Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation
Multiple sclerosis (MS) lesion segmentation is crucial for monitoring disease progression. Deep learning methods have shown promising results but suffer from domain shift problems when evaluated in data from different protocols or scanners. Transfer learning (TL) achieves successful domain adaptatio...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | NeuroImage: Clinical |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2213158225000658 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850235165135077376 |
|---|---|
| author | Luisana Álvarez Sergi Valverde Àlex Rovira Xavier Lladó |
| author_facet | Luisana Álvarez Sergi Valverde Àlex Rovira Xavier Lladó |
| author_sort | Luisana Álvarez |
| collection | DOAJ |
| description | Multiple sclerosis (MS) lesion segmentation is crucial for monitoring disease progression. Deep learning methods have shown promising results but suffer from domain shift problems when evaluated in data from different protocols or scanners. Transfer learning (TL) achieves successful domain adaptation, but can lead to catastrophic forgetting, resulting in a significant performance drop on the source domain. Continuous learning aims to address this issue by retaining knowledge from previous domains while adapting to new ones. This work applies Elastic Weight Consolidation (EWC) for the first time in the context of domain-incremental learning for MS lesion segmentation. The approach was evaluated using a 3D U-Net trained on public datasets (WMH2017 and Shifts) and fine-tuned on an in-house dataset using both TL and EWC, in both full training and few-shot scenarios. Results show that with only 3 training images from the target domain, EWC leads to a 10% improvement in F-score, while using 5 images achieves similar results to using all available training images. Catastrophic forgetting was reduced by 8%–19% compared to standard TL, where performance drops ranged from 20 to 37%. This work demonstrates that EWC enables models to adapt to new domains while preserving previous knowledge, with minimal data requirements, advancing towards more generalizable deep learning models for clinical MS applications. |
| format | Article |
| id | doaj-art-eaf41285de644766840a7f2d013f4a49 |
| institution | OA Journals |
| issn | 2213-1582 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage: Clinical |
| spelling | doaj-art-eaf41285de644766840a7f2d013f4a492025-08-20T02:02:24ZengElsevierNeuroImage: Clinical2213-15822025-01-014610379510.1016/j.nicl.2025.103795Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidationLuisana Álvarez0Sergi Valverde1Àlex Rovira2Xavier Lladó3Vicorob Institute, University of Girona, Girona, Spain; Tensor Medical, Girona, Spain; Corresponding author at: Tensor Medical, Girona, Spain.Tensor Medical, Girona, SpainSection of Neuroradiology, Department of Radiology (IDI), Vall d’Hebron University Hospital, Spain, Universitat Autònoma de Barcelona, Barcelona, SpainVicorob Institute, University of Girona, Girona, SpainMultiple sclerosis (MS) lesion segmentation is crucial for monitoring disease progression. Deep learning methods have shown promising results but suffer from domain shift problems when evaluated in data from different protocols or scanners. Transfer learning (TL) achieves successful domain adaptation, but can lead to catastrophic forgetting, resulting in a significant performance drop on the source domain. Continuous learning aims to address this issue by retaining knowledge from previous domains while adapting to new ones. This work applies Elastic Weight Consolidation (EWC) for the first time in the context of domain-incremental learning for MS lesion segmentation. The approach was evaluated using a 3D U-Net trained on public datasets (WMH2017 and Shifts) and fine-tuned on an in-house dataset using both TL and EWC, in both full training and few-shot scenarios. Results show that with only 3 training images from the target domain, EWC leads to a 10% improvement in F-score, while using 5 images achieves similar results to using all available training images. Catastrophic forgetting was reduced by 8%–19% compared to standard TL, where performance drops ranged from 20 to 37%. This work demonstrates that EWC enables models to adapt to new domains while preserving previous knowledge, with minimal data requirements, advancing towards more generalizable deep learning models for clinical MS applications.http://www.sciencedirect.com/science/article/pii/S2213158225000658Multiple sclerosisLesion segmentationContinuous learningTransfer learningCatastrophic forgetting |
| spellingShingle | Luisana Álvarez Sergi Valverde Àlex Rovira Xavier Lladó Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation NeuroImage: Clinical Multiple sclerosis Lesion segmentation Continuous learning Transfer learning Catastrophic forgetting |
| title | Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation |
| title_full | Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation |
| title_fullStr | Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation |
| title_full_unstemmed | Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation |
| title_short | Mitigating catastrophic forgetting in Multiple sclerosis lesion segmentation using elastic weight consolidation |
| title_sort | mitigating catastrophic forgetting in multiple sclerosis lesion segmentation using elastic weight consolidation |
| topic | Multiple sclerosis Lesion segmentation Continuous learning Transfer learning Catastrophic forgetting |
| url | http://www.sciencedirect.com/science/article/pii/S2213158225000658 |
| work_keys_str_mv | AT luisanaalvarez mitigatingcatastrophicforgettinginmultiplesclerosislesionsegmentationusingelasticweightconsolidation AT sergivalverde mitigatingcatastrophicforgettinginmultiplesclerosislesionsegmentationusingelasticweightconsolidation AT alexrovira mitigatingcatastrophicforgettinginmultiplesclerosislesionsegmentationusingelasticweightconsolidation AT xavierllado mitigatingcatastrophicforgettinginmultiplesclerosislesionsegmentationusingelasticweightconsolidation |