6 March 2018 Survival prediction of squamous cell head and neck cancer patients based on radiomic features selected from lung cancer patients using artificial neural network
Author Affiliations +
Abstract
The goal of this study was to investigate the survival prediction of squamous cell head and neck cancer (SCHNC) patients by using radiomic features that were selected using an artificial neural network (ANN). We employed computed tomography (CT) images of 86 squamous cell lung cancer (SCLC) patients for the feature selection, and 30 SCHNC patients for a test of the selected features. 486 radiomic features, i.e., statistic, texture, wavelet-based features, were extracted from the tumor regions in the CT images. The ANN was constructed for selecting 10 features that could classify the SCLC patients into shorter and longer survival groups than 2 years. The features were selected based on weights with strong links between the features and predicted survival in ANN. The survival times of the SCHNC patients, who were divided into two groups with respect to the median of each of the top 10 ranked features, were estimated using a Kaplan-Meier method. The statistical significant differences between survival curves of the two groups were assessed for the 10 features using a log-rank test. The homogeneity feature of the wavelet-based HHL image (HHL_Homogeneity) demonstrated a statistically significant difference (p < 0.01) between the two groups of SCHNC, but the other 9 features did not. Our results suggest that the 2-year survival of the SCHNC patients could be predicted by using at least the radiomic feature selected among the features for SCLC patients using the ANN-based feature selection approach.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
H. Kamezawa, H. Kamezawa, H. Arimura, H. Arimura, M. Soufi, M. Soufi, } "Survival prediction of squamous cell head and neck cancer patients based on radiomic features selected from lung cancer patients using artificial neural network", Proc. SPIE 10579, Medical Imaging 2018: Imaging Informatics for Healthcare, Research, and Applications, 1057918 (6 March 2018); doi: 10.1117/12.2293415; https://doi.org/10.1117/12.2293415
PROCEEDINGS
6 PAGES


SHARE
Back to Top