Lung-PNet: An Automated Deep Learning Model for the Diagnosis of Invasive Adenocarcinoma in Pure Ground-Glass Nodules on Chest CT

BACKGROUND. Pure ground-glass nodules (pGGNs) on chest CT representing invasive adenocarcinoma (IAC) warrant lobectomy with lymph node resection. For pGGNs representing other entities, close follow-up or sublobar resection without node dissection may be appropriate. OBJECTIVE. To develop and validate an automated deep learning model for differentiation of pGGNs on chest CT representing IAC from those representing atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), or minimally invasive adenocarcinoma (MIA). METHODS. This retrospective study included 402 patients (mean age, 53.2 years; 119 men, 283 women) with a total of 448 pGGNs on noncontrast chest CT that underwent resection from January 2019 to June 2022 and were histologically diagnosed as AAH (n=29), AIS (n=83), MIA (n=235), or IAC (n=101). We developed Lung-PNet, a 3D deep learning model for automatic segmentation and classification (probability of IAC vs other entities) of pGGNs on CT. Nodules from January 2019 to December 2021 were randomly allocated to training (n=327) and internal test (n=82) sets; nodules from January 2022 to June 22 formed a holdout test set (n=39). Segmentation performance was assessed by Dice coefficients, using radiologists' manual segmentations as reference. Classification performance was assessed by AUCROC and AUC under precision-recall curve (AUCPR), and compared with that of four readers (three radiologists, one surgeon). Code is publicly available: https://github.com/Xiaodong-Zhang-PKUFH/Lung-PNet.git. RESULTS. In the holdout test set, Dice coefficients for segmentation of IACs and of other lesions were 0.860 and 0838, and AUCROC and AUCPR for classification as IAC were 0.911 and 0.842. At threshold probability of ≥ 50.0% for prediction of IAC, Lung-PNet had sensitivity, specificity, accuracy, and F1 score of 50.0%, 92.0%, 76.9%, and 60.9% in the holdout test set. Accuracy and F1 score (with p values vs Lung-PNet) were, in the holdout test set: reader 1, 51.3% (p=.02) and 48.7% (p=.008); reader 2, 79.5% (p=.75) and 75.0% (p=.10); reader 3, 66.7% (p=.35) and 68.3% (p<.001); reader 4, 71.8% (p=.48) and 42.1% (p=.18). CONCLUSION. Lung-PNet exhibited robust performance for segmenting and classifying (IAC vs other entities) pGGNs on chest CT. CLINICAL IMPACT. This automated deep learning tool may help guide selection of surgical strategies for pGNN management.