Faculty Bibliography

A publicly available dataset containing k-space data as well as Digital Imaging and Communications in Medicine image data of knee images for accelerated MR image reconstruction using machine learning is presented.

PURPOSE/OBJECTIVE:To compare liver stiffness measurements obtained from MR elastography with liver T1 relaxation times obtained from T1 mapping and T2 relaxation times obtained from T2 mapping for detection and staging of liver fibrosis. MATERIALS AND METHODS/METHODS:223 patients with known or suspected liver disease underwent MRI of the liver with T1 mapping (Look-Locker sequence) and 2D SE-EPI MR elastography (MRE) sequences. 139 of these patients also underwent T2 mapping with radial T2 TSE sequence. Two readers (R1 & R2) measured liver stiffness, T1 relaxation times and T2 relaxation times. T1 and T2 times were correlated with stiffness measurements. ROC analysis was used to compare the performance of both techniques in discriminating fibrosis stage in 23 patients who underwent liver biopsy. RESULTS:For each reader there was significant moderate positive correlation between liver MRE and liver T1 mapping (r = 0.49 and 0.36). There was significant moderate positive correlation between liver T2 mapping and each of MRE and T1 mapping for one of the readers (r = 0.40 and 0.27). AUC for differentiating early (F0-F2) from advanced (F3-F4) fibrosis in biopsied patients was 0.975 (R1) and 0.925 (R2) for MRE, 0.671 (R1) and 0.642 (R2) for T1 mapping and 0.671 (R1) and 0.743 (R2) for T2 mapping. Inter-reader agreement was good for MRE (ICC = 0.84) substantial for T1 mapping (0.94) and T2 mapping (0.96). CONCLUSIONS:Liver T1 and T2 mapping showed moderate positive correlation with MR elastography. Accuracy of MRE is however superior to T1 and T2 mapping in the subset of patients who underwent liver biopsy. Accuracy of combination of MRE and T1 mapping/T2 mapping was not superior to MRE alone.

PURPOSE/OBJECTIVE:-RAVE) and to evaluate the multi relaxation components in the liver of healthy controls and chronic liver disease (CLD) patients. METHODS:components among patients (n = 3) and a control group (n = 10). RESULTS:relaxation time measurement relative to the reference on 2 different scanners. The coefficient of variation for test-retest scans performed on the same scanner was 5.7% and 2.4% for scans performed on 2 scanners. The comparison between healthy controls and CLD patients showed a significant difference (P < .05) in mono relaxation time (P = .002), stretched-exponential relaxation parameter (P = .04). The Akaike information criteria C criterion showed 2.53 ± 0.9% (2.3 ± 0.3% for CLD) of the voxels are bi-exponential while in 65.3 ± 5.8% (81.2 ± 0.06% for CLD) of the liver voxels, the stretched-exponential model was preferred. CONCLUSION/CONCLUSIONS:assessment of the liver during free breathing and can distinguish between healthy volunteers and CLD patients.

PURPOSE/OBJECTIVE:To propose a highly accelerated, high-resolution dynamic contrast-enhanced MRI (DCE-MRI) technique called GRASP-Pro (golden-angle radial sparse parallel imaging with imProved performance) through a joint sparsity and self-calibrating subspace constraint with automated selection of contrast phases. METHODS:GRASP-Pro reconstruction enforces a combination of an explicit low-rank subspace-constraint and a temporal sparsity constraint. The temporal basis used to construct the subspace is learned from an intermediate reconstruction step using the low-resolution portion of radial k-space, which eliminates the need for generating the basis using auxiliary data or a physical signal model. A convolutional neural network was trained to generate the contrast enhancement curve in the artery, from which clinically relevant contrast phases are automatically selected for evaluation. The performance of GRASP-Pro was demonstrated for high spatiotemporal resolution DCE-MRI of the prostate and was compared against standard GRASP in terms of overall image quality, image sharpness, and residual streaks and/or noise level. RESULTS:Compared to GRASP, GRASP-Pro reconstructed dynamic images with enhanced sharpness, less residual streaks and/or noise, and finer delineation of the prostate without prolonging reconstruction time. The image quality improvement reached statistical significance (P < 0.05) in all the assessment categories. The neural network successfully generated contrast enhancement curves in the artery, and corresponding peak enhancement indexes correlated well with that from the manual selection. CONCLUSION/CONCLUSIONS:GRASP-Pro is a promising method for rapid and continuous DCE-MRI. It enables superior reconstruction performance over standard GRASP and allows reliable generation of artery enhancement curve to guide the selection of desired contrast phases for improving the efficiency of GRASP MRI workflow.

OBJECTIVES/OBJECTIVE:The 8th International Forum for Liver Magnetic Resonance Imaging (MRI), held in Basel, Switzerland, in October 2017, brought together clinical and academic radiologists from around the world to discuss developments in and reach consensus on key issues in the field of gadoxetic acid-enhanced liver MRI since the previous Forum held in 2013. METHODS:Two main themes in liver MRI were considered in detail at the Forum: the use of gadoxetic acid for contrast-enhanced MRI in patients with liver cirrhosis and the technical performance of gadoxetic acid-enhanced liver MRI, both opportunities and challenges. This article summarises the expert presentations and the delegate voting on consensus statements discussed at the Forum. RESULTS AND CONCLUSIONS/CONCLUSIONS:It was concluded that gadoxetic acid-enhanced MRI has higher sensitivity for the diagnosis of hepatocellular carcinoma (HCC), when compared with multidetector CT, by utilising features of hyperenhancement in the arterial phase and hypointensity in the hepatobiliary phase (HBP). Recent HCC management guidelines recognise an increasing role for gadoxetic acid-enhanced MRI in early diagnosis and monitoring post-resection. Additional research is needed to define the role of HBP in predicting microvascular invasion, to better define washout during the transitional phase in gadoxetic acid-enhanced MRI for HCC diagnosis, and to reduce the artefacts encountered in the arterial phase. Technical developments are being directed to shortening the MRI protocol for reducing time and patient discomfort and toward utilising faster imaging and non-Cartesian free-breathing approaches that have the potential to improve multiphasic dynamic imaging. KEY POINTS/CONCLUSIONS:• Gadoxetic acid-enhanced MRI provides higher diagnostic sensitivity than CT for diagnosing HCC. • Gadoxetic acid-enhanced MRI has roles in early-HCC diagnosis and monitoring post-resection response. • Faster imaging and free-breathing approaches have potential to improve multiphasic dynamic imaging.

Accelerating Magnetic Resonance Imaging (MRI) by taking fewer measurements has the potential to reduce medical costs, minimize stress to patients and make MRI possible in applications where it is currently prohibitively slow or expensive. We introduce the fastMRI dataset, a large-scale collection of both raw MR measurements and clinical MR images, that can be used for training and evaluation of machine-learning approaches to MR image reconstruction. By introducing standardized evaluation criteria and a freely-accessible dataset, our goal is to help the community make rapid advances in the state of the art for MR image reconstruction. We also provide a self-contained introduction to MRI for machine learning researchers with no medical imaging background

Purpose: To determine the impact of DWI and T2WI acquisition parameters on first-order hepatic texture measures at 3T MRI.
Material(s) and Method(s): Five healthy volunteers (3 M/2F, mean 40 years old) were prospectively imaged at 3T using baseline liver free-breathing DWI and T2WI acquisition twice to assess scan-rescan repeatability. Three modifications to acquisition parameters were also performed individually: decreased averages (2 vs. 4); lower resolution (DWI: 128x96 vs. 192x144 and T2WI: 192x192 vs. 320x320); and increased slice thickness (8 vs. 4 mm). A single reader placed four co-registered hepatic ROIs using 3D Slicer v4.8.1 (https://urldefense.proofpoint.com/v2/url?u=http-3A__www.slicer.org&d=DwIFAg&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=EQR3KLkQ5UWCWWT7EfebH2P_dJeKQhvwk7yvrJe5GJY&m=VVljDEDjGLS_4z5jZ0uX9AVqXkAPM24hpGmZl06It_E&s=TQM-Y7ippXB-a-cXGwkMg-DnVAXTLHOB9hyiAIzdwXQ&e= ). 10 first-order histogram texture features (average of the four ROI) were compared to baseline acquisition. Percent difference (%diff) and coefficient of variance (CV) were computed using MedCalc.
Result(s): For ADC, 8 out of 10 parameters were repeat-able with <10% scan-rescan %diff; Skewness and Minimum were least repeatable with >10% scan-rescan %diff. Entropy was the only parameter that had < 10% CV and %diff for all acquisition schemes; all other parameters had >10% CV for at least one modified acquisition scheme. Skewness, Minimum, and Variance had the largest average CV. Change in slice thickness had the largest impact on most texture features. For T2WI, 9 out of 10 parameters were repeatable with <10% scan-rescan %diff; Skewness had >10% scan-rescan %diff. Entropy and Uniformity were the only two parameters that had <15% CV and %diff for all acquisition schemes. Change in slice thickness had the largest impact on most texture features.
Conclusion(s): ADC and T2WI first-order hepatic texture features, except for entropy, depend on acquisition parameters. Care must be taken to maintain identical acquisition schemes to compare changes in these features, such as after treatment

Incidence of kidney cancer is steadily rising likely due to incidental detection of small renal masses. Nearly 20% of all small renal masses (less than 4cm in size) are benign at pathology. Furthermore, these small renal cancers represents 70% of all newly diagnosed Renal Cell Cancers (RCC). Detection of small renal masses leads to management dilemma, as it is not always possible to characterise these lesions on conventional imaging. Inability to discriminate benign from malignant and indolent from aggressive tumour results in surgical treatment for many of these patients. Surgery provides excellent oncologic control but it is associated with increased morbidity. Treatment based on tumour aggressiveness will result in optimal outcome by selecting patients with aggressive tumour for surgery and avoiding unnecessary surgery in indolent tumours. Such a paradigm requires non-invasive methods to accurately diagnose tumours of different aggressiveness. Tumours of different histopathology differ in tumour aggressiveness. Imaging can help investigate renal tumour histopathology and aggressiveness and can impact treatment decision and lower treatment cost. Imaging can assist with: (A). Differentiating benign renal masses from malignant tumours. - Protenacious or hemorrhagic cysts are hyperdense on CT and T1 hyperintense on MRI. Differentiating these lesions from solid masses requires assessment of enhancement. To assess for enhancement we can obtain pre-and post-contrast imaging on CT or MRI. Dual energy CT permits a single phase CT imaging to assess for enhancement, thus decreasing radiation dose. Subtraction imaging on MRI is useful to assess for enhancement. - Benign angiomyolipoma (AML) contain bulk fat. This can be easily diagnosed on CT and MRI. - Lipid poor AML can be difficult to differentiate from other types of renal masses on CT imaging. MRI can be helpful in suggesting the diagnosis of lipid poor AML. Some of the MRI features that suggest diagnosis of AML include homogenous and uniform enhancement, homogenous T2 signal similar to that of muscle, and restricted diffusion with low ADC. Suggesting this diagnosis is important as diagnosis of lipid poor AML can be made confidently at core biopsy, thus avoiding surgery in these patients - It is nearly impossible to discriminate benign oncocytoma from chromophobe and clear cell subtypes of kidney cancers on conventional imaging. However, diffusion weighted imaging (DWI) and perfusion weighted imaging (PWI) has shown some promise in small pilot studies. (B). Tumour subtypes of solid R

PURPOSE/OBJECTIVE:To compare qualitative results obtained from computer-aided dual-ratio analysis on T1-weighted two-point Dixon, with T2*-corrected multi-echo Dixon and T2-corrected multi-echo single-voxel MR spectroscopy sequence (MRS) for evaluation of liver fat and iron at 3T. METHODS AND MATERIALS/METHODS:This retrospective, HIPAA-compliant, IRB-approved study included 479 patients with known or suspected liver disease. Two-point Dixon, multi-echo Dixon, and MR spectroscopy sequences were performed for each patient at 3T. A receiver-operating characteristic analysis was performed to compare the diagnostic performance in 80 patients using biopsy as the standard. Sensitivity, specificity, PPV, and NPV of qualitative two-point Dixon results, multi-echo Dixon (PDFF and R2*), and MRS (fat fraction and R2 water) for detection of hepatic steatosis and siderosis were assessed. RESULTS:, respectively. CONCLUSION/CONCLUSIONS:The computer-aided dual-ratio discrimination with two-point Dixon is a useful qualitative screening tool with high negative predictive value for hepatic steatosis and iron overload. Multi-echo Dixon and MRS have similar accuracy for detection of hepatic steatosis and iron overload at 3 Tesla.