Multi-Disease Segmentation of Gliomas and White Matter Hyperintensities in the BraTS Data Using a 3D Convolutional Neural Network.

Front Comput Neurosci 2019 20;13:84. Epub 2019 Dec 20.

Department of Radiology & Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States.

An important challenge in segmenting real-world biomedical imaging data is the presence of multiple disease processes within individual subjects. Most adults above age 60 exhibit a variable degree of small vessel ischemic disease, as well as chronic infarcts, which will manifest as white matter hyperintensities (WMH) on brain MRIs. Subjects diagnosed with gliomas will also typically exhibit some degree of abnormal T2 signal due to WMH, rather than just due to tumor. We sought to develop a fully automated algorithm to distinguish and quantify these distinct disease processes within individual subjects' brain MRIs. To address this multi-disease problem, we trained a 3D U-Net to distinguish between abnormal signal arising from tumors vs. WMH in the 3D multi-parametric MRI (mpMRI, i.e., native T1-weighted, T1-post-contrast, T2, T2-FLAIR) scans of the International Brain Tumor Segmentation (BraTS) 2018 dataset ( = 285, = 66). Our trained neuroradiologist manually annotated WMH on the BraTS training subjects, finding that 69% of subjects had WMH. Our 3D U-Net model had a 4-channel 3D input patch (80 × 80 × 80) from mpMRI, four encoding and decoding layers, and an output of either four [background, active tumor (AT), necrotic core (NCR), peritumoral edematous/infiltrated tissue (ED)] or five classes (adding WMH as the fifth class). For both the four- and five-class output models, the median for whole tumor (WT) extent (i.e., union of AT, ED, NCR) was 0.92 in both training and validation sets. Notably, the five-class model achieved significantly ( = 0.002) lower/better Hausdorff distances for WT extent in the training subjects. There was strong positive correlation between manually segmented and predicted volumes for WT ( = 0.96) and WMH ( = 0.89). Larger lesion volumes were positively correlated with higher/better scores for WT ( = 0.33), WMH ( = 0.34), and across all lesions ( = 0.89) on a log(10) transformed scale. While the median for WMH was 0.42 across training subjects with WMH, the median was 0.62 for those with at least 5 cm of WMH. We anticipate the development of computational algorithms that are able to model multiple diseases within a single subject will be a critical step toward translating and integrating artificial intelligence systems into the heterogeneous real-world clinical workflow.

Download full-text PDF

Source
http://dx.doi.org/10.3389/fncom.2019.00084DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6933520PMC
December 2019

Publication Analysis

Top Keywords

training subjects
12
wmh
11
subjects wmh
8
white matter
8
abnormal signal
8
brain mris
8
matter hyperintensities
8
disease processes
8
processes individual
8
subjects
6
output [background
4
peritumoral edematous/infiltrated
4
edematous/infiltrated tissue
4
[background active
4
ncr peritumoral
4
core ncr
4
active tumor
4
tumor necrotic
4
necrotic core
4
tissue ed]
4

Altmetric Statistics

References

(Supplied by CrossRef)
A novel focal tversky loss function with improved attention u-net for lesion segmentation
Abraham et al.
Proceedings of the IEEE Comput Soc Conf Comput Vis Pattern Recogn Proceedings - International Symposium on Biomedical Imaging 2019-April 2019
Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the BRATS challenge.
Bakas et al.
arXiv 2019
Fully convolutional deep residual neural networks for brain tumor segmentation.
Chang et al.
Brain Lesion 2017
Deep learning: a primer for radiologists.
Chartrand et al.
Radiographics 2017
3D U-net: learning dense volumetric segmentation from sparse annotation
Cicek et al.
Medical Image Computing and Computer-Assisted Intervention – MICCAI 2016 2016

Similar Publications