Diagnostic value of chest computed tomography scan based on artificial intelligence and deep learning in children with lobar pneumonia and analysis of image features before and after treatment: A retrospective cohort study

Background: A retrospective cohort study was conducted to analyze the diagnostic value and image features of chest computed tomography (CT) scan in children with lobar pneumonia (LP) before and after treatment. Materials and Methods: 172 children with lobar pneumonia treated from January 2016 to December 2021 were selected. The patients who underwent plain X-ray scan were divided into control group (n = 72) and the patients who underwent chest CT scan as study group (n = 100). The diagnostic value and image characteristics before and after treatment were compared between the two groups. Results: After treatment, the lesion area of the patient was absorbed in varying degrees, and the CT plain scan indicated that the solid shadow density decreased until it was completely absorbed. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value of chest X-ray were 66.67%, 58.33%, 63.89%, 76.19% and 46.67% respectively; and chest CT scan were 82.98%, 67.92%, 75.00%, 69.64% and 81.82%. The sensitivity, specificity, accuracy, and negative predictive value of chest CT plain scan were higher, and the positive predictive value was lower compared to those of chest X-ray plain film. The results of ROC curve study indicated that the AUC of chest CT plain scan was 0.755 (95%CI=0.657 -0.852), and the AUC of chest X-ray film was 0.625 (95%CI= 0.489-0.744). Conclusion: Chest CT has high sensitivity and specificity in the diagnosis of LP in children, which can clearly demonstrate the imaging features of LP before and after treatment.