Faculty Bibliography

OBJECTIVE. The purpose of this article is to perform an intraindividual comparison between 1.5 T and 3 T chemical-shift MRI in differentiating adrenal adenomas and nonadenomas, including comparison of quantitative thresholds. MATERIALS AND METHODS. In this retrospective study, 37 adrenal lesions in 36 patients (20 men and 16 women; mean [+/- SD] age, 66.7 +/- 12.9 years; 27 benign adenomas in 27 patients; 10 nonadenomas in nine patients) imaged at 1.5 T and 3 T were identified. Two readers qualitatively assessed intralesional signal loss between in- and opposed-phase images. One reader placed ROIs on adrenal lesions, spleen, liver, and muscle. Quantitative measures of signal loss, such as signal intensity (SI) index, adrenal-to-spleen ratio, adrenal-to-liver ratio, and adrenal-to-muscle ratio, were calculated. Qualitative and quantitative measures between field strengths were assessed with McNemar test and ROC analysis, respectively. RESULTS. Accuracy in qualitative adenoma identification (86.5% [32/37] at 1.5 T and 81.1% [30/37] at 3 T for reader 1; 81.1% [30/37] at 1.5 T and 83.8% [31/37] at 3 T for reader 2; both p >/= 0.180) was equivalent at both field strengths. AUCs were not statistically significantly different between field strengths for quantitative measures: AUCs at 1.5 T versus 3 T were 0.956 versus 0.915 for SI index, 0.963 versus 0.870 for adrenal-to-spleen ratio, 0.935 versus 0.852 for adrenal-to-liver ratio, and 0.948 versus 0.948 for adrenal-to-muscle ratio (all p > 0.11). The optimal threshold for SI index was lower at 3 T (> 7.4%) than at 1.5 T (> 21.6%) but had similar sensitivity (1.5 T, 92.6% [25/27]; 3 T, 88.9% [24/27]) and specificity (1.5 T, 90.0% [9/10]; 3 T, 90.0% [9/10]). CONCLUSION. Chemical-shift imaging has similar diagnostic efficacy for differentiating adrenal adenomas and nonadenomas at 1.5 T and 3 T. However, quantitative measures have different thresholds for this differentiation at 3 T; in particular, the commonly applied SI index is much lower at 3 T.

OBJECTIVE. The purpose of this study was to retrospectively evaluate the impact of multiparametric prostate MRI, including diffusion-weighted imaging (DWI) performed using different b values as well as dynamic contrast-enhanced MRI (DCE-MRI) on the accuracy, sensitivity, and specificity for transition zone (TZ) tumor detection and localization. MATERIALS AND METHODS. We included 106 prostate cancer patients (mean age [+/- SD], 62 +/- 7 years) who underwent 3-T MRI with a pelvic phased-array coil before radical prostatectomy. Three radiologists independently reviewed cases to record the likelihood of tumor in each of six TZ regions. Scores were initially assigned using T2-weighted imaging alone, reassigned after integration of DWI at b = 1000 s/mm(2) and corresponding apparent diffusion coefficient (ADC) maps, reassigned again after integration of DWI at b = 2000 s/mm(2), and reassigned a final time after integration of DCE-MRI. Generalized estimating equations based on binary logistic regression were used to compare sessions for TZ tumor detection, using prostatectomy findings as reference standard. RESULTS. Of the TZ sextants, 9.7% (62/636) contained tumor. All readers had higher sensitivity for T2-weighted imaging integrated with DWI at b = 1000 s/mm(2) and ADC compared with T2-weighted imaging alone (reader 1, 54.8% vs 33.9%; reader 2, 53.2% vs 22.6%; and reader 3, 50.0% vs 19.4% [p /= 0.054). Other measures were similar across the four sessions (reader 1, specificity 97.4-98.3% and accuracy 91.2-95.9%; reader 2, specificity 95.8-98.4% and accuracy 91.0-92.6%; reader 3, specificity 90.9-96.7% and accuracy 88.1-89.2%). CONCLUSION. DWI assists TZ tumor detection through higher sensitivity, particularly when using a very high b value; DCE-MRI lacks further additional benefit.

PURPOSE: To explore associations of whole-lesion histogram diffusion metrics with pathologic findings and subsequent metastatic disease in bladder cancer patients undergoing radical cystectomy. METHODS: Twenty-three bladder cancer patients (21M, 2F; mean 70 +/- 11 years) underwent MRI before cystectomy. A volume-of-interest was placed on all slices on the ADC map encompassing each lesion. Whole-lesion mean, kurtosis, and skewness of ADC were calculated and compared with T stage and pelvic nodal status at cystectomy and with subsequent metastasis in 20/25 patients with available follow-up. RESULTS: At cystectomy, 39 % (9/23) were stage T2, 61 % (14/23) >/=T3, and 28 % (5/23) exhibited positive nodes; 35 % (7/20) developed later metastases. Mean ADC was significantly lower in stage >/=T3 than in lower stage tumors (1.20 +/- 0.36 x 10-3 vs. 1.55 +/- 0.36 x 10-3 mm2/s; p = 0.044), but showed no association with nodal or metastatic disease (p = 0.362-0.709). Kurtosis was significantly lower in tumors with, compared to without, nodal disease (-0.05 +/- 0.29 vs. 0.91 +/- 1.16; p = 0.037), and showed a non-significant decrease in tumors with, compared to without, later metastases (0.23 +/- 0.63 vs. 0.83 +/- 0.89; p = 0.088). Kurtosis was not associated with T stage (p = 0.811), and skew was not associated with any outcome (p = 0.516-0.643). Mean ADC achieved highest AUC for identification of stage >/=T3 (AUC = 0.754 vs. 0.516-0.643 for other metrics). Kurtosis achieved highest AUC for nodal disease (AUC = 0.811 vs. 0.522-0.556 for other metrics) and metastases (AUC = 0.736 vs. 0.516-0.626 for other metrics). Only difference in AUC between skewness and kurtosis for nodal disease was significant (p = 0.031). CONCLUSION: While requiring larger studies, kurtosis has potential to complement mean ADC in bladder cancer prognosis using whole-lesion histogram analysis.

PURPOSE: To compare image quality of monopolar and bipolar diffusion-weighted imaging (DWI) sequences of the liver at 3T. METHODS: 32 healthy volunteers (mean 27 +/- 8 years; 27 M/5F) and 11 patients (mean age 58 +/- 14 years; 8 M/3F) underwent liver MRI using a 3T system incorporating 2-channel parallel transmission for B1-shimming and reduced B1-inhomogeneity. Scans included free-breathing DWI sequences (b-value 0, 400, 800 s/mm2) acquired using both monopolar and bipolar techniques. Estimated signal-to-noise ratio (eSNR), apparent diffusion coefficient (ADC), and measures of subjective image quality on b-800 images, scored on a 1-5 scale by two independent radiologists, were compared between sequences. RESULTS: Monopolar sequence demonstrated significantly higher eSNR (volunteers: 12.7 +/- 4.0 vs. 11.3 +/- 3.5, patients: 11.4 +/- 4.0 vs. 10.2 +/- 3.3; p /= 0.191). Hepatic ADC was significantly lower using monopolar technique only in volunteers (1.28 +/- 0.12 vs. 1.43 +/- 0.15, p < 0.001). CONCLUSION: In comparison with past studies performed at 1.5T, when using a modern 3T system, we observed improved image quality of liver DWI using a monopolar, rather than a bipolar, acquisition scheme, largely attributed to higher eSNR.

Purpose To assess the effect of the new Organ Procurement and Transplantation Network ( OPTN Organ Procurement and Transplantation Network )/United Network for Organ Sharing ( UNOS United Network for Organ Sharing ) policy on hepatocellular carcinoma ( HCC hepatocellular carcinoma ) detection and liver transplant allocation in patients with cirrhosis undergoing dynamic contrast material-enhanced liver magnetic resonance (MR) imaging. Materials and Methods In this HIPAA-compliant institutional review board-approved retrospective study with waiver of informed consent, 247 patients (196 men, 51 women; mean age, 60 years +/- 11 [standard deviation]) with liver cirrhosis who underwent evaluation for HCC hepatocellular carcinoma with MR imaging were identified via database search. Three radiologists independently reviewed images and identified number and size of HCC hepatocellular carcinoma based on criteria within either the prior or revised policy. Based on these interpretations, priority for liver transplantation for each patient was determined with prior and revised transplantation allocation criteria. HCC hepatocellular carcinoma detection was compared between sessions by using McNemar tests, and interreader agreement for detection of at least one HCC hepatocellular carcinoma was assessed by using kappa coefficients. Results All three readers detected significantly more 1-2-cm HCC hepatocellular carcinoma s with the revised policy (readers detected 22, eight, and 20 1-2-cm HCC hepatocellular carcinoma s) versus the prior policy (no reader detected 1-2-cm HCC hepatocellular carcinoma s) (P