2017
DOI: 10.3389/fninf.2017.00002
|View full text |Cite
|
Sign up to set email alerts
|

SEGMA: An Automatic SEGMentation Approach for Human Brain MRI Using Sliding Window and Random Forests

Abstract: Quantitative volumes from brain magnetic resonance imaging (MRI) acquired across the life course may be useful for investigating long term effects of risk and resilience factors for brain development and healthy aging, and for understanding early life determinants of adult brain structure. Therefore, there is an increasing need for automated segmentation tools that can be applied to images acquired at different life stages. We developed an automatic segmentation method for human brain MRI, where a sliding wind… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
8
0

Year Published

2017
2017
2022
2022

Publication Types

Select...
6
2

Relationship

0
8

Authors

Journals

citations
Cited by 16 publications
(8 citation statements)
references
References 63 publications
0
8
0
Order By: Relevance
“…Another aim was the development of novel computational methods for mapping growth and connectivity in development. While certain technical developments such as image segmentation and methods for studying crossing fibres are achievable with sample sizes of <100,19–22 larger sample sizes are needed to address other challenges. For example, larger atlases of the developing brain than are currently available are required to understand population diversity, and machine learning methods are being used to develop image biomarkers and to improve the interoperability of multisite acquisitions, which will enable researchers to increase study power, carry out essential replication studies and investigate risk and resilience in brain development conferred by the genome 23–25.…”
Section: Methods and Analysismentioning
confidence: 99%
See 1 more Smart Citation
“…Another aim was the development of novel computational methods for mapping growth and connectivity in development. While certain technical developments such as image segmentation and methods for studying crossing fibres are achievable with sample sizes of <100,19–22 larger sample sizes are needed to address other challenges. For example, larger atlases of the developing brain than are currently available are required to understand population diversity, and machine learning methods are being used to develop image biomarkers and to improve the interoperability of multisite acquisitions, which will enable researchers to increase study power, carry out essential replication studies and investigate risk and resilience in brain development conferred by the genome 23–25.…”
Section: Methods and Analysismentioning
confidence: 99%
“…Conventional images are reported by a paediatric radiologist using a structured system 29 30. We use image data to generate novel processing techniques optimised for neonatal data,11 19–21 31 and we will use these and other publicly available pipelines for processing neonatal data13 32 33 to derive image features for analyses with collateral data relating to exposures and outcomes. These include, but are not limited to, tract-based, morphometric and structural connectivity analyses 10–12 34–38…”
Section: Methods and Analysismentioning
confidence: 99%
“…Several classifiers can be used to relate the image features and the corresponding labels. Lately high performance classifiers such as random forest [12] have been used. In our proposed method, we have used a neural network-based classifier [13].…”
Section: Proposed Methodsmentioning
confidence: 99%
“…Patenaude et al [ 40 ] proposed a method that used manually labeled image training data, where the principles of both the active shape and appearance models were utilized within a Bayesian framework, allowing probabilistic relationships between shape and intensity to be fully used. Serag et al [ 41 ] employed high-dimensional feature vectors for segmenting brain subjects using a sliding window approach along with a multi-class random forest classifier. Wang et al [ 42 ] segmented T1, T2, and diffusion-weighted brain images using a sparse representation of the complementary tissue distribution.…”
Section: Introductionmentioning
confidence: 99%