Application value of artificial intelligence assessment of pulmonary nodule risk in assisting manual film reading
Objective To investigate the clinical utility of artificial intelligence(AI)utilizing CT images for screening and evaluating the risk of pulmonary nodules.Methods A total of 100 patients with pulmonary nodules who underwent surgery or puncture and pathological examination at Gezhouba Central Hospital of Sinopharm from January 2022 to December 2022 were selected.AI and physicians performed nodule diagnoses independently and compared the sensitivity,specificity,positive predictive value,negative predictive value,and accuracy of groups A(manual diagnosis,n=98),B(AI diagnosis,n=99),and C(manual+AI joint diagnosis,n=99).The value of AI-assisted physician prediction of benign and malignant was analyzed using the pathological results as the"gold standard".Results There were no significant differences in the risk of pulmonary nodules between groups A and B(P>0.05).Group C was superior to group A and B in terms of sensitivity(98.75%),negative predictive value(91.67%),and accuracy(90.00%).group C could significantly improve diagnostic efficiency,better than groups A and B in reading films alone,and the diagnosis was more consistent with pathological results(Kappa value=0.632,P<0.01).Conclusion AI-based screening and grading of pulmonary nodules can greatly enhance sensitivity and accuracy,decrease missed diagnosis rates,and provide substantial clinical value in assisting radiologists in image interpretation.AI-assisted risk assessment of pulmonary nodules also holds significant reference and clinical application value in the pre-judgment of benign and malignant nodules.
artificial intelligence diagnosticslow-dose spiral CTcomputer-assistedlung nodulebenign and malignant
万方数据
维普
