Validation of PI-RADS v.2 for prostate cancer diagnosis with MRI at 3T using an external phased-array coil

• Published in: Journal of magnetic resonance imaging 2016-11, Vol.44 (5), p.1354-1359
• Main Authors: Baldisserotto, Matteo, Neto, Eurico J. Dornelles, Carvalhal, Gustavo, de Toledo, Aloyso F., de Almeida, Clovis M., Cairoli, Carlos E.D., de Silva, Daniel O., Carvalhal, Eduardo, Paganin, Ricardo P., Agra, Alexandre, de Santos, Francisco S., Noronha, Jorge A.P.
• Format: Article
• Language: English
• Subjects: Accuracy; Aged; Aged, 80 and over; Agreements; Cancer; Confidence intervals; Diagnosis; Diagnostic systems; Equipment Design; Humans; Image Interpretation, Computer-Assisted - instrumentation; Image Interpretation, Computer-Assisted - standards; Internationality; Magnetic resonance imaging; Magnetic Resonance Imaging - instrumentation; Magnetic Resonance Imaging - standards; Male; Middle Aged; Patients; Practice Guidelines as Topic; Prostate cancer; Prostatic Neoplasms - diagnostic imaging; Prostatic Neoplasms - pathology; Radiation Dosage; Radiology Information Systems - standards; Reproducibility of Results; scoring method; Sensitivity and Specificity; Transducers - standards; validation studies
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.25284

Purpose To date, few studies have validated the Prostate Imaging Reporting and Data System Version 2 (PI‐RADS v. 2) for the diagnosis of prostate cancer. Our aim was to validate PI‐RADS v.2 using 3 Tesla (T) MRI. Materials and Methods This is a retrospective study of 54 consecutive patients who underwent 3T MRI with a body‐array coil for diagnostic confirmation of prostate cancer or cancer staging between June 2013 and June 2015. Sensitivity, specificity, and agreement were calculated based on a criterion of PI‐RADS score = 3. Inter‐examiner agreement was determined by the weighted kappa statistic. Results Histological findings were positive for cancer in 33 patients and negative in 21 patients. Considering a PI‐RADS score of 3 as positive for cancer, the accuracy of each reader was 85.20% and 70.40%, respectively, and agreement coefficients were κ = 0.69 and κ = 0.35. Considering PI‐RADS 3 as absence of cancer, the accuracy of each reader was 77.80% and 77.80%, respectively, and agreement was κ = 0.55 and κ = 0.54. Inter‐reader agreement was moderate/good (weighted κ = 0.53; 95% confidence interval: 0.39–0.66; P = 0.038). Conclusion High accuracy was obtained for the diagnosis of prostate cancer using 3T MRI with a body coil and the PI‐RADS v.2 score. J. Magn. Reson. Imaging 2016;44:1354–1359.