Search

Background: To establish a predictive model based on multisequence magnetic resonance imaging (MRI) using deep learning to identify wild-type (WT) epidermal growth factor receptor (EGFR), EGFR exon 19 deletion (19Del), and EGFR exon 21-point mutation (21L858R) simultaneously. Methods: A total of 399 patients with proven brain metastases of non-small cell lung cancer (NSCLC) were retrospectively enrolled and divided into training (n = 306) and testing (n = 93) cohorts separately based on two timepoints. All patients underwent 3.0-T brain MRI including T2-weighted, T2-weighted fluid-attenuated inversion recovery, diffusion-weighted imaging, and contrast-enhanced T1-weighted sequences. Radiomics features were extracted from each lesion based on four sequences. An algorithm combining radiomics approach with graph convolutional networks architecture (Radio-GCN) was designed for the prediction of EGFR mutation status and subtype. The area under the curve (AUC) at receiver operating characteristic analysis was used to evaluate the predication capabilities of each model. Results: We extracted 1,290 radiomics features from each MRI sequence. The AUCs of the Radio-GCN model for identifying EGFR 19Del, 21L858R, and WT for the lesion-wise analysis were 0.996 ± 0.004, 0.971 ± 0.013, and 1.000 ± 0.000 on the independent testing cohort separately. It also yielded AUCs of 1.000 ± 0.000, 0.991 ± 0.009, and 1.000 ± 0.000 for predicting EGFR mutations respectively for the patient-wise analysis. The κ coefficients were 0.735 and 0.812, respectively. Conclusions: The constructed Radio-GCN model is a new potential tool to predict the EGFR mutation status and subtype in NSCLC patients with brain metastases. Relevance statement: The study demonstrated that a deep learning approach based on multisequence MRI can help to predict the EGFR mutation status in NSCLC patients with brain metastases, which is beneficial to guide a personalized treatment. Key points: • This is the first study to predict the EGFR mutation subtype simultaneously. • The Radio-GCN model holds the potential to be used as a diagnostic tool. • This study provides an imaging surrogate for identifying the EGFR mutation subtype. [ABSTRACT FROM AUTHOR]

The frozen section (FS) diagnoses of pathology experts are used in China to determine whether sentinel lymph nodes of breast cancer have metastasis during operation. Direct implementation of a deep neural network (DNN) in clinical practice may be hindered by misdiagnosis of the algorithm, which affects a patient's treatment decision. In this study, we first obtained the prediction result of the commonly used patch-DNN, then we present a relative risk classification and regression tree (RRCART) to identify the misdiagnosed whole-slide images (WSIs) and recommend them to be reviewed by pathologists. Applying this framework to 2362 WSIs of breast cancer lymph node metastasis, test on frozen section results in the mean area under the curve (AUC) reached 0.9851. However, the mean misdiagnosis rate (0.0248), was significantly higher than the pathologists' misdiagnosis rate (p < 0.01). The RRCART distinguished more than 80% of the WSIs as a high-accuracy group with an average accuracy reached to 0.995, but the difference with the pathologists' performance was not significant (p > 0.01). However, the other low-accuracy group included most of the misdiagnoses of DNN models. Our research shows that the misdiagnosis from deep learning model can be further enriched by our method, and that the low-accuracy WSIs must be selected for pathologists to review and the high-accuracy ones may be ready for pathologists to give diagnostic reports. [ABSTRACT FROM AUTHOR]