Background: To establish a radiomic approach to identify epidermal growth factor receptor (EGFR) mutation status in lung adenocarcinoma patients based on CT images, and to distinguish exon-19 deletion and exon-21 L858R mutation.Methods: Two hundred sixty-three patients who underwent pre-surgical contrast-enhanced CT and molecular testing were included, and randomly divided into the training (80%) and test (20%) cohort.Tumor images were three-dimensionally segmented to extract 1,672 radiomic features. Clinical features (age, gender, and smoking history) were added to build classification models together with radiomic features.Subsequently, the top-10 most relevant features were used to establish classifiers. For the classifying tasks including EGFR mutation, exon-19 deletion, and exon-21 L858R mutation, four logistic regression models were established for each task. Results:The training and test cohort consisted of 210 and 53 patients, respectively. Among the established models, the highest accuracy and sensitivity among the four models were 75.5% (61.7-86.2%) and 92.9% (76.5-99.1%) to classify EGFR mutation, respectively. The highest specificity values were 86.7% (69.3-96.2%) and 70.4% (49.8-86.3%) to classify exon-19 deletion and exon-21 L858R mutation, respectively.Conclusions: CT radiomics can sensitively identify the presence of EGFR mutation, and increase the certainty of distinguishing exon-19 deletion and exon-21 L858R mutation in lung adenocarcinoma patients.CT radiomics may become a helpful non-invasive biomarker to select EGFR mutation patients for invasive sampling.

Purpose: To develop a machine learning-derived radiomics approach to simultaneously discriminate epidermal growth factor receptor (EGFR), Kirsten rat sarcoma viral oncogene (KRAS), Erb-B2 receptor tyrosine kinase 2 (ERBB2), and tumor protein 53 (TP53) genetic mutations in patients with non-small cell lung cancer (NSCLC). Methods: This study included consecutive patients from April 2018 to June 2020 who had histologically confirmed NSCLC, and underwent pre-surgical contrast-enhanced CT and post-surgical next-generation sequencing (NGS) tests to determine the presence of EGFR, KRAS, ERBB2, and TP53 mutations. A dedicated radiomics analysis package extracted 1672 radiomic features in three dimensions. Discriminative models were established using the least absolute shrinkage and selection operator to determine the presence of EGFR, KRAS, ERBB2, and TP53 mutations, based on radiomic features and relevant clinical factors. Results: In 134 patients (63.6 ± 8.9 years), the 20 most relevant radiomic features (13 for KRAS) to mutations were selected to construct models. The areas under the curve (AUCs) of the combined model (radiomic features and relevant clinical factors) for discriminating EGFR, KRAS, ERBB2, and TP53 mutations were 0.78 (95% CI: 0.70–0.86), 0.81 (0.69–0.93), 0.87 (0.78–0.95), and 0.84 (0.78–0.91), respectively. In particular, the specificity to exclude EGFR mutations was 0.96 (0.87–0.99). The sensitivity to determine KRAS, ERBB2, and TP53 mutations ranged from 0.82 (0.69–90) to 0.92 (0.62–0.99). Conclusions: Machine learning-derived 3D radiomics can simultaneously discriminate the presence of EGFR, KRAS, ERBB2, and TP53 mutations in patients with NSCLC. This noninvasive and low-cost approach may be helpful in screening patients before invasive sampling and NGS testing.

No abstract

PurposeThis study aimed to evaluate the feasibility of differentiating the atrial fibrillation (AF) subtype and preliminary explore the prognostic value of AF recurrence after ablation using radiomics models based on epicardial adipose tissue around the left atrium (LA-EAT) of cardiac CT images.MethodThe cardiac CT images of 314 patients were collected wherein 251 and 63 cases were randomly enrolled in the training and validation cohorts, respectively. Mutual information and the random forest algorithm were used to screen for the radiomic features and construct the radiomics signature. Radiomics models reflecting the features of LA-EAT were built to differentiate the AF subtype, and the multivariable logistic regression model was adopted to integrate the radiomics signature and volume information. The same methodology and algorithm were applied to the radiomic features to explore the ability for predicting AF recurrence.ResultsThe predictive model constructed by integrating the radiomic features and volume information using a radiomics nomogram showed the best ability in differentiating AF subtype in the training [AUC, 0.915; 95% confidence interval (CI), 0.880–0.951] and validation (AUC, 0.853; 95% CI, 0.755–0.951) cohorts. The radiomic features have shown convincible predictive ability of AF recurrence in both training (AUC, 0.808; 95% CI, 0.750–0.866) and validation (AUC, 0.793; 95% CI, 0.654–0.931) cohorts.ConclusionsThe LA-EAT radiomic signatures are a promising tool in the differentiation of AF subtype and prediction of AF recurrence, which may have clinical implications in the early diagnosis of AF subtype and disease management.

ObjectiveTo develop and validate a dual-energy computed tomography (DECT) derived radiomics model to predict peritoneal metastasis (PM) in patients with gastric cancer (GC).MethodsThis retrospective study recruited 239 GC (non-PM = 174, PM = 65) patients with histopathological confirmation for peritoneal status from January 2015 to December 2019. All patients were randomly divided into a training cohort (n = 160) and a testing cohort (n = 79). Standardized iodine-uptake (IU) images and 120-kV-equivalent mixed images (simulating conventional CT images) from portal-venous and delayed phases were used for analysis. Two regions of interest (ROIs) including the peritoneal area and the primary tumor were independently delineated. Subsequently, 1691 and 1226 radiomics features were extracted from the peritoneal area and the primary tumor from IU and mixed images on each phase. Boruta and Spearman correlation analysis were used for feature selection. Three radiomics models were established, including the R_IU model for IU images, the R_MIX model for mixed images and the combined radiomics model (the R_comb model). Random forest was used to tune the optimal radiomics model. The performance of the clinical model and human experts to assess PM was also recorded.ResultsFourteen and three radiomics features with low redundancy and high importance were extracted from the IU and mixed images, respectively. The R_IU model showed significantly better performance to predict PM than the R_MIX model in the training cohort (AUC, 0.981 vs. 0.917, p = 0.034). No improvement was observed in the R_comb model (AUC = 0.967). The R_IU model was the optimal radiomics model which showed no overfitting in the testing cohort (AUC = 0.967, p = 0.528). The R_IU model demonstrated significantly higher predictive value on peritoneal status than the clinical model and human experts in the testing cohort (AUC, 0.785, p = 0.005; AUC, 0.732, p <0.001, respectively).ConclusionDECT derived radiomics could serve as a non-invasive and easy-to-use biomarker to preoperatively predict PM for GC, providing opportunity for those patients to tailor appropriate treatment.