Radio-pathomic Maps of Epithelium and Lumen Density Predict the Location of High-Grade Prostate Cancer

• DOI: 10.1016/j.ijrobp.2018.04.044
• Publication Date: 2018-04-24T16:37:56Z
• Authors: Sean D. McGarry; Sarah L. Hurrell; Kenneth A. Iczkowski; William Hall; Amy L. Kaczmarowski; Anjishnu Banerjee; Tucker Keuter; Kenneth Jacobsohn; John D. Bukowy; Marja T. Nevalainen; Mark D. Hohenwalter; William A. See; Peter S. LaViolette
• Keywords: Male; 610; Contrast Media; Article; Epithelium; Machine Learning; 03 medical and health sciences; Computer-Assisted; 0302 clinical medicine; Image Interpretation, Computer-Assisted; Humans; False Positive Reactions; Prospective Studies; Least-Squares Analysis; Image Interpretation; Aged; Neoplasm Staging; Prostatectomy; Radiotherapy; Prostate; Reproducibility of Results; Prostatic Neoplasms; Middle Aged; Prostate-Specific Antigen; Magnetic Resonance Imaging; 3. Good health; ROC Curve; Three-Dimensional; Printing, Three-Dimensional; Printing; Regression Analysis; Learning Curve

This study aims to combine multiparametric magnetic resonance imaging (MRI) and digitized pathology with machine learning to generate predictive maps of histologic features for prostate cancer localization.Thirty-nine patients underwent MRI prior to prostatectomy. After surgery, tissue was sliced according to MRI orientation using patient-specific 3-dimensionally printed slicing jigs. Whole-mount sections were annotated by our pathologist and digitally contoured to differentiate the lumen and epithelium. Slides were co-registered to the T2-weighted MRI scan. A learning curve was generated to determine the number of patients required for a stable machine-learning model. Patients were randomly stratified into 2 training sets and 1 test set. Two partial least-squares regression models were trained, each capable of predicting lumen and epithelium density. Predicted density values were calculated for each patient in the test dataset, mapped into the MRI space, and compared between regions confirmed as high-grade prostate cancer.The learning-curve analysis showed that a stable fit was achieved with data from 10 patients. Maps indicated that regions of increased epithelium and decreased lumen density, generated from each independent model, corresponded with pathologist-annotated regions of high-grade cancer.We present a radio-pathomic approach to mapping prostate cancer. We find that the maps are useful for highlighting high-grade tumors. This technique may be relevant for dose-painting strategies in prostate radiation therapy.