Prostate Cancer Detection Using Computed Very High b-value Diffusion-weighted Imaging: How High Should We Go?
Rosenkrantz, Andrew B; Parikh, Nainesh; Kierans, Andrea S; Kong, Max Xiangtian; Babb, James S; Taneja, Samir S; Ream, Justin M
RATIONALE AND OBJECTIVES: The aim of this study was to assess prostate cancer detection using a broad range of computed b-values up to 5000 s/mm2. MATERIALS AND METHODS: This retrospective Health Insurance Portability and Accountability Act-compliant study was approved by an institutional review board with consent waiver. Forty-nine patients (63 +/- 8 years) underwent 3T prostate magnetic resonance imaging before prostatectomy. Examinations included diffusion-weighted imaging (DWI) with b-values of 50 and 1000 s/mm2. Seven computed DWI image sets (b-values: 1000, 1500, 2000, 2500, 3000, 4000, and 5000 s/mm2) were generated by mono-exponential fit. Two blinded radiologists (R1 [attending], R2 [fellow]) independently evaluated diffusion weighted image sets for image quality and dominant lesion location. A separate unblinded radiologist placed regions of interest to measure tumor-to-peripheral zone (PZ) contrast. Pathologic findings from prostatectomy served as reference standard. Measures were compared between b-values using the Jonckheere-Terpstra trend test, Spearman correlation coefficient, and generalized estimating equations based on logistic regression for correlated data. RESULTS: As b-value increased, tumor-to-PZ contrast and benign prostate suppression for both readers increased (r = +0.65 to +0.71, P = 0.001), whereas anatomic clarity, visualization of the capsule, and visualization of peripheral-transition zone edge decreased (r = -0.69 to -0.75, P = 0.003). Sensitivity for tumor was highest for R1 at b1500-3000 (84%-88%) and for R2 at b1500-2500 (70%-76%). Sensitivities for both pathologic outcomes were lower for both readers at both b1000 and the highest computed b-values. Sensitivity for Gleason >6 tumor was highest for R1 at b1500-3000 (90%-93%) and for R2 at 1500-2500 (78%-80%). The positive predictive value for tumor for R1 was similar from b1000 to 4000 (93%-98%) and for R2 was similar from b1500 to 4000 (88%-94%). CONCLUSIONS: Computed b-values in the range of 1500-2500 s/mm2 (but not higher) were optimal for prostate cancer detection; b-values of 1000 or 3000-5000 exhibited overall lower performance.
PMID: 26992738
ISSN: 1878-4046
CID: 2032222
Use of a Quality Improvement Initiative to Achieve Consistent Reporting of Level of Suspicion for Tumor on Multiparametric Prostate MRI
Rosenkrantz, Andrew B; Pujara, Akshat C; Taneja, Samir S
OBJECTIVE: The purpose of this article is to evaluate the utility of a quality improvement (QI) initiative in achieving long-term adherence to an evolving structured format for reporting the level of suspicion for tumor on prostate MRI examinations. MATERIALS AND METHODS: The original QI initiative occurred over a 4-month period in 2010, before which prostate MRI was reported using free text. The initiative consisted of development of a section-wide macro, an initial group training session, ordering physician input regarding the structured report's value, subsequent weekly sessions for ongoing review, and timely individualized feedback in instances of nonuse. The initial structured report included pick lists for describing the level of suspicion for tumor as negative, low, medium, or high. Pick lists were modified in 2011 to incorporate a 5-point Likert scale and again in 2015 to incorporate Prostate Imaging Data and Reporting System (PI-RADS) version 2. These refinements were implemented after accelerated training periods. The frequency of reports providing an MRI-based suspicion level during these periods was assessed. RESULTS: Fifty-five percent of reports provided an MRI-based level of suspicion for tumor before the initiative. For various cohorts evaluated after the initiative (using structured reports based on the low, medium, or high scheme; a numeric Likert scale; or PI-RADS), this frequency improved to 95-100% (p < 0.001). Among reports without a suspicion level, potential confounding factors included marked artifact from hip prosthesis and overt diffuse tumor. CONCLUSION: The QI initiative achieved excellent adherence in reporting a suspicion level for tumor on prostate MRI examinations. The described components of the initiative were useful for maintaining long-term adherence that persisted after serial modifications to the report lexicon.
PMID: 27105339
ISSN: 1546-3141
CID: 2080232