Prostate cancer is the 2nd most common malignancy in the world as well as the leading cancer in male population in Taiwan. The treatment selections of prostate cancer are limited by the uncertainty of its aggressiveness (i.e.: histological graded) and staging before treatment. Although prostate mpMRI has much better ability for detection and localization of prostate cancers than other imaging modalities and diagnostic tests, there is still gap for risk stratifications and treatment selection based on prostate mpMRI findings. Thus, a robust radiomics prediction models based on imaging biomarkers on prostate mpMRI with high prediction accuracy could fill the gap of misclassification of risk stratifications of prostate cancers, guides treatment selections and providing monitoring schedules for treated patients as well as early timely additional treatments (i.e.: target therapy or immunotherapy) for patients with high risk of early recurrence. Furthermore, radiomics could provide consistent information which help in decreasing interobserver and intra-observer variability of interpretating prostate cancer even in the use of PIRADS. In this way, this would save the fee of inappropriate or ineffective treatment and avoid unnecessary time and cost of monitoring low risk patients as well as improve patients' survivals and possibly life-quality as well.
