Prostate Cancer Detection in Multiparametric MRI Using Random Forest with Instance Weighting and MR Prostate Segmentation by Deep Learning Using Holistic Nested Networks

摘要

The disclosed computer-aided diagnosis of prostate (CAD) system uses a random forest classifier to detect prostate cancer. The system classifies individual pixels inside the prostate as potential cancer sites using a combination of spatial, intensity, and texture features extracted from the three sequences. Random forest training considers instance-level weighting for small and large cancerous lesions, as well as for equivalent treatment of small and large prostate backgrounds. Two other approaches are based on the AutoContext pipeline, which aims to better use sequence-specific patterns. Methods and systems for accurate automatic segmentation of the prostate in MRI are also disclosed.