Privacy-preserving distributed learning of radiomics to predict overall survival and HPV status in head and neck cancer.

A major challenge in radiomics is assembling data from multiple centers. Sharing data between hospitals is restricted by legal and ethical regulations. Distributed learning is a technique, enabling training models on multicenter data without data leaving the hospitals ("privacy-preserving" distributed learning). This study tested feasibility of distributed learning of radiomics data for prediction of two year overall survival and HPV status in head and neck cancer (HNC) patients. Pretreatment CT images were collected from 1174 HNC patients in 6 different cohorts. 981 radiomic features were extracted using Z-Rad software implementation. Hierarchical clustering was performed to preselect features. Classification was done using logistic regression. In the validation dataset, the receiver operating characteristics (ROC) were compared between the models trained in the centralized and distributed manner. No difference in ROC was observed with respect to feature selection. The logistic regression coefficients were identical between the methods (absolute difference <10-7). In comparison of the full workflow (feature selection and classification), no significant difference in ROC was found between centralized and distributed models for both studied endpoints (DeLong p > 0.05). In conclusion, both feature selection and classification are feasible in a distributed manner using radiomics data, which opens new possibility for training more reliable radiomics models.

Auto-delineation of oropharyngeal clinical target volumes using 3D convolutional neural networks.

Accurate clinical target volume (CTV) delineation is essential to ensure proper tumor coverage in radiation therapy. This is a particularly difficult task for head-and-neck cancer patients where detailed knowledge of the pathways of microscopic tumor spread is necessary. This paper proposes a solution to auto-segment these volumes in oropharyngeal cancer patients using a two-channel 3D U-Net architecture. The first channel feeds the network with the patient's CT image providing anatomical context, whereas the second channel provides the network with tumor location and morphological information. Radiation therapy simulation computer tomography scans and their corresponding manually delineated CTV and gross tumor volume (GTV) delineations from 285 oropharyngeal patients previously treated at MD Anderson Cancer Center were used in this study. CTV and GTV delineations underwent rigorous group peer-review prior to the start of treatment delivery. The convolutional network's parameters were fine-tuned using a training set of 210 patients using 3-fold cross-validation. During hyper-parameter selection, we use a score based on the overlap (dice similarity coefficient (DSC)) and missed volumes (false negative dice (FND)) to minimize any possible under-treatment. Three auto-delineated models were created to estimate tight, moderate, and wide CTV margin delineations. Predictions on our test set (75 patients) resulted in auto-delineations with high overlap and close surface distance agreement (DSC > 0.75 on 96% of cases for tight and moderate auto-delineation models and 97% of cases having mean surface distance ⩽ 5.0 mm) to the ground-truth. We found that applying a 5 mm uniform margin expansion to the auto-delineated CTVs would cover at least 90% of the physician CTV volumes for a large majority of patients; however, determination of appropriate margin expansions for auto-delineated CTVs merits further investigation.

Utilization of short-course radiation therapy for patients with nonmetastatic rectal adenocarcinoma in the United States.

PURPOSE: Preoperative short-course radiation therapy (SCRT) for patients with nonmetastatic rectal adenocarcinoma has been studied in European trials, but is not often used in the United States. We aim to describe the utilization of preoperative SCRT among patients with nonmetastatic rectal cancer in the National Cancer Database and describe factors associated with its use. METHODS AND MATERIALS: The National Cancer Database was queried for patients treated with preoperative radiation therapy followed by surgery for nonmetastatic rectal adenocarcinoma between 2004 and 2014. Patient, tumor, and treatment-related characteristics were compared between patients treated with SCRT (20-25 Gy in <7 fractions) and patients treated with long-course radiation therapy (45-70 Gy in ≥ 25 fractions). Univariate and multivariate Cox regression analyses were used to evaluate factors associated with overall survival. Survival rates were compared using an inverse-probability-weighted regression adjustment method. RESULTS: A total of 42,336 patients were included for analysis of which 41,867 patients (98.9%) were treated with long-course radiation therapy and 469 patients (1.1%) with SCRT. Patients treated with SCRT were older, had more comorbidities, had earlier T-stage, and were more likely to be clinically node-negative. Patients treated with SCRT were more likely to be treated at an academic center, have Medicare insurance, and be treated without chemotherapy. Patients treated with SCRT had lower pathological complete response rates (4.3% vs 6.9%; P < .001) and higher rates of positive circumferential resection margins (8.3% vs 5.2%; P = .001). On multivariate analysis, radiation fractionation was not significantly associated with overall survival. CONCLUSIONS: SCRT is used for only approximately 1% of patients treated preoperatively for nonmetastatic rectal cancer in the United States. The results of recently completed randomized trials may further inform patterns of practice in the United States and abroad.