental Neuroscience
Background:
- Focal cortical dysplasia type II (FCD-II) is a significant brain malformation linked to drug-resistant epilepsy and cognitive deficits.
- Magnetic Resonance Imaging (MRI) analysis is vital for FCD-II diagnosis, surgical planning, and postoperative care.
- Deep learning (DL) offers potential for enhancing the accuracy and efficiency of FCD-II analysis.
Purpose of the Study:
- To identify the most suitable deep learning (DL) model for analyzing Magnetic Resonance Imaging (MRI) data of Focal Cortical Dysplasia Type II (FCD-II).
- To evaluate the performance of different DL models across various MRI modalities and image planes for FCD-II detection.
Main Methods:
- A comprehensive study evaluated six distinct deep learning (DL) models.
- Analysis included T1w and FLAIR MRI modalities across axial, coronal, and sagittal planes.
- Demographic (age, sex) and clinical (hemisphere, lobes) data were incorporated into the analysis.
Main Results:
- The DenseNet201 model exhibited superior performance in classifying FCD-II.
- DenseNet201 achieved high precision, F1-score, and a large area under the ROC and PR curves.
- The model's effectiveness was demonstrated across different imaging parameters and patient characteristics.
Conclusions:
- DenseNet201 is a highly suitable DL model for the accurate analysis of FCD-II from MRI data.
- This finding supports the integration of advanced DL techniques for improved FCD-II diagnosis and patient management.
- Optimized DL analysis can significantly enhance presurgical planning and postoperative care for FCD-II patients.
