Background During COVID-19, the main manifestations of the disease are not only pneumonia but also coagulation disorders. The purpose of this study was to evaluate pulmonary vascular abnormalities 3 months after hospitalization for SARS-CoV-2 pneumonia in patients with persistent respiratory symptoms. Methods Among the 320 patients who participated in a systematic follow-up 3 months after hospitalization, 76 patients had residual symptoms justifying a specialized follow-up in the department of pulmonology. Among them, dual-energy CT angiography (DECTA) was obtained in 55 patients. Findings The 55 patients had partial ( n = 40; 72.7%) or complete ( n = 15; 27.3%) resolution of COVID-19 lung infiltration. DECTA was normal in 52 patients (52/55; 94.6%) and showed endoluminal filling defects in 3 patients (3/55; 5.4%) at the level of one ( n = 1) and two ( n = 1) segmental arteries of a single lobe and within central and peripheral arteries ( n = 1). DECT lung perfusion was rated as non-interpretable ( n = 2;3.6%), normal ( n = 17; 30.9%) and abnormal ( n = 36; 65.5%), the latter group comprising 32 patients with residual COVID-19 opacities (32/36; 89%) and 4 patients with normal lung parenchyma (4/36; 11%). Perfusion abnormalities consisted of (a) patchy defects (30/36; 83%), (b) PE-type defects (6/36; 16.6%) with ( n = 1) or without proximal thrombosis ( n = 5); and (c) focal areas of hypoperfusion (2/36; 5.5%). Increased perfusion was seen in 15 patients, always matching GGOs, bands and/or vascular tree-in- bud patterns. Interpretation DECT depicted proximal arterial thrombosis in 5.4% of patients and perfusion abnormalities suggestive of widespread microangiopathy in 65.5% of patients. Lung microcirculation was abnormal in 4 patients with normal lung parenchyma.

No abstract

The radiologic community is rapidly integrating a revolution that has not fully entered daily practice. It necessitates a close collaboration between computer scientists and radiologists to move from concepts to practical applications. This article reviews the current littérature on machine learning and deep neural network applications in the field of pulmonary embolism, chronic thromboembolic pulmonary hypertension, aorta, and chronic obstructive pulmonary disease.