Automatic detection of prostate cancer grades and chronic prostatitis in biparametric MRI.

•Inclusion of prostatitis class into the training improves the detection of clinically-significant prostate cancer lesions.•One-hot encoding outperforms ordinal encoding in the prostate lesion detection context as a segmentation data format.•Proposed optimal parameter range for data simulated shift to improve robustness against intermodal image displacement.•High-quality biparametric magnetic resonance imaging (MRI) dataset with ground-truth from precise MRI-guided biopsies.