Automated Quantification of CT Patterns Associated with COVID-19 from Chest CT
Shikha Chaganti, Philippe Greniér, Abishek Balachandran, Guillaume Chabin, Stuart L. Cohen, Thomas Flohr, Bogdan Georgescu, Saša Grbić, Siqi Liu, François Mellot, Nicolas Murray, Savvas Nicolaou, William Parker, Thomas J. Re, Pina C. Sanelli, Alexander Sauter, Zhoubing Xu, Youngjin Yoo, Valentin Ziebandt, Dorin Comaniciu
Abstract
Purpose To present a method that automatically segments and quantifies abnormal CT patterns commonly present in COVID-19, namely ground-glass opacities and consolidations. Materials and Methods In this retrospective study, the proposed method takes as input a noncontrast chest CT and segments the lesions, lungs, and lobes in three dimensions, based on a dataset of 9749 chest CT volumes. The method outputs two combined measures of the severity of lung and lobe involvement, quantifying both the extent of COVID-19 abnormalities and presence of high opacities, based on deep learning and deep reinforcement learning. The first measure of (percentage of opacity, percentage of high opacity [PO, PHO]) is global, while the second of (lung severity score, lung high opacity score [LSS, LHOS]) is lobe-wise. Evaluation of the algorithm is reported on CT studies of 200 participants (100 COVID-19 confirmed patients and 100 healthy controls) from institutions from Canada, Europe, and the United States collected between 2002 and April 2020. Ground truth is established by manual annotations of lesions, lungs, and lobes. Correlation and regression analyses were performed to compare the prediction to the ground truth. Results Pearson correlation coefficient between method prediction and ground truth for COVID-19 cases was calculated as 0.92 for PO (P < .001), 0.97 for PHO (P < .001), 0.91 for LSS (P < .001), and 0.90 for LHOS (P < .001). Ninety-eight of 100 healthy controls had a predicted PO of less than 1%; two had between 1% and 2%. Automated processing time to compute the severity scores was 10 seconds per case compared with 30 minutes required for manual annotations. Conclusion A new method segments regions of CT abnormalities associated with COVID-19 and computes (PO, PHO), as well as (LSS, LHOS) severity scores. Keywords: CT, Lung, Segmentation/Vision/Application Domain, Quantification/Vision/Application Domain, Supervised Learning, Reinforcement Learning © RSNA, 2021 Supplemental material is available for this article.