Radiomics is defined as the use of automated or semi-automated post-processing and analysis of multiple features derived from imaging exams. Extracted features might generate models able to predict the molecular profile of solid tumors. The aim of this study was to develop a predictive algorithm to define the mutational status of EGFR in treatment-naïve patients with advanced non–small cell lung cancer (NSCLC). CT scans from 109 treatment-naïve patients with NSCLC (21 EGFR-mutant and 88 EGFR-wild type) underwent radiomics analysis to develop a machine learning model able to recognize EGFR-mutant from EGFR-WT patients via CT scans. A “test–retest” approach was used to identify stable radiomics features. The accuracy of the model was tested on an external validation set from another institution and on a dataset from the Cancer Imaging Archive (TCIA). The machine learning model that considered both radiomic and clinical features (gender and smoking status) reached a diagnostic accuracy of 88.1% in our dataset with an AUC at the ROC curve of 0.85, whereas the accuracy values in the datasets from TCIA and the external institution were 76.6% and 83.3%, respectively. Furthermore, 17 distinct radiomics features detected at baseline CT scan were associated with subsequent development of T790M during treatment with an EGFR inhibitor. In conclusion, our machine learning model was able to identify EGFR-mutant patients in multiple validation sets with globally good accuracy, especially after data optimization. More comprehensive training sets might result in further improvement of radiomics-based algorithms. Significance: These findings demonstrate that data normalization and “test–retest” methods might improve the performance of machine learning models on radiomics images and increase their reliability when used on external validation datasets.

Objective The first-line therapeutic approach for oral cavity squamous cell carcinoma (OCSCC) is complete surgical resection. Preoperative assessment of depth of invasion (cDOI) is crucial to plan the surgery. Magnetic resonance (MR) and intraoral ultrasonography (IOUS) have been shown to be useful tools for assessment of DOI. The present analysis investigates the accuracy of MR and IOUS in evaluating DOI in OCSCC compared to histological evaluation (pDOI). Materials and methods Forty-nine previously untreated patients with cT1-T3 OCSCC were reviewed. Nine patients were staged with MR alone, 10 with IOUS alone, and 30 with both MR and IOUS. Results Mean difference between cDOIMR and pDOI values of 0.2 mm (95% CI − 1.0–1.3 mm) and between cDOIIOUS and pDOI of 0.3 mm (95% CI − 1.0–1.6 mm). Spearman R between cDOIMR and pDOI was R = 0.83 and between cDOIIOUS and pDOI was R = 0.76. Both radiological techniques showed high performance for the correct identification, with the optimum cut-off of 5 mm, of patients with a pDOI ≥ 4 mm and amenable to a neck dissection, with an AUC of 0.92 and 0.82 for MR and IOUS, respectively. Conclusion Both examinations were valid approaches for preoperative determination of DOI in OCSCC, although with different cost-effectiveness profiles and indications.

Delta-radiomics is a branch of radiomics in which features are confronted after time or after introducing an external factor (such as treatment with chemotherapy or radiotherapy) to extrapolate prognostic data or to monitor a certain condition. Immune checkpoint inhibitors (ICIs) are currently revolutionizing the treatment of non-small cell lung cancer (NSCLC); however, there are still many issues in defining the response to therapy. Contrast-enhanced CT scans of 33 NSCLC patients treated with ICIs were analyzed; altogether, 43 lung lesions were considered. The radiomic features of the lung lesions were extracted from CT scans at baseline and at first reassessment, and their variation (delta, Δ) was calculated by means of the absolute difference and relative reduction. This variation was related to the final response of each lesion to evaluate the predictive ability of the variation itself. Twenty-seven delta features have been identified that are able to discriminate radiologic response to ICIs with statistically significant accuracy. Furthermore, the variation of nine features significantly correlates with pseudo-progression.

• Pectoralis muscle area can be estimated on breast MRI • Total psoas area on CT and pectoralis muscle area on MRI are strongly correlated • Pectoralis muscle area on breast MRI could estimate the skeletal muscle mass.