Ivan Coronado1, Refaat E Gabr1, Ponnada A Narayana1. 1. Department of Diagnostic and Interventional Imaging, The University of Texas Health Science Center at Houston (UTHealth), Houston, TX, USA.
Abstract
OBJECTIVE: The aim of this study is to assess the performance of deep learning convolutional neural networks (CNNs) in segmenting gadolinium-enhancing lesions using a large cohort of multiple sclerosis (MS) patients. METHODS: A three-dimensional (3D) CNN model was trained for segmentation of gadolinium-enhancing lesions using multispectral magnetic resonance imaging data (MRI) from 1006 relapsing-remitting MS patients. The network performance was evaluated for three combinations of multispectral MRI used as input: (U5) fluid-attenuated inversion recovery (FLAIR), T2-weighted, proton density-weighted, and pre- and post-contrast T1-weighted images; (U2) pre- and post-contrast T1-weighted images; and (U1) only post-contrast T1-weighted images. Segmentation performance was evaluated using the Dice similarity coefficient (DSC) and lesion-wise true-positive (TPR) and false-positive (FPR) rates. Performance was also evaluated as a function of enhancing lesion volume. RESULTS: The DSC/TPR/FPR values averaged over all the enhancing lesion sizes were 0.77/0.90/0.23 using the U5 model. These values for the largest enhancement volumes (>500 mm3) were 0.81/0.97/0.04. For U2, the average DSC/TPR/FPR values were 0.72/0.86/0.31. Comparable performance was observed with U1. For all types of input, the network performance degraded with decreased enhancement size. CONCLUSION: Excellent segmentation of enhancing lesions was observed for enhancement volume ⩾70 mm3. The best performance was achieved when the input included all five multispectral image sets.
OBJECTIVE: The aim of this study is to assess the performance of deep learning convolutional neural networks (CNNs) in segmenting gadolinium-enhancing lesions using a large cohort of multiple sclerosis (MS) patients. METHODS: A three-dimensional (3D) CNN model was trained for segmentation of gadolinium-enhancing lesions using multispectral magnetic resonance imaging data (MRI) from 1006 relapsing-remitting MS patients. The network performance was evaluated for three combinations of multispectral MRI used as input: (U5) fluid-attenuated inversion recovery (FLAIR), T2-weighted, proton density-weighted, and pre- and post-contrast T1-weighted images; (U2) pre- and post-contrast T1-weighted images; and (U1) only post-contrast T1-weighted images. Segmentation performance was evaluated using the Dice similarity coefficient (DSC) and lesion-wise true-positive (TPR) and false-positive (FPR) rates. Performance was also evaluated as a function of enhancing lesion volume. RESULTS: The DSC/TPR/FPR values averaged over all the enhancing lesion sizes were 0.77/0.90/0.23 using the U5 model. These values for the largest enhancement volumes (>500 mm3) were 0.81/0.97/0.04. For U2, the average DSC/TPR/FPR values were 0.72/0.86/0.31. Comparable performance was observed with U1. For all types of input, the network performance degraded with decreased enhancement size. CONCLUSION: Excellent segmentation of enhancing lesions was observed for enhancement volume ⩾70 mm3. The best performance was achieved when the input included all five multispectral image sets.
Entities:
Keywords:
Convolutional neural networks; MRI; active lesions; artificial intelligence; false positive; white matter lesions
Authors: Balasrinivasa Rao Sajja; Sushmita Datta; Renjie He; Meghana Mehta; Rakesh K Gupta; Jerry S Wolinsky; Ponnada A Narayana Journal: Ann Biomed Eng Date: 2006-03-09 Impact factor: 3.934
Authors: Refaat E Gabr; Ivan Coronado; Melvin Robinson; Sheeba J Sujit; Sushmita Datta; Xiaojun Sun; William J Allen; Fred D Lublin; Jerry S Wolinsky; Ponnada A Narayana Journal: Mult Scler Date: 2019-06-13 Impact factor: 6.312
Authors: Y Miki; R I Grossman; J K Udupa; S Samarasekera; M A van Buchem; B S Cooney; S N Pollack; D L Kolson; C Constantinescu; M Polansky; L J Mannon Journal: AJNR Am J Neuroradiol Date: 1997-04 Impact factor: 3.825
Authors: Fred D Lublin; Stephen C Reingold; Jeffrey A Cohen; Gary R Cutter; Per Soelberg Sørensen; Alan J Thompson; Jerry S Wolinsky; Laura J Balcer; Brenda Banwell; Frederik Barkhof; Bruce Bebo; Peter A Calabresi; Michel Clanet; Giancarlo Comi; Robert J Fox; Mark S Freedman; Andrew D Goodman; Matilde Inglese; Ludwig Kappos; Bernd C Kieseier; John A Lincoln; Catherine Lubetzki; Aaron E Miller; Xavier Montalban; Paul W O'Connor; John Petkau; Carlo Pozzilli; Richard A Rudick; Maria Pia Sormani; Olaf Stüve; Emmanuelle Waubant; Chris H Polman Journal: Neurology Date: 2014-05-28 Impact factor: 9.910
Authors: L Danieli; L Roccatagliata; D Distefano; E Prodi; G C Riccitelli; A Diociasi; L Carmisciano; A Cianfoni; T Bartalena; A Kaelin-Lang; C Gobbi; C Zecca; E Pravatà Journal: AJNR Am J Neuroradiol Date: 2022-05-26 Impact factor: 4.966
Authors: Hugo Vrenken; Mark Jenkinson; Dzung L Pham; Charles R G Guttmann; Deborah Pareto; Michel Paardekooper; Alexandra de Sitter; Maria A Rocca; Viktor Wottschel; M Jorge Cardoso; Frederik Barkhof Journal: Neurology Date: 2021-10-04 Impact factor: 9.910