RESUMEN
Objectives: 1) To develop a deep learning (DL) pipeline allowing quantification of COVID-19 pulmonary lesions on low-dose computed tomography (LDCT). 2) To assess the prognostic value of DL-driven lesion quantification. Methods: This monocentric retrospective study included training and test datasets taken from 144 and 30 patients, respectively. The reference was the manual segmentation of 3 labels: normal lung, ground-glass opacity(GGO) and consolidation(Cons). Model performance was evaluated with technical metrics, disease volume and extent. Intra- and interobserver agreement were recorded. The prognostic value of DL-driven disease extent was assessed in 1621 distinct patients using C-statistics. The end point was a combined outcome defined as death, hospitalization>10 days, intensive care unit hospitalization or oxygen therapy. Results: The Dice coefficients for lesion (GGO+Cons) segmentations were 0.75±0.08, exceeding the values for human interobserver (0.70±0.08; 0.70±0.10) and intraobserver measures (0.72±0.09). DL-driven lesion quantification had a stronger correlation with the reference than inter- or intraobserver measures. After stepwise selection and adjustment for clinical characteristics, quantification significantly increased the prognostic accuracy of the model (0.82 vs. 0.90; p<0.0001). Conclusions: A DL-driven model can provide reproducible and accurate segmentation of COVID-19 lesions on LDCT. Automatic lesion quantification has independent prognostic value for the identification of high-risk patients.
RESUMEN
BACKGROUND: Low-dose chest CT (LDCT) showed high sensitivity and ability to quantify lung involvement of COVID-19 pneumopathy. The aim of this study was to describe the prevalence and risk factors for lung involvement in 247 patients with a visual score and assess the prevalence of incidental findings. METHODS: For 12 days in March 2020, 250 patients with RT-PCR positive tests and who underwent LDCT were prospectively included. Clinical and imaging findings were recorded. The extent of lung involvement was quantified using a score ranging from 0 to 40. A logistic regression model was used to explore factors associated with a score ≥ 10. RESULTS: A total of 247 patients were analyzed; 138 (54%) showed lung involvement. The mean score was 4.5 ± 6.5, and the mean score for patients with lung involvement was 8.1 ± 6.8 [1-31]. The mean age was 43 ± 15 years, with 121 males (48%) and 17 asymptomatic patients (7%). Multivariate analysis showed that age > 54 years (odds ratio 4.4[2.0-9.6] p < 0.001) and diabetes (4.7[1.0-22.1] p = 0.049) were risk factors for a score ≥ 10. Multivariate analysis including symptoms showed that only age > 54 years (4.1[1.7-10.0] p = 0.002) was a risk factor for a score ≥ 10. Rhinitis (0.3[0.1-0.7] p = 0.005) and anosmia (0.3[0.1-0.9] p = 0.043) were protective against lung involvement. Incidental imaging findings were found in 19% of patients, with a need for follow-up in 0.6%. CONCLUSION: The prevalence of lung involvement was 54% in a predominantly paucisymptomatic population. Age ≥ 55 years and diabetes were risk factors for significant parenchymal lung involvement. Rhinitis and anosmia were protective against LDCT abnormalities.