Follow-up Of COVID-19 Patients During Recovery And Dynamic Study Of Lung CT Based On AI Quantitative Analysis

Objective: Analyze the characteristics of pulmonary CT of COVID-19 patients and explore the independent risk factors of residual lung lesions.At the same time,Artificial intelligence quantitative analysis technology was used to analyze the chest CT of covid-19 patients.The characteristics of quantitative indicators such as lesion volume and density over time were studied to establish the change curve of patient lung lesions over time,and combined with the patient’s clinical symptoms and the laboratory related indicators,auxiliary to recommend the appropriate lung CT examination time for COVID-19 patients.Materials and methods: A total of 319 patients were diagnosed by RT-PCR in three designated hospitals in Changsha,Xiangtan and Shaoyang,Hunan Province between January 2019 and March 2020.Patients younger than 18 years and without abnormal pulmonary manifestations were excluded.Clinical data(gender,age,epidemiological history,basic diseases,etc.),laboratory(blood routine,blood sedimentation,etc.)and pulmonary CT data during and after discharge were collected.All the patients were divided into mild and severe groups according to the novel coronavirus diagnosis and treatment guidelines.And comparing the differences in clinical,laboratory and pulmonary CT changes between the two groups.AI was used to quantify the previous lung CT to obtain quantitative indicators such as volume and density of lung lesions.SPSS25.0 was used to compare and analyze the differences.Results: About 107(50.95%)and 20(36.36%)patients had residual lung lesions at 2 and 6 months after the onset of the disease.The severe group was more likely to develop residual lesions than the mild group.High proportion of residual pulmonary lesions at 2 months and high complete absorption of pulmonary lesions at 6 months(p<0.05).The residual lesions were mainly ground glass and fiber bars.Age ≥ 46 years,Maximum CT score ≥10 were independent risk factors for residual CT abnormalities at 2、6 months(odds ratios of 6.5 and 0.1;12.9 and 32.7,p<0.05,each).AI was used to quantify the previous lung CT to obtain quantitative indicators such as volume and density of lung lesions.The maximum volume fraction of pulmonary lesions in the severe group was on the 16.28±9.55 days,which are later than mild patients(12.38±6.40days),and ababsorption is slower than mild patients.Elderly patients(>50 years old)and patients with leukocytosis had a higher percentage of lung lesion volume,and the absorption rate was slower than other groups.There was no significant difference in the extent of pulmonary lesions between male and female patients(p > 0.05).Conclusion: In the follow-up of COVID-19 patients,most of the patients’ lung lesions have different degrees of absorption.1/2 of the patients at 2 months and about 1/3 of the patients at 6 months have residual lesions,and the residual lesions are mostly fibrous cords,subpleural lines,grid shadows,etc.Patients with residual pulmonary lesions are older and have higher initial CT and maximum CT scores.These patients may have fibrotic lesions in the lungs for a long time,and long-term review of changes in pulmonary CT lesions and changes in pulmonary function is required.AI can effectively evaluate the changes of thoracic lesions in patients with COVID-19,which is expected to guide patients to perform lung CT reexamination,reduce radiation damage to patients and reduce the economic burden of the country and individuals.