Purpose: Discriminating the invasiveness of early lung adenocarcinomas clinically important, and several institutions have developed their own approaches by using computed tomography images of patients. The purpose of this study is to evaluate Computer-Aided Analysis of Risk Yield (CANARY), a validated virtual biopsy and risk-stratification machine-learning tool for lung adenocarcinomas, in a Korean patient population.
Methods: A total of 380 lesions were reviewed from 360 patients who underwent lung resection between January 2018 and July 2020. The lesions were classified as "indolent” (atypical adenomatous hyperplasia, adenocarcinomas in situ, or minimally invasive adenocarcinoma) or "invasive” (invasive adenocarcinoma) by following the Score Indicative of Lung Cancer Aggression (SILA) from the CANARY analysis results, and compared with the pathology report.
Results: The mean age of the patients was 63 ± 9.8 years, and 219 (60.9%) were female. Each of the area under the curve (AUC) from the entire cohort and a dataset upsampled for an even distribution of two cases were 0.814 and 0.809, while the CANARY developer group announced the AUC of 0.912 with the cut-off SILA of 0.338. At the optimal threshold from our data, which was SILA of 0.229, CANARY found 47 out of 65 (72.3%) indolent cases and 242 out of 315 (76.8%) invasive cases.
Conclusion: The AUCs of CANARY in Korean patients were considered excellent even though there was a meaningful difference in the optimal thresholds from the developer group. We found that adjusting the cut-off threshold is needed to use CANARY in other countries or races.