No abstract

Background: We aim to analyze the ability to detect epithelial growth factor receptor (EGFR) mutations on chest CT images of patients with lung adenocarcinoma using radiomics and/or multi-level residual convolutionary neural networks (MCNNs). Methods: We retrospectively collected 1,010 consecutive patients in Shanghai Chest Hospital from 2013 to 2017, among which 510 patients were EGFR-mutated and 500 patients were wild-type. The patients were randomly divided into a training set (810 patients) and a validation set (200 patients) according to a balanced distribution of clinical features. The CT images and the corresponding EGFR status measured by Amplification Refractory Mutation System (ARMS) method of the patients in the training set were utilized to construct both a radiomics-based model (M Radiomics) and MCNNs-based model (M MCNNs). The M Radiomics and M MCNNs were combined to build the Model Radiomics+MCNNs (M Radiomics+MCNNs). Clinical data of gender and smoking history constructed the clinical features-based model (M Clinical). M Clinical was then added into M Radiomics , M MCNNs , and M Radiomics+MCNNs to establish the Model Radiomics+Clinical (M Radiomics+Clinical), the Model MCNNs+Clinical (M MCNNs+Clinical) and the Model Radiomics+MCNNs+Clinical (M Radiomics+MCNNs+Clinical). All the seven models were tested in the validation set to ascertain whether they were competent to detect EGFR mutations. The detection efficiency of each model was also compared in terms of area under the curve (AUC), sensitivity and specificity. Results: The AUC of the M Radiomics , M MCNNs and M Radiomics+MCNNs to predict EGFR mutations was 0.740, 0.810 and 0.811 respectively. The performance of M MCNNs was better than that of M Radiomics (P=0.0225). The addition of clinical features did not improve the AUC of the M Radiomics (P=0.623), the M MCNNs (P=0.114) and the M Radiomics+MCNNs (P=0.058). The M Radiomics+MCNNs+Clinical demonstrated the highest AUC value of 0.834. The M MCNNs did not demonstrate any inferiority when compared with the M Radiomics+MCNNs (P=0.742) and the M Radiomics+MCNNs+Clinical (P=0.056). Conclusions: Both of the M Radiomics and the M CNNs could predict EGFR mutations on CT images of patients with lung adenocarcinoma. The M MCNNs outperformed the M Radiomics in the detection of EGFR mutations. The combination of these two models, even added with clinical features, is not significantly more efficient than M MCNNs alone.

The CNN showed some diagnostic power and its performance could be further improved by increasing the training set, optimizing the network structure and training strategy. Medical image based CNN has the potential to reflect spatial heterogeneity.