Automated Detection Of Prostate Cancer from Multiparametric MRI Using Deep Convolutional Neural Networks

Abstract

One of the most prevalent cancers and the second largest cause of cancer-related deaths in males is prostate cancer (PCa).Prostate cancer age-standardized incidence rates have grown in Uganda, rising from 41.6 to 60.5 per 100,000 men. Compared to 98% of African American patients. In Africa only 46.9%[1] of patients diagnosed can live up to 5 years after diagnosis compared to 98% [2] of African American patients. This is caused by late diagnosis of victims who are found at later stages with incurable tumours.PCa can be treated if detected at its earlier stages.[3]. When a male above 50 experiences symptoms like blood in the urine and frequent urination a urologist carries out the following tests which include Prostate Specific Antigen (PSA) and Digital Rectal Examination (DRE). Because high concentrations of PSA can also be linked to false positive results which causes unnecessary anxiety to patients, it is not highly recommended. Following a positive PSA test, a DRE is performed to determine the need for a biopsy.[4]Due to the highly invasive nature of biopsies, the results from the above tests are first sent to a radiologist for Magnetic Resonance Imaging (MRI) scan for detection of cancerous tissues in the prostate before a biopsy is done. Recent advancements in imaging technology led to the emergence of multiparametric Magnetic Resonance Imaging (mpMRI) which has improved selectivity of cancerous regions by providing a more detailed picture of the prostate than an MRI scan.