Automatic Classification and Segmentation of Low-Grade Gliomas in Magnetic Resonance Imaging

Barbosa, Marta L. S.; Ribeiro, Rogério; Coelho, Luís

doi:10.1007/978-3-030-17065-3_5

Cited by 2 publications

(1 citation statement)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After the full-text screening, 223 studies are included for synthesis. Among them, 61 are physics or mathematics-based, 13–74 156 are deep learning-based and six are software-based or semi-automatic 75–80 methods articles.…”

Section: Resultsmentioning

confidence: 99%

Magnetic resonance image-based brain tumour segmentation methods: A systematic review

2022

View full text Add to dashboard Cite

Background Image segmentation is an essential step in the analysis and subsequent characterisation of brain tumours through magnetic resonance imaging. In the literature, segmentation methods are empowered by open-access magnetic resonance imaging datasets, such as the brain tumour segmentation dataset. Moreover, with the increased use of artificial intelligence methods in medical imaging, access to larger data repositories has become vital in method development. Purpose To determine what automated brain tumour segmentation techniques can medical imaging specialists and clinicians use to identify tumour components, compared to manual segmentation. Methods We conducted a systematic review of 572 brain tumour segmentation studies during 2015–2020. We reviewed segmentation techniques using T1-weighted, T2-weighted, gadolinium-enhanced T1-weighted, fluid-attenuated inversion recovery, diffusion-weighted and perfusion-weighted magnetic resonance imaging sequences. Moreover, we assessed physics or mathematics-based methods, deep learning methods, and software-based or semi-automatic methods, as applied to magnetic resonance imaging techniques. Particularly, we synthesised each method as per the utilised magnetic resonance imaging sequences, study population, technical approach (such as deep learning) and performance score measures (such as Dice score). Statistical tests We compared median Dice score in segmenting the whole tumour, tumour core and enhanced tumour. Results We found that T1-weighted, gadolinium-enhanced T1-weighted, T2-weighted and fluid-attenuated inversion recovery magnetic resonance imaging are used the most in various segmentation algorithms. However, there is limited use of perfusion-weighted and diffusion-weighted magnetic resonance imaging. Moreover, we found that the U-Net deep learning technology is cited the most, and has high accuracy (Dice score 0.9) for magnetic resonance imaging-based brain tumour segmentation. Conclusion U-Net is a promising deep learning technology for magnetic resonance imaging-based brain tumour segmentation. The community should be encouraged to contribute open-access datasets so training, testing and validation of deep learning algorithms can be improved, particularly for diffusion- and perfusion-weighted magnetic resonance imaging, where there are limited datasets available.

show abstract

Section: Resultsmentioning

confidence: 99%

Magnetic resonance image-based brain tumour segmentation methods: A systematic review

2022

View full text Add to dashboard Cite

show abstract

Analysis of Lower-Grade Gliomas in MRI Through Segmentation and Genomic Cluster-Shape Feature Correlation

Paradkar

2022

Preprint

View full text Add to dashboard Cite

Gliomas, which originate from glial cells, are considered the most aggressive type of brain tumors. Currently, glioma research efforts are focused mainly on high-grade gliomas. This project aims to analyze lower-grade gliomas (LGG) in MRI and extend the understanding of LGGs. LGG segmentation, which outlines tumors in MRIs, is crucial to developing effective treatment plans. However, segmentation performed manually by radiologists is tedious, time-consuming, and often leads to inter-observer variability. Another unexplored area of LGG research is genomic subtypes. These subtypes can play a large factor in how LGGs can be treated, however there is currently no noninvasive method of identifying these subtypes. Recent studies suggest that LGG shape features have a correlation with genomic subtypes and should be investigated as a viable factor in LGG treatment options. This presents a need for additional research as most LGGs eventually develop into high-grade gliomas. The specific aims of this project include analyzing LGGs through deep learning-based segmentation, shape feature extraction, and statistical analysis to identify correlation between selected shape features and genomic subtypes. To realize these goals, programs were written and run using a publicly available LGG dataset. In terms of automatic segmentation, two models were created using different convolutional neural networks (CNN). The highest performing model used U-Net with a ResNeXt-50 backbone and yielded a 91.4% accuracy in terms of mean intersection over union (IoU) after testing. Shape feature extraction included three geometric features and 4 radiomic features which quantified tumor shape in 2D and 3D. Angular Standard Deviation (ASD), Margin Fluctuation (MF), Bounding Ellipsoid Volume Ratio (BEVR), Elongation, Major and Minor Axis Length, and Volume were tested for correlation with genomic subtype data using 49 Pearson chi-squared tests. P-values less than or equal to 0.05 suggested correlation. Statistical analysis found 16 statistically significant associations. The strongest associations were between MF and the RNASeq cluster (p < 0.00003), ASD and the RNASeq Cluster (p < 0.0005), and volume and the RPPA cluster (p < 0.0035).

show abstract

Automatic Classification and Segmentation of Low-Grade Gliomas in Magnetic Resonance Imaging

Cited by 2 publications

References 11 publications

Magnetic resonance image-based brain tumour segmentation methods: A systematic review

Magnetic resonance image-based brain tumour segmentation methods: A systematic review

Analysis of Lower-Grade Gliomas in MRI Through Segmentation and Genomic Cluster-Shape Feature Correlation

Contact Info

Product

Resources

About