Evaluation of Rapid T2-weighted and DWI MR Sequences Reconstructed by Deep Learning for Prostate Imaging (DLRPRO)

Evaluation of Rapid T2-weighted and DWI MR Sequences Reconstructed by Deep Learning for Prostate Imaging

MR prostate exam is essential for the diagnosis, workup and follow-up of prostate cancer. It allows to detect subclinical prostate cancer following an increase in the level of PSA. The investigators can score the lesion according to the PIRADS classification and obtain an estimate of lesion malignancy. To perform this classification, T2 and DWI sequences are essential. Detection and characterization of malignant lesion is important to address appropriate patient care pathway. The purpose of this project is to evaluate novel deep learning (DL) T2-weighted TSE (T2DL) and Diffusion (DWIDL) sequences for prostate MR exam and investigate its impact on diagnostic, examination time, image quality, and PI-RADS classification compared to standard T2-weighted TSE (T2S) and standard Diffusion (DWIS) sequences.

Subjects will lie in supine position. The systematic use of a headset will reduce the acoustic noise inherent to the machine. We are going to carry out the standard MR prostate protocol which patients usually benefit from in clinical routine. This protocol consists of morphological sequences (T2 weighting with spin echo readout), Diffusion MR and dynamic contrast-enhanced sequences. We will then perform an additional faster enhanced T2-weighting SE and DWI sequences combined with Deep Learning reconstruction

Inclusion Criteria: Age ≥ 18 ans Healthy subject without history of hepatic disease Patient addressed for an prostate MRI Ability to give consent Exclusion Criteria: claustrophobia, major obesity (>140 kg), Patient under guardianship or curators Age < 18 years, Women, History of prostatectomy or irradiation of the prostate any contraindication to MRI exam