CNN Filter Learning from Drawn Markers for the Detection of Suggestive Signs of COVID-19 in CT Images.

Early detection of COVID-19 is vital to control its spread. Deep learning methods have been presented to detect suggestive signs of COVID-19 from chest CT images. However, due to the novelty of the disease, annotated volumetric data are scarce. Here we propose a method that does not require either large annotated datasets or backpropagation to estimate the filters of a convolutional neural network (CNN). For a few CT images, the user draws markers at representative normal and abnormal regions. The method generates a feature extractor composed of a sequence of convolutional layers, whose kernels are specialized in enhancing regions similar to the marked ones, and the decision layer of our CNN is a support vector machine. As we have no control over the CT image acquisition, we also propose an intensity standardization approach. Our method can achieve mean accuracy and kappa values of 0.97 and 0.93, respectively, on a dataset with 117 CT images extracted from different sites, surpassing its counterpart in all scenarios.

Automatic algorithm for quantifying lung involvement in patients with chronic obstructive pulmonary disease, infection with SARS-CoV-2, paracoccidioidomycosis and no lung disease patients.

In this work, we aimed to develop an automatic algorithm for the quantification of total volume and lung impairments in four different diseases. The quantification was completely automatic based upon high resolution computed tomography exams. The algorithm was capable of measuring volume and differentiating pulmonary involvement including inflammatory process and fibrosis, emphysema, and ground-glass opacities. The algorithm classifies the percentage of each pulmonary involvement when compared to the entire lung volume. Our algorithm was applied to four different patients groups: no lung disease patients, patients diagnosed with SARS-CoV-2, patients with chronic obstructive pulmonary disease, and patients with paracoccidioidomycosis. The quantification results were compared with a semi-automatic algorithm previously validated. Results confirmed that the automatic approach has a good agreement with the semi-automatic. Bland-Altman (B&A) demonstrated a low dispersion when comparing total lung volume, and also when comparing each lung impairment individually. Linear regression adjustment achieved an R value of 0.81 when comparing total lung volume between both methods. Our approach provides a reliable quantification process for physicians, thus impairments measurements contributes to support prognostic decisions in important lung diseases including the infection of SARS-CoV-2.