Peripheral Zone Prostate Cancer Localization by Multiparametric Magnetic Resonance at 3 T

• DOI: 10.1097/rli.0b013e318263f0fd
• Publication Date: 2012-09-25T18:15:22Z
• Authors: Kirsten M. Selnæs; Arend Heerschap; Line R. Jensen; May-Britt Tessem; Gregor Jarosch-Vo...; Pål E. Goa; Trond Viset; Anders Angelsen; Ingrid S. Gribbestad
• Keywords: Cut-off; Histopathology; Flip angle

Objectives The aim of this study was to assess the diagnostic accuracy peripheral zone prostate cancer localization by multiparametric magnetic resonance (MR) at 3 T using segmental matching histopathology and MR images avoid bias image features in selection noncancer regions. Materials Methods Forty-eight patients underwent imaging (MRI) on a system phased array body coil spine elements for signal detection before prostatectomy. examination included T2-weighted (T2WI), diffusion-weighted (DWI), dynamic contrast-enhanced (DCE-MRI), spectroscopic (MRSI). Histopathology slides were correlated T2W stringent procedure performed define areas without influence appearance. Mean intensity, apparent diffusion coefficient, area under enhancement curve, choline + creatine-to-citrate ratio calculated areas. Receiver operating characteristic (ROC) analysis MR-derived parameters from selected Logistic regression used create models based best combination parameters. A simplified approach assigning parametric score each segment cutoff values ROC also explored. Results By procedure, 138 41 segments transferred other methods. significant difference between mean observed all analyzed (P < 0.001). Apparent coefficient performing single parameter, with an curve Az,DWI 0.90 detection. Any T2WI, DWI, DCE-MRI significantly better than T2WI alone separating (Az,T2WI DWI = 0.94, Az,T2WI 0.92, 0.91, 0.85). MRSI (Az, 0.90); however, logistic including did not have combining gave similar results (Az 0.95 0.97, respectively). Conclusion selecting defined contrast, objectively reveals that investigated ability separate individually any is alone.