tetano
Editor, Senior Moderator
Cureus
. 2022 Feb 14;14(2):e22203.
doi: 10.7759/cureus.22203. eCollection 2022 Feb.
Comparison of Chest Computed Tomography Between the Two Waves of Coronavirus Disease 2019 in Belgium Using Artificial Intelligence
Federico De Lucia[SUP] 1 [/SUP], Rahim Amer Ouali[SUP] 2 1 [/SUP], Arnaud Devriendt[SUP] 1 [/SUP], Said Sanoussi[SUP] 1 [/SUP], Mieke Cannie[SUP] 1 3 [/SUP]
Affiliations
Abstract
Background In this study, we aimed to compare two outbreaks of coronavirus disease 2019 (COVID-19) in Belgium in tomographic and biological-clinical aspects with artificial intelligence (AI). Methodology We performed an observational retrospective study. Adult patients who were symptomatic in the first seven days with COVID-19 infection, diagnosed by chest computed tomography (CT) and/or reverse transcription-polymerase chain reaction, were included in this study. The first wave of the pandemic lasted from March 25, 2020, to May 25, 2020, and the second wave lasted from October 7, 2020, to December 7, 2020. For each wave, two subgroups were defined depending on whether respiratory failure occurred during the course of the disease. The quantitative estimation of COVID-19 lung lesions was performed by AI, radiologists, and radiology residents. The chest CT severity score was calculated by AI. Results In the 202 patients included in this study, we found statistically significant differences for obesity, hypertension, and asthma. The differences were predominant in the second wave. Moreover, a mixed distribution (central and peripherical) of pulmonary lesions was noted in the second wave, but no differences were noted regarding mortality, respiratory failure, complications, and other radiological and biological elements. Chest CT severity score was among the risk factors of mortality and respiratory failure. There was a mild agreement between AI and visual evaluation of pulmonary lesion extension (K = 0.4). Conclusions Between March and December 2020, in our cohort, for the majority of the parameters analyzed, we did not record significant changes between the two waves. AI can reduce the experience and performance gap of radiologists and better establish a hospitalization criterion.
Keywords: artificial intelligence in radiology; chest computed tomography; chest ct; coronavirus disease-19 (covid-19); pandemic.
. 2022 Feb 14;14(2):e22203.
doi: 10.7759/cureus.22203. eCollection 2022 Feb.
Comparison of Chest Computed Tomography Between the Two Waves of Coronavirus Disease 2019 in Belgium Using Artificial Intelligence
Federico De Lucia[SUP] 1 [/SUP], Rahim Amer Ouali[SUP] 2 1 [/SUP], Arnaud Devriendt[SUP] 1 [/SUP], Said Sanoussi[SUP] 1 [/SUP], Mieke Cannie[SUP] 1 3 [/SUP]
Affiliations
- PMID: 35308674
- PMCID: PMC8926029
- DOI: 10.7759/cureus.22203
Abstract
Background In this study, we aimed to compare two outbreaks of coronavirus disease 2019 (COVID-19) in Belgium in tomographic and biological-clinical aspects with artificial intelligence (AI). Methodology We performed an observational retrospective study. Adult patients who were symptomatic in the first seven days with COVID-19 infection, diagnosed by chest computed tomography (CT) and/or reverse transcription-polymerase chain reaction, were included in this study. The first wave of the pandemic lasted from March 25, 2020, to May 25, 2020, and the second wave lasted from October 7, 2020, to December 7, 2020. For each wave, two subgroups were defined depending on whether respiratory failure occurred during the course of the disease. The quantitative estimation of COVID-19 lung lesions was performed by AI, radiologists, and radiology residents. The chest CT severity score was calculated by AI. Results In the 202 patients included in this study, we found statistically significant differences for obesity, hypertension, and asthma. The differences were predominant in the second wave. Moreover, a mixed distribution (central and peripherical) of pulmonary lesions was noted in the second wave, but no differences were noted regarding mortality, respiratory failure, complications, and other radiological and biological elements. Chest CT severity score was among the risk factors of mortality and respiratory failure. There was a mild agreement between AI and visual evaluation of pulmonary lesion extension (K = 0.4). Conclusions Between March and December 2020, in our cohort, for the majority of the parameters analyzed, we did not record significant changes between the two waves. AI can reduce the experience and performance gap of radiologists and better establish a hospitalization criterion.
Keywords: artificial intelligence in radiology; chest computed tomography; chest ct; coronavirus disease-19 (covid-19); pandemic.