Related Experiment Video
Updated: Nov 2, 2025
Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
Published on: September 25, 2019
icobrain ms 5.1: Combining unsupervised and supervised approaches for improving the detection of multiple sclerosis
Mladen Rakić1, Sophie Vercruyssen2, Simon Van Eyndhoven2
1icometrix, Leuven, Belgium; KU Leuven, Department of Electrical Engineering (ESAT), Processing Speech and Images (PSI) and Medical Imaging Research Center, 3001 Leuven, Belgium.
Abstract:
Multiple sclerosis (MS) is a chronic autoimmune, inflammatory neurological disease of the central nervous system. Its diagnosis nowadays commonly includes performing an MRI scan, as it is the most sensitive imaging test for MS. MS plaques are commonly identified from fluid-attenuated inversion recovery (FLAIR) images as hyperintense regions that are highly varying in terms of their shapes, sizes and locations, and are routinely classified in accordance to the McDonald criteria. Recent years have seen an increase in works that aimed at development of various semi-automatic and automatic methods for detection, segmentation and classification of MS plaques. In this paper, we present an automatic combined method, based on two pipelines: a traditional unsupervised machine learning technique and a deep-learning attention-gate 3D U-net network. The deep-learning network is specifically trained to address the weaker points of the traditional approach, namely difficulties in segmenting infratentorial and juxtacortical plaques in real-world clinical MRIs. It was trained and validated on a multi-center multi-scanner dataset that contains 159 cases, each with T1 weighted (T1w) and FLAIR images, as well as manual delineations of the MS plaques, segmented and validated by a panel of raters. The detection rate was quantified using lesion-wise Dice score. A simple label fusion is implemented to combine the output segmentations of the two pipelines. This combined method improves the detection of infratentorial and juxtacortical lesions by 14% and 31% respectively, in comparison to the unsupervised machine learning pipeline that was used as a performance assessment baseline.
Insights
This study introduces an automated method combining traditional machine learning and deep learning to improve the detection of multiple sclerosis (MS) brain lesions on MRI scans. The new approach enhances the identification of challenging infratentorial and juxtacortical MS plaques.
Area of Science:
- Neurology
- Medical Imaging
- Artificial Intelligence
Background:
- Multiple sclerosis (MS) is a chronic autoimmune neurological disease impacting the central nervous system.
- Magnetic Resonance Imaging (MRI), particularly fluid-attenuated inversion recovery (FLAIR) sequences, is crucial for diagnosing MS by identifying hyperintense plaques.
- Accurate segmentation of MS plaques, especially infratentorial and juxtacortical lesions, remains a challenge for automated methods.
Purpose of the Study:
- To develop and evaluate an automatic combined method for detecting and segmenting multiple sclerosis (MS) plaques in MRI scans.
- To improve the detection rates of infratentorial and juxtacortical MS plaques, which are often difficult to segment.
- To leverage both traditional unsupervised machine learning and deep learning techniques for enhanced lesion segmentation.
Main Methods:
- A hybrid approach combining a traditional unsupervised machine learning pipeline with a deep-learning attention-gate 3D U-net network was developed.
- The deep learning model was specifically trained to overcome limitations of the traditional method in segmenting challenging MS plaque locations.
- The method was trained and validated on a multi-center, multi-scanner dataset of 159 cases with T1-weighted and FLAIR MRI images and expert-annotated MS plaques.
Main Results:
- The combined method demonstrated improved detection rates for MS plaques compared to the baseline unsupervised machine learning pipeline.
- Specifically, the detection of infratentorial lesions improved by 14%, and juxtacortical lesions improved by 31%.
- Lesion-wise Dice score was used to quantify the detection rate, indicating robust performance.
Conclusions:
- The proposed automatic combined method effectively enhances the segmentation of multiple sclerosis plaques, particularly in challenging anatomical regions.
- This hybrid approach integrating deep learning with traditional machine learning offers a promising advancement for automated MS lesion detection in clinical practice.
- The improved detection of infratentorial and juxtacortical lesions signifies a step forward in the automated analysis of MS neuroimaging data.
