A Machine Learning based Preoperatively Grading Rectal Method with Continue Time Random Walk DWI Model.

Machine learning offers a principled approach for developing automatic algorithms for analysis of high-dimensional biomedical data. The continuous-time random-walk model (CTRW) is novel non-Gaussian diffusion model that provides promising evidence indicating a possible link between voxel-level spatiotemporal diffusion heterogeneity and microscopic intravoxel tissue heterogeneity giving the model advantages in diagnosing many diseases1. In this study, we apply a machine-learning algorithm, the principal component analysis (PCA), for quantitative automatic diagnosis to grade rectal cancer using parameters from CTRW model. Our study shows that PCA has the potential to grade rectal cancer with higher accuracy than the original parameters.