Background: Direct intraindividual comparison of dynamic CT myocardial perfusion imaging (MPI) and machine learning (ML)-based CT fractional flow reserve (FFR) has not been explored for diagnosing hemodynamically significant coronary artery disease. Purpose: To investigate the diagnostic performance of dynamic CT MPI and ML-based CT FFR for functional assessment of coronary stenosis. Materials and Methods: Between January 2, 2017, and October 17, 2018, consecutive participants with stable angina were prospectively enrolled. All participants underwent dynamic CT MPI coronary CT angiography and invasive conventional coronary angiography (CCA) FFR within 2 weeks. Receiver operating characteristic (ROC) curve analysis was used to assess diagnostic performance. Results: Eighty-six participants (mean age, 67 years +/- 12 [standard deviation]; 67 men) with 157 target vessels were included for final analysis. The mean radiation doses for dynamic CT MPI and coronary CT angiography were 3.6 mSv +/- 1.1 and 2.7 mSv +/- 0.8, respectively. Myocardial blood flow (MBF) was lower in ischemic segments compared with nonischemic segments and reference segments (defined as the territory of vessels without stenosis) (75 mL/100 mL/min +/- 20 vs 148 mL/100 mL/min +/- 22 and 169 mL/100 mL/min +/- 34, respectively, both P < .001). Similarly, CT FFR was also lower for hemodynamically significant lesions than for hemodynamically nonsignificant lesions (0.68 +/- 0.1 vs 0.83 +/- 0.1, respectively, P < .001). MBF had the largest area under the ROC curve (AUC) (using 99 mL/100 mL/min as a cutoff) among all parameters, outperforming ML-based CT FFR (AUC = 0.97 vs 0.85, P < .001). The vessel-based specificity and diagnostic accuracy of MBF were higher than those of ML-based CT FFR (93% vs 68%, P < .001 and 94% vs 78%, respectively, P = .04) whereas the sensitivity of both methods was similar (96% vs 88%, respectively, P = .11). Conclusion: Dynamic CT myocardial perfusion imaging was able to help accurately evaluate the hemodynamic significance of coronary stenosis using a reduced amount of radiation. In addition, the myocardial blood flow derived from dynamic CT myocardial perfusion imaging outperformed machine learning-based CT fractional flow reserve for identifying lesions causing ischemia. (C) RSNA, 2019
Purpose: To determine the diagnostic performance of dual-energy virtual noncalcium (VNCa) CT in the detection of bone marrow edema in study participants with sacroiliitis associated with axial spondyloarthritis. Materials and Methods: In this prospective study, 47 consecutive participants (mean age, 27 years; age range, 14-41 years [28 male; mean age, 24 years; age range, 14-37 years] [19 female; mean age, 29 years; age range, 17-41 years]) underwent dual-energy CT and 3.0-T MRI between April 2016 and December 2017. Two independent readers visually evaluated all sacroiliac joints for the presence of abnormal marrow attenuation on dual-energy VNCa images using a four-point classification system (0, no edema; 1, mild edema; 2, moderate edema; 3, severe edema). CT numbers on VNCa images were determined with region-of-interestbased quantitative analysis. MRI was the reference standard for presence of bone marrow edema. Results: Sensitivity, specificity, and accuracy of readers 1 and 2, respectively, in the identification of bone edema at CT were 87% and 93% (48 and 51 of 55), 94% and 91% (32 and 31 of 34), and 90% and 92% (80 and 82 of 89). Interobserver agreement was excellent (kappa = 0.81). CT numbers from VNCa images increased from no edema to severe edema (P<.001). The area under the receiver operating characteristic curve was 0.93 for reader 1 and 0.91 for reader 2 in differentiation of the presence of bone marrow edema from no edema. A cutoff value of 233 HU derived from reader 1 yielded overall sensitivity, specificity, and accuracy of 90% (49 of 55), 83% (28 of 34), and 87% (77 of 89) in the detection of any extent of edema in the sacroiliac joints. Conclusion: Dual-energy VNCa CT images had excellent diagnostic performance in evaluation of the extent of bone marrow edema in study participants with sacroiliitis associated with axial spondyloarthritis. (c) RSNA, 2018
Purpose: To evaluate a radiomics model for predicting lymph node (LN) metastasis in biliary tract cancers (BTCs) and to determine its prognostic value for disease-specific and recurrence-free survival. Materials and Methods: For this retrospective study, a radiomics model was developed on the basis of a primary cohort of 177 patients with BTC who underwent resection and LN dissection between June 2010 and December 2016. Radiomic features were extracted from portal venous CT scans. A radiomics signature was built on the basis of reproducible features by using the least absolute shrinkage and selection operator method. Multivariable logistic regression model was adopted to establish a radiomics nomogram. Nomogram performance was determined by its discrimination, calibration, and clinical usefulness. The model was internally validated in 70 consecutive patients with BTC between January 2017 and February 2018. Results: The radiomics signature, composed of three LN-status-related features, was associated with LN metastasis in primary and validation cohorts (P<.001). The radiomics nomogram that incorporated radiomics signature and CT-reported LN status showed good calibration and discrimination in primary cohort (area under the curve, 0.81) and validation cohort (area under the curve, 0.80). Patients at high risk of LN metastasis portended lower disease-specific and recurrence-free survival than did those at low risk after surgery (both P<.001). High-risk LN metastasis was an independent preoperative predictor of disease-specific survival (hazard ratio, 3.37; P<.001) and recurrence-free survival (hazard ratio, 1.98; P=.003). Conclusion: A radiomics model derived from portal phase CT of the liver has good performance for predicting lymph node metastasis in biliary tract cancer and may help to improve clinical decision making. (c) RSNA, 2018
Purpose: To develop and validate a computational model for estimating hepatic venous pressure gradient (HVPG) based on CT angiographic images, termed virtual HVPG, to enable the noninvasive diagnosis of portal hypertension in patients with cirrhosis. Materials and Methods: In this prospective multicenter diagnostic trial (ClinicalTrials.gov identifier: NCT02842697), 102 consecutive eligible participants (mean age, 47 years [range, 21-75 years]; 68 men with a mean age of 44 years [range, 21-73 years] and 34 women with a mean age of 52 years [range, 24-75 years]) were recruited from three high-volume liver centers between August 2016 and April 2017. All participants with cirrhosis of various causes underwent transjugular HVPG measurement, Doppler US, and CT angiography. Virtual HVPG was developed with a three-dimensional reconstructed model and computational fluid dynamics. Results: In the training cohort (n = 29), the area under the receiver operating characteristic curve (AUC) of virtual HVPG in the prediction of clinically significant portal hypertension (CSPH) was 0.83 (95% confidence interval [CI]: 0.58, 1.00). The diagnostic performance was prospectively confirmed in the validation cohort (n = 73), with an AUC of 0.89 (95% CI: 0.81, 0.96). Inter-and intraobserver agreement was 0.88 and 0.96, respectively, suggesting the good reproducibility of virtual HVPG measurements. There was good correlation between virtual HVPG and invasive HVPG (R = 0.61, P<.001), with a satisfactory performance to rule out (7.3 mm Hg) and rule in (13.0 mm Hg) CSPH. Conclusion: The accuracy of a computational model of virtual hepatic venous pressure gradient (HVPG) shows significant correlation with invasive HVPG. The virtual HVPG also showed a good performance in the noninvasive diagnosis of clinically significant portal hypertension in cirrhosis. (c) RSNA, 2018
Purpose: To develop a quantitative assessment of collateral perfusion at CT and to investigate its value in the prediction of outcome in patients with acute ischemic stroke (AIS). Materials and Methods: This retrospective study reviewed data from consecutive patients with AIS and an occluded M1 segment of the middle cerebral artery who underwent pretreatment perfusion CT between May 2009 and August 2017. The maximum cerebral blood flow (CBF) of collateral vessels (cCBF(max)) within the Sylvian fissure was calculated for each patient. Good outcome was defined as a 90-day modified Rankin scale score of 0-2. Multivariable logistic regression analysis was used to determine the relationship between cCBF(max) and (a) hemorrhagic transformation and (b) clinical outcome. Results: The final analysis included 204 patients (median age, 73 years; interquartile range, 62-80 years; 82 [40.2%] women). Multivariable logistic regression analysis showed that higher cCBF(max) was an independent predictor for (a) a lower risk of hemorrhagic transformation (odds ratio [OR], 0.99; 95% confidence interval [CI]: 0.98, 1.00; P = .009) after adjusting for baseline National Institutes of Health Stroke Scale (NIHSS), endovascular thrombectomy, baseline infarct core volume, and recanalization and (b) better outcome (OR, 1.02; 95% CI: 1.01, 1.03; P = .001) after adjusting for age, baseline NIHSS score, endovascular thrombectomy, hypertension, baseline infarct core volume, and recanalization, respectively. Conclusion: The measurement of maximum cerebral blood flow of collateral vessels within the Sylvian fissure is a feasible quantitative collateral assessment at perfusion CT. Maximum cerebral blood flow of collateral vessels was associated with clinical outcome in patients with acute ischemic stroke. (C) RSNA, 2019
Background: A quantifiable imaging measure to gauge the intensity of individual inflammatory lesions in axial spondyloarthritis (SpA) has not been well established. Previous studies have shown that diffusion-weighted (DW) MRI reflects disease activity in axial SpA. Purpose: To determine- the association between apparent diffusion coefficient (ADC) at MRI of discovertebral lesions and disease in individuals with axial SpA. Materials and Methods: In this prospective study, 243 study participants (mean age +/- standard deviation, 43.2 years +/- 13.5) with back pain who fulfilled the Assessment of Spondylo Arthritis International Society criteria for SpA were recruited from four rheu-matology centers between April 2014 and March 2018. There were 132 men (mean age, 41.4 years +/- 13.3) and 111 women (mean age, 45.3 years +/- 13.4). Clinical, biochemical, and radiologie parameters were collected. All participants underwent whole-spine MR1 by using a short inversion time inversion-recovery sequence and DW imaging.Two independent readers identified the presence of discovertebral lesions. ADCs were measured and normalized with normal bone marrow. Regression analysis was performed to determine association between the mean, maximum, and normalized mean and maximum ADGs of the discovertebral lesions and disease activity and functional parameters (Barth Ankylosing Spondylitis Disease Activity Index [BASDAI], Bath Ankylosing Spondylitis Functional Index [BASFI], and Bath Ankylosing Spondylitis Global index [BASGI]). Results: Ninety-one discovertebral lesions (five cervical, 61 thoracic, 25 lumbar) were present in 55 of the 243 study participants (22.6%). After adjusting for confounding factors, increased maximum ADC was independently associated with increased BASFI (regression coefficient [beta] = 1.94 [x 10(-3) mm(2) sec]. P .04). Increased normalized maximum ADC was independently associated with BASDAI question 2 (ie, back pain score) (beta = 0.45, P = .01), mean stiffness score (beta = 0.41, P = .04), and BASGI (beta = 0.43, P = .04). Increased normalized mean ADC was independently associated with BASDAI question 2 (beta = 0.61, P = .04). Conclusion: Apparent diffusion coefficients at MRI of discovertebral lesions were associated with disease activity, functional impairment, and patient global assessment in axial spondyloarthritis. (C) RSNA, 2019
There are increasing reports of a type of lung cancer that manifests as solitary cystic airspaces. The purpose of this case series was to identify the CT features and possible mechanisms of solitary cystic lung cancer, on the basis of CT observations and pathologic characteristics. The clinical, imaging, and pathologic data of 106 patients with solitary cystic lung cancer were collected and analyzed between January 2011 and December 2017. CT images were reviewed independently by three radiologists who were blinded to pathologic findings. Demographic data and clinical and smoking status were extracted from the medical records. The mean age was 58.8 years +/- 10.6 (standard deviation) (range, 30-82 years). CT features in the 106 patients included nonuniform cystic walls in 96 (90.6%) patients, cyst septations in 62 (58.5%) patients, nodular walls in 58 54.7%) patients, ground-glass opacity around the cyst in 53 patients (50.0%), and irregular margins in 42 (39.6%) patients. At histologic examination, the majority of cases (81 [87.1%] of 93) were adenocarcinoma. (c) RSNA, 2019
Background: A major technical challenge of prostatic arterial embolization (PAE) is the identification and catheterization of the prostatic arteries (PAs). Recently, MR angiography has been shown to help visualize PAs, but the clinical utility of MR angiography for this purpose is not known. Purpose: To determine the efficacy of contrast material-enhanced MR angiography in identifying the PA and to evaluate its role in PAE for benign prostatic hyperplasia (BPH). Materials and Methods: In this prospective study, 100 consecutive men who were scheduled to undergo PAE for BPH from January 2015 to May 2017 were assigned by using a randomized block design to either group A (n = 50; mean age, 71.7 years 6 11.9 [standard deviation]) without MR angiography or group B (n = 50; mean age, 72.3 years +/- 12.2) with MR angiography prior to PAE. MR angiography findings of the PA anatomy were compared with those of digital subtraction angiography (DSA). The Student t test and Wilcoxon rank-sum test were used to compare the differences between the parameters indicating the performance of PAE. Results: The mean age of the 100 men in the study was 72.0 years +/- 11.8 (range, 51-88 years). Compared with DSA as the reference standard, MR angiography identified PAs with a sensitivity of 91.5% (97 of 106) and a positive predictive value of 100% (97 of 97). With the knowledge of tube obliquity and anatomy, group B had lower procedure times than group A (82.3 minutes +/- 5.4 vs 123.9 minutes +/- 12.4, P < .001) and shorter fluoroscopy times (13.8 minutes +/- 2.7 vs 28.5 minutes +/- 8.0, P < .001). Additionally, radiation dose was reduced for group A versus group B, from a median of 920 to 339 mGy (P = .004). Conclusion: Contrast-enhanced MR angiography can accurately show anatomy for the prostate arteries, leading to shorter prostatic artery embolization times and lower radiation dose than when preprocedural prostate MR angiography is not performed. Published under a CC BY 4.0 license.
Background: Nasopharyngeal carcinoma (NPC) may be cured with radiation therapy. Tumor proximity to critical structures demands accuracy in tumor delineation to avoid toxicities from radiation therapy; however, tumor target contouring for head and neck radiation therapy is labor intensive and highly variable among radiation oncologists. Purpose: To construct and validate an artificial intelligence (AI) contouring tool to automate primary gross tumor volume (GTV) contouring in patients with NPC. Materials and Methods: In this retrospective study, MRI data sets covering the nasopharynx from 1021 patients (median age, 47 years; 751 male, 270 female) with NPC between September 2016 and September 2017 were collected and divided into training, validation, and testing cohorts of 715, 103, and 203 patients, respectively. GTV contours were delineated for 1021 patients and were defined by consensus of two experts. A three-dimensional convolutional neural network was applied to 818 training and validation MRI data sets to construct the AI tool, which was tested in 203 independent MRI data sets. Next, the AI tool was compared against eight qualified radiation oncologists in a multicenter evaluation by using a random sample of 20 test MRI examinations. The Wilcoxon matched-pairs signed rank test was used to compare the difference of Dice similarity coefficient (DSC) of pre-versus post-AI assistance. Results: The AI-generated contours demonstrated a high level of accuracy when compared with ground truth contours at testing in 203 patients (DSC, 0.79; 2.0-mm difference in average surface distance). In multicenter evaluation, AI assistance improved contouring accuracy (five of eight oncologists had a higher median DSC after AI assistance; average median DSC, 0.74 vs 0.78; P < .001), reduced intra-and interobserver variation (by 36.4% and 54.5%, respectively), and reduced contouring time (by 39.4%). Conclusion: The AI contouring tool improved primary gross tumor contouring accuracy of nasopharyngeal carcinoma, which could have a positive impact on tumor control and patient survival. (C) RSNA, 2019
Background: The Vesical Imaging-Reporting and Data System (VI-RADS) scoring system was created in 2018 to standardize imaging and reporting of bladder cancer staging with multiparametric MRI. The system provides a five-point VI-RADS score, which suggests the likelihood of detrusor muscle invasion. Muscle-invasive disease carries a worse prognosis and requires radical surgery. Purpose: To determine the performance of the VI-RADS score in detecting muscle-invasive bladder cancer in a cohort of patients undergoing multiparametric MRI before surgery. Materials and Methods: In this retrospective study, a total of 340 patients with bladder cancer were identified from a database of consecutive patients undergoing multiparametric MRI from November 2011 to August 2018. The tumor with the largest burden was selected in those patients with multifocal tumors. Bladder tumors were retrospectively categorized according to the VI-RADS five-point scoring system by two readers, independently and in consensus, who were blinded to histologic findings. The VI-RADS score was compared with postoperative pathology for each tumor, and the performance of VI-RADS for determining detrusor muscle invasion was analyzed by using the Cochran-Armitage test. Results: Among the 340 patients, there were 296 men and 44 women; the median age was 64.0 years (interquartile range [IQR], 57.0-87.0 years). Of 340 tumors, 255 (75.0%) were verified as non-muscle-invasive and 85 (25.0%) as muscle-invasive bladder cancer. Both the VI-RADS score and its components were associated with muscle-invasive condition (P < .001). The area under the receiver operating characteristic curve for VI-RADS for muscle invasion was 0.94 (95% confidence interval [CI]: 0.90, 0.98). The sensitivity and specificity of a VI-RADS score of 3 or greater were 87.1% (95% CI: 78%, 93%) and 96.5% (95% CI: 93%, 98%), respectively. Conclusion: The Vesical Imaging-Reporting and Data System score effectively defines the likelihood of detrusor muscle invasion in bladder cancer and should be considered for evaluation of tumors prior to surgery. (C) RSNA, 2019
Background: Renal impairment is common in patients with coronary artery disease and, if severe, late gadolinium enhancement (LGE) imaging for myocardial infarction (MI) evaluation cannot be performed. Purpose: To develop a fully automatic framework for chronic MI delineation via deep learning on non-contrast material-enhanced cardiac cine MRI. Materials and Methods: In this retrospective single-center study, a deep learning model was developed to extract motion features from the left ventricle and delineate MI regions on nonenhanced cardiac cine MRI collected between October 2015 and March 2017. Patients with chronic MI, as well as healthy control patients, had both nonenhanced cardiac cine (25 phases per cardiac cycle) and LGE MRI examinations. Eighty percent of MRI examinations were used for the training data set and 20% for the independent testing data set. Chronic MI regions on LGE MRI were defined as ground truth. Diagnostic performance was assessed by analysis of the area under the receiver operating characteristic curve (AUC). MI area and MI area percentage from nonenhanced cardiac cine and LGE MRI were compared by using the Pearson correlation, paired t test, and Bland-Altman analysis. Results: Study participants included 212 patients with chronic MI (men, 171; age, 57.2 years +/- 12.5) and 87 healthy control patients (men, 42; age, 43.3 years +/- 15.5). Using the full cardiac cine MRI, the per-segment sensitivity and specificity for detecting chronic MI in the independent test set was 89.8% and 99.1%, respectively, with an AUC of 0.94. There were no differences between nonenhanced cardiac cine and LGE MRI analyses in number of MI segments (114 vs 127, respectively; P = .38), per-patient MI area (6.2 cm(2) +/- 2.8 vs 5.5 cm(2) +/- 2.3, respectively; P = .27; correlation coefficient, r = 0.88), and MI area percentage (21.5% +/- 17.3 vs 18.5% +/- 15.4; P = .17; correlation coefficient, r = 0.89). Conclusion: The proposed deep learning framework on nonenhanced cardiac cine MRI enables the confirmation (presence), detection (position), and delineation (transmurality and size) of chronic myocardial infarction. However, future larger-scale multicenter studies are required for a full validation. Published under a CC BY 4.0 license
Background: The evidence of combining sorafenib with transarterial chemoembolization (TACE) for intermediate-stage recurrent hepatocellular carcinoma (HCC) is limited. Patient responses to this treatment varied because of the heterogeneous nature of intermediate-stage recurrent HCC, making it important to identify patients who are most likely to benefit from this combination therapy. Purpose: To compare sorafenib administered in combination with TACE versus TACE alone in the treatment of recurrent intermediate-stage HCC after initial hepatectomy and to determine the relationship of microvascular invasion (MVI) to survival. Materials and Methods: In this retrospective multicenter study, 3652 consecutive patients were found to have intrahepatic recurrences after initial hepatectomy of primary HCC from January 2010 to December 2016. Of these, 260 patients with intermediatestage recurrent HCC underwent combination treatment with sorafenib and TACE or TACE alone. Overall survival (OS) and progression-free survival (PFS) were compared between these two treatments according to MVI status by using log-rank tests. Results: A total of 128 patients were administered combination therapy (mean age, 55 years +/- 7.6 [standard deviation]; 107 men) and 132 patients were administered TACE alone (mean age, 56 years +/- 8.3; 110 men). The 5-year OS and PFS were higher in the combination group than in the TACE group (OS: 38.9% vs 20.5%, respectively, P =.01; PFS, 37.5% vs 18.7%, respectively, P =.003). For patients with MVI-positive lesions, the median OS and PFS after combination treatment (n = 55) were longer than those after TACE alone (n = 72; OS: 17.2 months vs 12.1 months, respectively, P =.02; PFS: 17.0 months vs 11.0 months, respectively, P = .02). Multivariable analysis showed that tumor number, MVI status, and treatment allocation were significant predictors of OS and PFS, whereas tumor size was a prognostic factor for PFS. Conclusion: Patients with recurrent intermediate-stage hepatocellular carcinoma and lesions positive for microvascular invasion (MVI) had longer survival times by using a combined treatment of sorafenib with transarterial chemoembolization (TACE) compared with TACE alone; patients with MVI-negative lesions did not show survival benefit from combined therapy. (C) RSNA, 2019.
Background: It is well known that white matter injuries observed at birth are associated with adverse neurodevelopmental outcomes later in life. Whether white matter developmental variations in healthy newborns are also associated with changes in later neurodevelopment remains to be established. Purpose: To evaluate whether developmental variations of white matter microstructures identified by MRI correlate with neurodevelopmental outcomes in healthy full-term infants. Materials and Methods: In this prospective study, pregnant women were recruited and their healthy full-term newborns underwent a brain MRI including diffusion tensor imaging at approximately 2 weeks of age. These infants were tested at approximately 2 years of age with the Bayley Scales of Infant Development (BSID). Voxel-wise correlation analyses of fractional anisotropy (FA), measured with diffusion tensor MRI, and neurodevelopmental test scores, measured by using BSID, were performed by using tract-based spatial statistics (TBSS), followed by region-of-interest (ROI) analyses of correlations between mean FA in selected white matter ROIs and each BSID subscale score. Results: Thirty-eight full-term infants (20 boys, 18 girls) underwent MRI examination at 2 weeks of age (14.3 days +/- 1.6) and BSID measurement at 2 years of age (732 days +/- 6). TBSS analyses showed widespread clusters in major white matter tracts, with positive correlations (P <= .05, corrected for the voxel-wise multiple comparisons) between FA values and multiple BSID subscale scores. These correlations were largely independent of several demographic parameters as well as family environment. Gestational age at birth appeared to be a confounding factor as TBSS-observed correlations weakened when it was included as a covariate; however, after controlling for gestational age at birth, ROI analyses still showed positive correlations (P <= .05, R = 0.35 to 0.48) between mean FA in many white matter ROIs and BSID cognitive, language, and motor scores. Conclusion: There were significant associations between white matter microstructure developmental variations in healthy full-term newborns and their neurodevelopmental outcomes. (C) RSNA, 2019