No abstract

No abstract

The APS Journal Legacy Content is the corpus of 100 years of historical scientific research from the American Physiological Society research journals. This package goes back to the first issue of each of the APS journals including the American Journal of Physiology, first published in 1898. The full text scanned images of the printed pages are easily searchable. Downloads quickly in PDF format.

No abstract

Purpose: Intrahepatic failure is a major concern in the treatment of hepatocellular carcinoma (HCC) but lacks clinically reliable predictive markers. This work aims to identify a radiomic signature for risk prediction of intrahepatic progression in HCC patients treated with stereotactic body radiation therapy (SBRT) by extracting magnetic resonance imaging (MRI) and contrast-enhanced CT phenotypes from tumor(s) and liver parenchyma. Materials/Methods: 94 HCC patients treated with SBRT were retrospectively analyzed. The range of follow-up time is 57-1768 days. Samples were right censored if there was no intrahepatic progression within the follow-up time. The event rates is 43/94=45.7%. 112 radiomics features were extracted from tumors and liver parenchyma from CECT and MRI using pyradiomics platform. CT images were first added a 1000 HU shift and MRI images were standardized by the mean and standard deviation. Univariate Cox analysis was applied for radiomics features analysis. Multivariate Cox models and survival neural networks were also used to predict the outcome. Multiple folds cross-validation was used for model evaluation. Models’ performances were assessed using Harrell’s c-index and Kaplan-Meier plot. Results: Sphericity, small dependence emphasis, small dependence high gray level emphasis, run high gray level emphasis and coarseness are significant predictors from MR tumors for the outcome. The Cox models showed c-index of 0.600 (CI 95%: 0.558-0.642). Sphericity and maximal correlation coefficient are significant predictors from CT tumors for the outcome, with a c-index of 0.613 (CI 95%: 0.576-0.649). Volume, elongation and sphericity are significant predictors for overall survival, with a c-index of 0.584 (CI 95%: 0.532-0.636). None of the liver parenchyma features were important predictors. Interestingly, convolutional neural networks (CNN) were applied on the contour of the tumors and the c-index was 0.668 (CI 95%: 0.612-0.723). The 2 channel (MR-tumor region and contour mask) CNN network showed better performance of 0.698 (CI 95%: 0.658-0.749). Conclusions: Shape features showed predictive power for both intrahepatic progression-free and overall survival endpoints, which was proven using both radiomics feature based Cox modeling and CNN based contour methods. Adding MR imaging improved the performance. Citation Format: Lise Wei, Jue Jiang, Harini Veeraraghavan, Issam El Naqa. Shape features predicting intrahepatic progression-free and overall survival for SBRT treated HCC patients using radiomics and deep learning based survival models [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-041.