2020
DOI: 10.48550/arxiv.2012.08317
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

Multiple Sclerosis Lesion Segmentation -- A Survey of Supervised CNN-Based Methods

Abstract: Lesion segmentation is a core task for quantitative analysis of MRI scans of Multiple Sclerosis patients. The recent success of deep learning techniques in a variety of medical image analysis applications has renewed community interest in this challenging problem and led to a burst of activity for new algorithm development. In this survey, we investigate the supervised CNN-based methods for MS lesion segmentation. We decouple these reviewed works into their algorithmic components and discuss each separately. F… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1

Citation Types

0
4
0

Year Published

2021
2021
2023
2023

Publication Types

Select...
2

Relationship

0
2

Authors

Journals

citations
Cited by 2 publications
(4 citation statements)
references
References 90 publications
0
4
0
Order By: Relevance
“…Compared Methods. Many state-of-the-art methods, most of which are fully-supervised, 16 have provided their results on the ISBI challenge data set. Among the transfer learning methods, Valverde et al 17 trained the networks with two public datasets (MICCAI08, 18 MICCAI16) and then fine-tuned with a single subject from the ISBI dataset.…”
Section: Experimental Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…Compared Methods. Many state-of-the-art methods, most of which are fully-supervised, 16 have provided their results on the ISBI challenge data set. Among the transfer learning methods, Valverde et al 17 trained the networks with two public datasets (MICCAI08, 18 MICCAI16) and then fine-tuned with a single subject from the ISBI dataset.…”
Section: Experimental Methodsmentioning
confidence: 99%
“…Middle, transfer learning (TL) methods. Bottom, the top 5 fully-supervised methods listed by 1,16. Other top-10 methods scored in the [90.07, 92.12] range.Here, we trained models on MICCAI16+UMCL and experimented with Focal loss, L2 loss, batch normalization (BN) and instance normalization (IN).…”
mentioning
confidence: 99%
“…Danelakis et al [9] reviewed MS segmentation methods published between 2013 to 2018; and Kaur et al [10] included three methods described in 2019 in addition to earlier methods. Finally, Zhang et al [11] reviewed deeplearning methods up to 2020. However, a combined analysis of both statistical and deep-learning methods is lacking, as is a comprehensive discussion of their clinical integration and application.…”
Section: Introductionmentioning
confidence: 99%
“…In this work, we present a comprehensive survey of MS lesion segmentation methods published prior to 2021, unifying previous statistical modeling approaches with stateof-the-art deep-learning frameworks. We specifically discuss the potential for these segmentation methods to address an urgent clinical need for quantitative lesion activity analysis (the measurement of longitudinal lesion change); and also review domain adaptation solutions to lesion segmentation in multicenter, multi-scanner user scenarios [5]- [11].…”
Section: Introductionmentioning
confidence: 99%