ResectVol: A tool to automatically segment and characterize lacunas in brain images
Raphael F Casseb1, Brunno M de Campos1, Marcia Morita-Sherman2
1Neuroimaging Laboratory, Department of Neurology, University of Campinas, Campinas, Brazil.
Epilepsia Open
|October 5, 2021
Summary
A new tool, ResectVol, automatically segments surgical cavities in epilepsy patient MRIs, improving accuracy over manual methods for better surgical outcome prediction.
Area of Science:
- Neurosurgery
- Medical Imaging Analysis
- Epilepsy Research
Background:
- Epilepsy surgery is crucial for pharmacoresistant focal epilepsies.
- Accurate prediction of surgical outcomes remains challenging.
- Manual segmentation of resected tissue (lacuna) is common but biased and lacks detail.
Purpose of the Study:
- To develop and validate a novel automated tool for segmenting and characterizing surgical lacunas in postoperative MRI scans of epilepsy patients.
- To provide a user-friendly, automatic solution for quantifying resected tissue volume and characteristics.
Main Methods:
- A MATLAB-based pipeline using SPM12 was developed to create 3D masks and estimate lacuna volume.
- The automated segmentation tool, ResectVol, was compared against manual segmentations.
- Validation involved 51 MRI scans from epilepsy patients who underwent temporal lobe resections.
Main Results:
- The ResectVol tool provides a user-friendly graphical or command-line interface.
- Automated and manual segmentations achieved a median Dice similarity coefficient of 0.77 (IQR: 0.71-0.81).
Conclusions:
- ResectVol offers a reliable and automated method for surgical lacuna segmentation and volume estimation.
- This tool facilitates advanced analytical techniques, including machine learning, for improved surgical outcome prediction in epilepsy.
- Automated analysis enables better coregistration with preoperative findings and other imaging modalities.
