Unsupervised segmentation and quantification of COVID-19 lesions on computed Tomography scans using CycleGAN.

BACKGROUND: Lesion segmentation is a critical step in medical image analysis, and methods to identify pathology without time-intensive manual labeling of data are of utmost importance during a pandemic and in resource-constrained healthcare settings. Here, we describe a method for fully automated segmentation and quantification of pathological COVID-19 lung tissue on chest Computed Tomography (CT) scans without the need for manually segmented training data. METHODS: We trained a cycle-consistent generative adversarial network (CycleGAN) to convert images of COVID-19 scans into their generated healthy equivalents. Subtraction of the generated healthy images from their corresponding original CT scans yielded maps of pathological tissue, without background lung parenchyma, fissures, airways, or vessels. We then used these maps to construct three-dimensional lesion segmentations. Using a validation dataset, Dice scores were computed for our lesion segmentations and other published segmentation networks using ground truth segmentations reviewed by radiologists. RESULTS: The COVID-to-Healthy generator eliminated high Hounsfield unit (HU) voxels within pulmonary lesions and replaced them with lower HU voxels. The generator did not distort normal anatomy such as vessels, airways, or fissures. The generated healthy images had higher gas content (2.45 ± 0.93 vs 3.01 ± 0.84 L, P < 0.001) and lower tissue density (1.27 ± 0.40 vs 0.73 ± 0.29 Kg, P < 0.001) than their corresponding original COVID-19 images, and they were not significantly different from those of the healthy images (P < 0.001). Using the validation dataset, lesion segmentations scored an average Dice score of 55.9, comparable to other weakly supervised networks that do require manual segmentations. CONCLUSION: Our CycleGAN model successfully segmented pulmonary lesions in mild and severe COVID-19 cases. Our model's performance was comparable to other published models; however, our model is unique in its ability to segment lesions without the need for manual segmentations.

Lung distribution of gas and blood volume in critically ill COVID-19 patients: a quantitative dual-energy computed tomography study.

BACKGROUND: Critically ill COVID-19 patients have pathophysiological lung features characterized by perfusion abnormalities. However, to date no study has evaluated whether the changes in the distribution of pulmonary gas and blood volume are associated with the severity of gas-exchange impairment and the type of respiratory support (non-invasive versus invasive) in patients with severe COVID-19 pneumonia. METHODS: This was a single-center, retrospective cohort study conducted in a tertiary care hospital in Northern Italy during the first pandemic wave. Pulmonary gas and blood distribution was assessed using a technique for quantitative analysis of dual-energy computed tomography. Lung aeration loss (reflected by percentage of normally aerated lung tissue) and the extent of gas:blood volume mismatch (percentage of non-aerated, perfused lung tissue-shunt; aerated, non-perfused dead space; and non-aerated/non-perfused regions) were evaluated in critically ill COVID-19 patients with different clinical severity as reflected by the need for non-invasive or invasive respiratory support. RESULTS: Thirty-five patients admitted to the intensive care unit between February 29th and May 30th, 2020 were included. Patients requiring invasive versus non-invasive mechanical ventilation had both a lower percentage of normally aerated lung tissue (median [interquartile range] 33% [24-49%] vs. 63% [44-68%], p < 0.001); and a larger extent of gas:blood volume mismatch (43% [30-49%] vs. 25% [14-28%], p = 0.001), due to higher shunt (23% [15-32%] vs. 5% [2-16%], p = 0.001) and non-aerated/non perfused regions (5% [3-10%] vs. 1% [0-2%], p = 0.001). The PaO2/FiO2 ratio correlated positively with normally aerated tissue (ρ = 0.730, p < 0.001) and negatively with the extent of gas-blood volume mismatch (ρ = - 0.633, p < 0.001). CONCLUSIONS: In critically ill patients with severe COVID-19 pneumonia, the need for invasive mechanical ventilation and oxygenation impairment were associated with loss of aeration and the extent of gas:blood volume mismatch.