Automatización del diagnóstico de cáncer de próstata mediante aprendizaje supervisado

Abstract

The rigorous, invasive, uncomfortable and ineffective that the current procedures for diagnosing prostate cancer on men can be, prevents this important process from being fully efficient, exemplifying diagnoses such as PSA, which have a low degree of accuracy, and DRE, which is highly dependent on the experience and sensitivity of the doctor who performs it. As a result, we present the project of a web information system by which, using magnetic resonance imaging together with neural networks, diagnoses the degree of cancer that a patient has on his prostate. To this end, we implemented the concept of automated training of artificial intelligence training training training the neural networks so that they were able to predict where the prostate is the image and decide on the Diagnostic, showing the Gleason grade result so that the physician can make decisions about the patient.