COVID-19 Volumetric Pulmonary Lesion Estimation on CT Images Based on Probabilistic Active Contour and CNN Segmentation

Purpose: A semiautomatic two-step methodology is proposed to obtain a volumetric estimation of the COVID-19 related lesions on CT images.Methods: The first step consists of lesion segmentation using a non-supervised approach based on probabilistic active contours, making it interesting because it does not need a large dataset to train the model. The second step uses a CNN to segment the lung parenchyma and then obtain a whole-lung volume estimation. Finally, the resultant masks are used to compute the percentage of the lesion in the lungs. The proposed approach is validated using a publicly available dataset with 20 CT COVID-19 labeled images. A comparison of the amount of lesion on survived and deceased patients was performed with other 295 low and high-resolution CT images from COVID-19 patients.Results: The median dice coefficient value for the 20 segmentation evaluation images was 0.66. The results show a significant difference in percentages of lesion between deceased and survived patients, with a p-value of 9.1 x 10-4 in low-resolution and 5.1 x 10-5 for high-resolution images. Furthermore, an average difference of 10% in the percentages of lesion between high and low resolutions images was found.Conclusion: This methodology could be useful to estimate the amount of lesion and this estimated lesion could be useful to differentiate between survived and deceased patients. The low variation between the estimated percentage of lesion in high and low resolution could be important to take into account in those places where high-resolution CT is difficult to access.