Faculty Bibliography

PURPOSE: The aim of this study was to evaluate the feasibility of using Twitter polls to assess public opinion regarding session content at a national specialty society meeting. METHODS: Twitter polls allow users to embed multiple-choice questions within tweets and automatically aggregate responses. Two radiologists attending the 2016 annual meeting of the ACR posted a Twitter poll containing the hashtag #ACR2016 during 10 meeting sessions addressing socioeconomics/advocacy, patient experience, and social media/informatics (20 polls total). Each poll contained a question asking for an opinion regarding the session's content. Polls were open for responses for 24 hours. RESULTS: The average number of responses per poll was significantly higher for the user with the larger number of Twitter followers (24.3 +/- 14.4 versus 11.2 +/- 9.8, P = .015). A total of 57% of respondents agreed that radiologists' payments should shift to value-based payments, and 86% agreed that radiologists should routinely survey their patients to monitor quality; however, 83% disagreed with basing physician payments on patient satisfaction scores. A total of 85% disagreed that the artificial intelligence supercomputer Watson will entirely replace radiologists. A total of 76% agreed that social media can drive business at less cost than standard marketing. A total of 56% agreed with the direction of the ACR's advocacy and regulatory efforts, whereas 74% considered the ACR's advocacy efforts to be moderately or very useful for their practice. A total of 50% planned to change their practice on the basis of keynote remarks by Dr Ezekiel Emanuel. CONCLUSIONS: Twitter polls provide a free and easy infrastructure to potentially capture global public sentiment during the course of a medical society meeting. Their use may enrich and promote discussions of key session content.

OBJECTIVE: The purpose of this study was to apply a visual assessment of the intensity and pattern of T1 hyperintensity at MRI to differentiate hemorrhagic renal cysts from renal cell carcinoma (RCC). MATERIALS AND METHODS: A total of 144 T1-hyperintense renal lesions (62 cysts, all showing no enhancement on subtracted contrast-enhanced images and either 2-year stability or unenhanced CT density > 70 HU, and 82 histologically confirmed RCCs) in 144 patients were included. Two radiologists independently characterized qualitative features of the T1 hyperintensity in each lesion on unenhanced T1-weighted images. An additional radiologist placed ROIs to measure lesions' T1 signal intensity normalized to that of the psoas muscle. Chi-square and unpaired t tests were performed to compare the pattern of T1 hyperintensity between groups. RESULTS: The T1 hyperintensity was considered marked in 62.9% of cysts and 17.1% of RCCs for reader 1 and in 46.8% of cysts and 8.5% of RCCs for reader 2 (p < 0.001). The T1 hyperintensity exhibited a diffusely homogeneous distribution in 88.7% of cysts and 7.3% of RCCs for reader 1 and in 72.6% of cysts and 4.9% of RCCs for reader 2 (p < 0.001). The combination of both diffusely homogeneous distribution and marked degree of T1 hyperintensity achieved sensitivities of 40.3-56.5%, specificities of 97.6-98.8%, and accuracies of 73.6-79.9% for the diagnosis of T1-hyperintense cysts. The two cases of RCC exhibiting this imaging pattern for at least one reader were both papillary RCCs. Normalized signal intensity was 2.39 +/- 0.99 in T1-hyperintense cysts versus 2.12 +/- 0.84 in T1-hyperintense RCCs (p = 0.088). CONCLUSION: Diffuse T1 hyperintensity, particularly when marked, strongly indicates a hemorrhagic cyst rather than an RCC. Deferral of follow-up imaging may be considered when this imaging appearance is encountered at unenhanced MRI.

PURPOSE: The purpose of the study is to perform a meta-analysis of studies investigating the diagnostic performance of apparent diffusion coefficient (ADC) values in separating high-risk from low-risk prostate cancer (PCa). METHODS: MEDLINE and EMBASE databases were searched in December 2015 for studies reporting diagnostic performance of ADC values for discriminating high-risk from low-risk PCa and providing sufficient data to construct 2 x 2 contingency tables. Diagnostic performance was quantitatively pooled using a bivariate random-effects model including subgroup analysis and assessment of study heterogeneity and methodological quality. RESULTS: 13 studies were included, providing 1107 tumor foci in 705 patients. Heterogeneity among studies was moderate (tau2 = 0.222). Overall sensitivity was 76.9% (95% CI 68.6-83.6%); overall specificity was 77.0% (95% CI 69.9-82.8%); and summary AUC was 0.67. Inverse correlation between sensitivity and specificity (rho = -0.58) indicated interstudy heterogeneity was partly due to variation in threshold for test positivity. Primary biases were readers' knowledge of Gleason score during ADC measurement, lack of prespecified ADC thresholds, and lack of prostatectomy as reference in some studies. Higher sensitivity was seen in studies published within the past 2 years and studies not using b value of at least 2000; higher specificity was associated with involvement of one, rather than two, readers measuring ADC. Field strength, coil selection, and advanced diffusion metrics did not significantly impact diagnostic performance. CONCLUSION: ADC values show moderate accuracy in separating high-risk from low-risk PCa, although important biases may overestimate performance and unexplained sources of heterogeneity likely exist. Further studies using a standardized methodology and addressing identified weaknesses may help guide the use of ADC values for clinical decision-making.

OBJECTIVES: To determine whether a combination of PCA3 and MRI suspicion score (mSS) could further optimize detection of prostate cancer on MRF-TB among men with no previous history of biopsy. MATERIALS AND METHODS: 187 men presenting to our institution between 6/12 and 8/14 who underwent multiparametric MRI and PCA3 prior to MRF-TB. Biopsy results, stratified by biopsy indication and PCA3 score, were recorded. Receiver operating characteristics (ROC) curves and multivariable logistic regressions were utilized to model the association of PCA3 and mSS with cancer detection on MRF-TB. RESULTS: PCA3 is associated with cancer detection on MRF-TB for men with no prior biopsies (AUC = 0.67, 95% CI 0.59-0.76). Using a cutoff of >/=35, PCA3 was associated with cancer risk among men with mSS 2-3 (p=0.004), but not among those with mSS 4-5 (p=0.340). The interaction of PCA3 and mSS demonstrated significantly higher discrimination for cancer than mSS alone (AUC: 0.83 vs. 0.79, p=0.0434). CONCLUSIONS: Urinary PCA3 is associated with mSS and the detection of cancer on MRF-TB for men with no prior biopsies. PCA3 notably demonstrates a high negative predictive value among mSS 2-3. However, in the case of high suspicion mpMRI, PCA3 is not associated with cancer detection on MRF-TB adding little to cancer diagnosis. Further studies are needed to evaluate the utility of PCA3 in predicting cancer among men with normal mpMRI.

PURPOSE: The purpose of the study was to identify the frequency and reasons for extra sequences in clinical liver MRI and MRCP examinations. METHODS: A total of 250 consecutive liver MRI and 250 consecutive MRCP examinations performed at a single institution were reviewed. Extra sequences performed in comparison with our standard institutional protocol were identified. Reasons for the extra sequences were identified. Overall trends were assessed. RESULTS: In significantly greater fractions of exams (p = 0.009-0.030), MRCP had >/=1 extra sequence (40.8% vs. 29.2%) and >/=2 extra sequences (16.0% vs. 5.6%) than liver MRI. The average number of extra sequences was significantly higher (p = 0.004) for MRCP (0.73 +/- 1.2) than liver MRI (0.44 +/- 0.88). Reasons for extra sequences were as follows: sequence repeated for patient motion (33.8% for liver MRI; 31.9% for MRCP); sequence repeated for anatomic coverage (24.3% for liver MRI; 19.8% for MRCP); sequence added by the radiologist (15.3% for liver MRI; 33.0% for MRCP); sequence repeated for other reason (17.1% for liver MRI; 12.6% for MRCP); and sequence added by the technologist (5.4% for liver MRI; 2.7% for MRCP). The most commonly repeated sequence due to motion was the axial fat-saturated turbo spin-echo T2-weighted sequence for both liver MRI and MRCP (54.7% and 29.3% of sequences repeated due to motion, respectively). CONCLUSION: For liver MRI and MRCP exams, sequences were most often repeated due to motion artifact (most often occurring on TSE T2WI), and sequences were most often added by the radiologist. The findings may help guide sequence optimization, quality improvement initiatives, and standardization of operations, for improving efficiency in abdominal MRI workflow.

PURPOSE: To compare image quality and diagnostic performance for detecting local recurrence (LR) of prostate cancer after radical prostatectomy (RP) between standard dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) and a high spatiotemporal resolution, continuously acquired Golden-angle RAdial Sparse Parallel acquisition employing compressed sensing reconstruction ("GRASP"). METHODS: A search was conducted for prostate MRI examinations performed in patients with PSA >/=0.2 ng/mL after RP in whom follow-up evaluation allowed classification as positive (>/=50% PSA reduction after pelvic radiation or positive biopsy) or negative (<50% PSA reduction after pelvic radiation; spontaneous PSA normalization) for LR, yielding 13 patients with standard DCE (11 LR+) and 12 with GRASP (10 LR+). Standard DCE had voxel size 3.0 x 1.9 x 1.9 mm and temporal resolution 5.5 s. GRASP had voxel size 1.0 x 1.1 x 1.1 cm and was retrospectively reconstructed at 2.3 s resolution. Two radiologists evaluated DCE sequences for image quality measures (1-5 scale) and the presence of LR. RESULTS: GRASP achieved higher scores than standard DCE from both readers (p < 0.001-0.136) for anatomic clarity (R1: 4.4 +/- 0.8 vs. 2.8 +/- 0.67 R2: 4.8 +/- 0.5 vs. 3.2 +/- 0.6), sharpness (3.6 +/- 0.9 vs. 2.5 +/- 0.7; 4.6 +/- 0.5 vs. 2.6 +/- 0.5), confidence in interpretation (3.8 +/- 0.8 vs. 3.1 +/- 0.9; 3.8 +/- 1.0 vs. 3.1 +/- 1.2), and conspicuity of detected lesions (4.7 +/- 0.5 vs. 3.8 +/- 1.1; 4.5 +/- 0.5 vs. 3.8 +/- 1.0). For detecting LR, GRASP also achieved higher sensitivity (70% vs. 36%; 80% vs. 45%), specificity (R1 and R2: 100% vs. 50%), and accuracy (75% vs. 38%; 83% vs. 46%) for both readers. CONCLUSION: Although requiring larger studies, high spatiotemporal resolution GRASP achieved substantially better image quality and diagnostic performance than standard DCE for detecting LR in patients with elevated PSA after prostatectomy.

OBJECTIVE: The purpose of this article is to compare radiologists', referring physicians', and patients' interpretations of expressions within radiology reports to describe findings of likely low clinical significance. SUBJECTS AND METHODS: Surveys were completed by abdominal radiologists (n = 13), physicians referring patients for abdominal CT (n = 59), and outpatients awaiting imaging (n = 51) at a large urban academic medical center. Surveys presented 10 expressions for describing an incidental 5-mm liver lesion and asked respondents to select from a list of choices their perceived likelihood that the lesion represented malignancy. Radiologists and referrers were asked supplemental questions. RESULTS: Compared with radiologists' concern, referrers' and patients' concerns were higher for four and seven of the 10 expressions. Only the expression "benign cyst" was associated with no concern in all groups; "most likely a cyst" and "too small to characterize" were associated with median levels of concern of 0% for radiologists, > 0% to 1% for referrers, and > 2% to 5% for patients. Expressions containing the phrase "not excluded" had the highest concern in all groups. Referrers' likelihood of ordering follow-up imaging varied widely for the expressions (e.g., "benign cyst," 2%; "cyst," 22%; "most likely a cyst," 46%; "most likely a cyst, although tumor not excluded," 75%). Overall, the preferred phrase for a 5-mm liver lesion with benign features in normal-risk patients was "cyst" among radiologists and "benign cyst" among referrers. Seventy-six percent of referring physicians thought that radiology reports should indicate whether follow-up imaging is recommended for such lesions. CONCLUSION: Ambiguity in radiologists' language for incidental low-risk findings may contribute to increased patient anxiety and follow-up testing, warranting greater radiologist attention and potentially new practice or reporting strategies.

PURPOSE: The aim of this study was to conduct a prospective pilot study comparing the diagnostic performance of MRI alone and F-FDG simultaneous PET/MRI using a diuresis protocol in bladder cancer patients. METHODS: Twenty-two bladder cancer patients underwent F-FDG PET/MRI, using intravenous furosemide and oral hydration for bladder clearance. A radiologist scored probability of tumor in 3 locations (urinary bladder, pelvic lymph nodes, nonnodal pelvis) using 1- to 3-point scale (1 = negative, 2 = equivocal, 3 = definite tumor). A nuclear medicine physician reviewed fused PET/MRI images, after which scores were reassigned based on combined findings. Follow-up pathologic and imaging data served as reference. Performances of MRI alone and PET/MRI were compared. RESULTS: Of these patients, 82%, 38%, and 18% were positive for bladder, pelvic nodal, and nonnodal pelvic tumor, respectively. At a score of 3, PET/MRI exhibited greater accuracy for detection of bladder tumor (86% vs 77%), metastatic pelvic lymph nodes (95% vs 76%), and nonnodal pelvic malignancy (100% vs 91%). In the bladder, PET changed the level of suspicion in 36% of patients (50% increased suspicion, 50% decreased suspicion), with 75% of these changes deemed correct based on reference standard. For pelvic lymph nodes, PET changed suspicion in 52% (36% increase, 64% decrease), with 95% of changes deemed correct. For nonnodal pelvis, PET changed suspicion in 9% (100% increase), with 100% deemed correct. CONCLUSIONS: Additional PET information helped to appropriately determine level of suspicion in multiple anatomic sites for otherwise equivocal findings on MRI alone. Although requiring larger studies, findings suggest a possible role for simultaneous PET/MRI to assist bladder cancer management.

OBJECTIVE: Our aim was to evaluate the frequency and outcomes of incidental breast lesions detected on abdominal MRI examinations. MATERIALS AND METHODS: Abdominal MRI reports for 11,462 women imaged at our institution from November 2007 through December 2014 were reviewed to identify those reporting an incidental breast lesion. Available breast imaging and pathology results were assessed to identify outcomes in these lesions. RESULTS: Incidental breast lesions were described in the MRI reports of 292 (3%) patients who underwent abdominal MRI during the study period; breast imaging was recommended for 192 of these 292 (66%) patients. Sixty-three of the 192 (33%) patients for whom follow-up breast imaging was recommended underwent such imaging at our institution. Twenty-one of these 63 (33%) lesions underwent biopsy or surgery; histologic sampling of these lesions yielded seven incidental cancers (invasive ductal, n = 6; invasive lobular, n = 1) and 14 benign diagnoses. Three additional cancers (invasive ductal, n = 2; invasive lobular, n = 1) and three benign diagnoses were discovered at pathology at outside institutions. Of the remaining 165 patients without a histologic diagnosis, the lesions in 95 (58%) patients were presumed to be benign because of stability over time. Seven of the 10 patients with a diagnosis of incidental cancer (age range, 53-86 years; mean +/- SD, 67.0 +/- 10.6 years) had not undergone screening mammography at our institution. The frequency of incidental breast cancer was 11% of patients subsequently undergoing follow-up breast imaging at our institution, 3% of all patients with reported breast lesions, and 0.09% of patients undergoing abdominal MRI examinations. CONCLUSION: Although incidental breast lesions were rarely detected on abdominal MRI, a considerable number of these lesions were found to represent breast cancer, particularly when leading to a recommendation for follow-up breast imaging. Therefore, it is important for radiologists interpreting abdominal MRI examinations to carefully evaluate for the presence of breast abnormalities.