Detection of white matter lesion regions in MRI using SLIC0 and convolutional neural network

Diniz, Pedro Henrique Bandeira; Valente, Thales Levi Azevedo; Diniz, João Otávio Bandeira; Silva, Aristófanes Corrêa; Gattass, Marcelo; Ventura, Nina; Muniz, Bernardo Carvalho; Gasparetto, Emerson L.

doi:10.1016/j.cmpb.2018.04.011

Cited by 31 publications

(20 citation statements)

References 96 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…All studies which reported ground truth generation details, validated the WMH segmentation method with ground truth binary masks generated using the Fluid Attenuated Inversion Recovery (FLAIR) MRI sequence. However, only three studies reported having used only the FLAIR sequence in their segmentation framework (Bandeira Diniz et al, 2018;Knight et al, 2018;Schirmer et al, 2019). From the rest (i.e., 34/37) that described using data from different sequences, 28 used a combination of more than one sequence (i.e., also known as "multispectral approach"), generally T1-weighted and FLAIR, to generate the final outcome.…”

Section: Pre-processing Methodsmentioning

confidence: 99%

“…From the 10 studies that proposed an unsupervised segmentation method (i.e., 27% of the total number of studies included), one used deep learning (Atlason et al, 2019). In total, eleven studies used Convolutional Neural Networks (Rachmadi, et al, 2018;Li et al, 2018;Guerrero et al, 2017;Moeskops et al, 2018;Ghafoorian et al, 2016;Hong et al, 2020;Manjón et al, 2018;Liu et al, 2020;Diniz et al, 2018;Schirmer et al, 2019), four studies proposed a method based on k-nearest neighbours (k-NN) (Sundaresan et al, 2019;Jiang et al, 2018;Ling et al, 2018;Griffanti et al, 2016), four studies proposed regression models (Knight et al, 2018;Dadar et al, 2017a;Zhan et al, 2017;Ding et al, 2020), and three studies used Random forest (RF) in their proposed algorithms (Stone et al, 2016;Park et al, 2018;Roy et al, 2015). Two studies proposed a method based on Fuzzy C mean algorithm (Zhan et al, 2015;Valverde et al, 2017) and three proposed improvements to a Gaussian Mixture Model framework (Sudre et al, 2015(Sudre et al, , 2017Fiford et al, 2020), both unsupervised methods.…”

Section: Segmentation Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Balakrishnan

Hernández

Farrall

2020

Preprint

View full text Add to dashboard Cite

Background: White matter hyperintensities (WMH), of presumed vascular origin, are visible and quantifiable neuroradiological markers of brain parenchymal change. These changes may range from damage secondary to inflammation and other neurological conditions, through to healthy ageing. Fully automatic WMH quantification methods are promising, but still, traditional semi-automatic methods seem to be preferred in clinical research. We systematically reviewed the literature for fully automatic methods developed in the last five years, to assess what are considered state-of-the-art techniques, as well as trends in the analysis of WMH of presumed vascular origin. Method: We registered the systematic review protocol with the International Prospective Register of Systematic Reviews (PROSPERO), registration number - CRD42019132200. We conducted the search for fully automatic methods developed from 2015 to July 2020 on Medline, Science direct, IEE Explore, and Web of Science. We assessed risk of bias and applicability of the studies using QUADAS 2. Results: The search yielded 2327 papers after removing 104 duplicates. After screening titles, abstracts and full text, 37 were selected for detailed analysis. Of these, 16 proposed a supervised segmentation method, 10 proposed an unsupervised segmentation method, and 11 proposed a deep learning segmentation method. Average DSC values ranged from 0.538 to 0.93, being the highest value obtained from a deep learning segmentation method. Only four studies validated their method in longitudinal samples, and eight performed an additional validation using clinical parameters. Only 8/37 studies made available their method in public repositories. Conclusions: Although deep learning methods reported highly accurate results, we found no evidence that favours them over the more established k-NN, linear regression and unsupervised methods in this task. Data and code availability, bias in study design and ground truth generation influence the wider validation and applicability of these methods in clinical research.

show abstract

Section: Pre-processing Methodsmentioning

confidence: 99%

Section: Segmentation Methodsmentioning

confidence: 99%

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Balakrishnan

Hernández

Farrall

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…All studies which reported ground truth generation details, validated the WMH segmentation method with ground truth binary masks, generated using the FLAIR MRI sequence. However, only three studies reported having used only the FLAIR sequence in their segmentation framework (Bandeira Diniz et al, 2018;Knight et al, 2018;Schirmer et al, 2019). Oft the rest (i.e., 34/37) which described using data from different sequences, 28 used a combination of more than one sequence (i.e., also known as "multispectral approach"), generally T1-weighted and FLAIR, to generate the final outcome.…”

Section: Pre-processing Methodsmentioning

confidence: 99%

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Balakrishnan¹,

Valdés-Hernández²,

Farrall³

2020

Preprint

View full text Add to dashboard Cite

Background: White matter hyperintensities (WMH), of presumed vascular origin, are visible and quantifiable neuroradiological markers of brain parenchymal change. These changes may range from damage secondary to inflammation and other neurological conditions, through to healthy ageing. Fully automatic WMH quantification methods are promising, but still, traditional semi-automatic methods seem to be preferred in clinical research. We systematically reviewed the literature for fully automatic methods developed in the last five years, to assess what are considered state-of-the-art techniques, as well as trends in the analysis of WMH of presumed vascular origin. Method: We registered the systematic review protocol with the International Prospective Register of Systematic Reviews (PROSPERO), registration number - CRD42019132200. We conducted the search for fully automatic methods developed from 2015 to July 2020 on Medline, Science direct, IEE Explore, and Web of Science. We assessed risk of bias and applicability of the studies using QUADAS 2. Results: The search yielded 2327 papers after removing 104 duplicates. After screening titles, abstracts and full text, 37 were selected for detailed analysis. Of these, 16 proposed a supervised segmentation method, 10 proposed an unsupervised segmentation method, and 11 proposed a deep learning segmentation method. Average DSC values ranged from 0.538 to 0.91, being the highest value obtained from an unsupervised segmentation method. Only four studies validated their method in longitudinal samples, and eight performed an additional validation using clinical parameters. Only 8/37 studies made available their method in public repositories. Conclusions: We found no evidence that favours deep learning methods over the more established k-NN, linear regression and unsupervised methods in this task. Data and code availability, bias in study design and ground truth generation influence the wider validation and applicability of these methods in clinical research.

show abstract

“…A técnica de superpixel reduz significativamente o custo de memória, enquanto potencialmente aumenta a precisão de detecção. As técnicas de superpixels analisadas foram o agrupamento iterativo linear simples (SLIC) [11], e a sua variação SLIC otimizada (SLICO) [12].…”

Section: Extração Dos Superpixels Baseado No Algoritmo Psounclassified

Classification of brain lesions on magnetic resonance imaging using superpixel, PSO and convolutional neural network

Cipriano

Silva

Ferreira

et al. 2019

Anais Do XV Workshop De Visão Computacional (WVC 2019)

View full text Add to dashboard Cite

One of the most severe and common brain tumors is gliomas. Manual classiﬁcation of injuries of this type is a laborious task in the clinical routine. Therefore, this work proposes an automatic method to classify lesions in the brain in 3D MR images based on superpixels, PSO algorithm and convolutional neural network. The proposed method obtained results for the complete, central and active regions, an accuracy of 87.88%, 70.51%, 80.08% and precision of 76%, 84%, 75% for the respective regions. The results demonstrate the difﬁculty of the network in the classiﬁcation of the regions found in the lesions.

show abstract

Detection of white matter lesion regions in MRI using SLIC0 and convolutional neural network

Cited by 31 publications

References 96 publications

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Automatic Segmentation of White Matter Hyperintensities from Brain Magnetic Resonance Images in the Era of Deep Learning and Big Data – A Systematic Review

Classification of brain lesions on magnetic resonance imaging using superpixel, PSO and convolutional neural network

Contact Info

Product

Resources

About