2018
DOI: 10.1007/978-3-030-00320-3_11
|View full text |Cite
|
Sign up to set email alerts
|

Transfer Learning for Task Adaptation of Brain Lesion Assessment and Prediction of Brain Abnormalities Progression/Regression Using Irregularity Age Map in Brain MRI

Abstract: The Irregularity Age Map (IAM) for the unsupervised assessment of brain white matter hyperintensities (WMH) opens several opportunities in machine learning-based MRI analysis, including transfer task adaptation learning in the segmentation and prediction of brain lesion progression and regression. The lack of need for manual labels is useful for transfer learning. Whereas the nature of IAM itself can be exploited for predicting lesion progression/regression. In this study, we propose the use of task adaptation… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3

Citation Types

0
3
0

Year Published

2019
2019
2021
2021

Publication Types

Select...
5

Relationship

3
2

Authors

Journals

citations
Cited by 5 publications
(3 citation statements)
references
References 10 publications
0
3
0
Order By: Relevance
“…the task domain changes), 3) when performing tasks that are related to but not the same as that on which they were trained (e.g. lesion segmentation vs. lesion assessment) ( Rachmadi et al, 2018 ). To overcome these limitations, there are several ways to enhance the performance of CNN architectures without modifying the architecture itself.…”
Section: Introductionmentioning
confidence: 99%
“…the task domain changes), 3) when performing tasks that are related to but not the same as that on which they were trained (e.g. lesion segmentation vs. lesion assessment) ( Rachmadi et al, 2018 ). To overcome these limitations, there are several ways to enhance the performance of CNN architectures without modifying the architecture itself.…”
Section: Introductionmentioning
confidence: 99%
“…Another study pre-trained CNN using natural images for segmentation of neonatal to adult brain images (Xu et al, 2017), and other study pre-trained a CNN for brain lesion segmentation using MRI data acquired with other protocols (Ghafoorian et al, 2017). Task adaptation transfer learning has been applied to WMH segmentation, by teaching a CNN to “learn" to detect texture irregularities instead of binary expert-delineated WMH segmentations (Rachmadi et al, 2018a).…”
Section: Introductionmentioning
confidence: 99%
“…In this paper, we propose to use IAM as an additional input data to train a U-Net neural network architecture for WMH segmentation, owed to the fact that LOTS-IM can easily produce IAM without the need for training using manually marked WMH ground-truth data. U-Net architecture is selected as a base model for our experiments as it has shown the best learning performance using IAM (Rachmadi et al, 2018a). To address the incorporation of IAM to U-Net for WMH segmentation, we propose feed-forwarding IAM as regional map to a Saliency U-Net architecture.…”
Section: Introductionmentioning
confidence: 99%