Deep Learning for Predicting Subtype Classification and Survival of Lung Adenocarcinoma on Computed Tomography

Abstract

Background: The subtype classification of lung adenocarcinoma is of paramount importance for preoperative adjuvant therapy. The precise evaluation in histopathologic classification of lung adenocarcinoma depends on full nodule lesion resection. This study aimed to investigate the deep learning and radiomics networks for predicting histologic subtype classification and survival of lung adenocarcinoma diagnosed with small biopsy specimen through computed tomography (CT) images. Methods: A dataset of 1222 lung adenocarcinoma patients including the clinicopathological data were retrospectively enrolled from three institutions. The anonymised preoperative CT images and pathological labels of atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), invasive adenocarcinoma (IAC) with predominant components classified into lepidic (pLAC), acinar(pAAC), papillary(pPAC), micropapillary(pMAC), and solid (pSAC) ones were obtained. These pathological labels were divided into 2-category classification (IAC; non-IAC), 3-category (Non-MAC/SAC, Non-pMAC/pSAC and pMAC/pSAC) and 8-category as aforementioned. We modeled the classification task of histological subtypes based on modified ResNet-34 deep learning network, radiomics strategies and deep radiomics combined algorithm. Then we established the prognostic models in lung adenocarcinoma patients with survival outcomes. The accuracy (ACC), area under ROC curves (AUCs) and C-index were primarily performed to evaluate the algorithms. Findings: This study included the a training set (n=802) and two validation cohorts (internal, n=196; external, n=224). The ACC of deep radiomics algorithm in internal validation achieved 0.8776, 0.8061 in the 2-category, 3-category classification, respectively. Even in 8 classifications, the AUC ranged from 0.739 to 0.940 in internal test. Further, we constructed a prognosis model that C-index was 0.892(95% Confidence Intervals: 0.846-0.937) in internal validation set. Interpretation: Our results reveal that the automated deep radiomics based triage system has achieved the great performance in the subtype classification and survival predictability in patients with CT-detected lung adenocarcinoma nodules, providing the clinical guide for preoperative adjuvant therapy. Funding Statement: National Natural Science Foundation of China, Science and Technology Project of Chengdu, and Science and Technology Project of Sichuan Declaration of Interests: The authors declare no conflict of interest. Ethics Approval Statement: This study was approved by the institutional ethics committee of participating institutions.