Comparison study for computer assisted detection and diagnosis ‘CAD’ systems dedicated to prostate cancer detection using MRImp modalities

摘要

Prostate cancer could be considered among the leading causes of men death [1]. Prostate cancer early detection could be hence so necessary in order to confirm the existence of a tumor. It was traditionally based on conventional methods as DRE (Digital Rectal Examination), dosing Prostatic Specific Antigen (PSA) and biopsy. However, these techniques often suffer from lack of specificity, could be invasive, or linked to non-cancerous pathologies. To overcome these difficulties, it was necessary to use imaging techniques that can estimate the position, the volume, and the tumor aggressiveness. Magnetic Resonance Imaging (MRI) has emerged as a promising technique for the prostate cancer diagnosis in terms of detection and localization thanks to an excellent tissue contrast. Multi-parametric Magnetic Resonance Imaging (MRImp) was considered as a reference for detection, localization and evaluation of tumor aggressiveness. This was therefore a promising research domain to diagnose and to quickly adapt a patient monitoring. In addition, its localization capabilities help other treatment techniques, allowing local treatment of the tumor and no longer systematically resort to total removal of the gland. Computer Assisted Detection and Diagnosis systems (CAD) are intended to assist the radiologist in making decisions. We present in this paper a comparison study between the most recent CAD systems in order to clarify usefulness as well as to propose an emerging methodology for implementation that could certainly help clinicians during their diagnosis. Such a proposition was hence forth implemented and tested as one preliminary research in this promoting domain.