Workflow for automatic renal perfusion quantification using ASL-MRI and machine learning.

PURPOSE: Clinical applicability of renal arterial spin labeling (ASL) MRI is hampered because of time consuming and observer dependent post-processing, including manual segmentation of the cortex to obtain cortical renal blood flow (RBF). Machine learning has proven its value in medical image segmentation, including the kidneys. This study presents a fully automatic workflow for renal cortex perfusion quantification by including machine learning-based segmentation. METHODS: Fully automatic workflow was achieved by construction of a cascade of 3 U-nets to replace manual segmentation in ASL quantification. All 1.5T ASL-MRI data, including M0 , T1 , and ASL label-control images, from 10 healthy volunteers was used for training (dataset 1). Trained cascade performance was validated on 4 additional volunteers (dataset 2). Manual segmentations were generated by 2 observers, yielding reference and second observer segmentations. To validate the intended use of the automatic segmentations, manual and automatic RBF values in mL/min/100 g were compared. RESULTS: Good agreement was found between automatic and manual segmentations on dataset 1 (dice score = 0.78 ± 0.04), which was in line with inter-observer variability (dice score = 0.77 ± 0.02). Good agreement was confirmed on dataset 2 (dice score = 0.75 ± 0.03). Moreover, similar cortical RBF was obtained with automatic or manual segmentations, on average and at subject level; with 211 ± 31 mL/min/100 g and 208 ± 31 mL/min/100 g (P < .05), respectively, with narrow limits of agreement at -11 and 4.6 mL/min/100 g. RBF accuracy with automated segmentations was confirmed on dataset 2. CONCLUSION: Our proposed method automates ASL quantification without compromising RBF accuracy. With quick processing and without observer dependence, renal ASL-MRI is more attractive for clinical application as well as for longitudinal and multi-center studies.

Feasibility of Velocity-Selective Arterial Spin Labeling in Breast Cancer Patients for Noncontrast-Enhanced Perfusion Imaging.

BACKGROUND: Dynamic contrast-enhanced (DCE) MRI is the most sensitive method for detection of breast cancer. However, due to high costs and retention of intravenously injected gadolinium-based contrast agent, screening with DCE-MRI is only recommended for patients who are at high risk for developing breast cancer. Thus, a noncontrast-enhanced alternative to DCE is desirable. PURPOSE: To investigate whether velocity selective arterial spin labeling (VS-ASL) can be used to identify increased perfusion and vascularity within breast lesions compared to surrounding tissue. STUDY TYPE: Prospective. POPULATION: Eight breast cancer patients. FIELD STRENGTH/SEQUENCE: A 3 T; VS-ASL with multislice single-shot gradient-echo echo-planar-imaging readout. ASSESSMENT: VS-ASL scans were independently assessed by three radiologists, with 3-25 years of experience in breast radiology. Scans were scored on lesion visibility and artifacts, based on a 3-point Likert scale. A score of 1 corresponded to "lesions being distinguishable from background" (lesion visibility), and "no or few artifacts visible, artifacts can be distinguished from blood signal" (artifact score). A distinction was made between mass and nonmass lesions (based on BI-RADS lexicon), as assessed in the standard clinical exam. STATISTICAL TESTS: Intra-class correlation coefficient (ICC) for interobserver agreement. RESULTS: The ICC was 0.77 for lesion visibility and 0.84 for the artifact score. Overall, mass lesions had a mean score of 1.27 on lesion visibility and 1.53 on the artifact score. Nonmass lesions had a mean score of 2.11 on lesion visibility and 2.11 on the artifact score. DATA CONCLUSION: We have demonstrated the technical feasibility of bilateral whole-breast perfusion imaging using VS-ASL in breast cancer patients. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 1.

Exploring label dynamics of velocity-selective arterial spin labeling in the kidney.

PURPOSE: Velocity-selective arterial spin labeling (VSASL) has been proposed for renal perfusion imaging to mitigate planning challenges and effects of arterial transit time (ATT) uncertainties. In VSASL, label generation may shift in the vascular tree as a function of cutoff velocity. Here, we investigate label dynamics and especially the ATT of renal VSASL and compared it with a spatially selective pulsed arterial spin labeling technique, flow alternating inversion recovery (FAIR). METHODS: Arterial spin labeling data were acquired in 7 subjects, using free-breathing dual VSASL and FAIR with five postlabeling delays: 400, 800, 1200, 2000, and 2600 ms. The VSASL measurements were acquired with cutoff velocities of 5, 10, and 15 cm/s, with anterior-posterior velocity-encoding direction. Cortical perfusion-weighted signal, temporal SNR, quantified renal blood flow, and arterial transit time were reported. RESULTS: In contrast to FAIR, renal VSASL already showed fairly high signal at the earliest postlabeling delays, for all cutoff velocities. The highest VSASL signal and temporal SNR was obtained with a cutoff velocity of 10 cm/s at postlabeling delay = 800 ms, which was earlier than for FAIR at 1200 ms. Fitted ATT on VSASL was ≤ 0 ms, indicating ATT insensitivity, which was shorter than for FAIR (189 ± 79 ms, P < .05). Finally, the average cortical renal blood flow measured with cutoff velocities of 5 cm/s (398 ± 84 mL/min/100 g) and 10 cm/s (472 ± 160 mL/min/100 g) were similar to renal blood flow measured with FAIR (441 ± 84 mL/min/100 g) (P > .05) with good correlations on subject level. CONCLUSION: Velocity-selective arterial spin labeling in the kidney reduces ATT sensitivity compared with the recommended pulsed arterial spin labeling method, as well as if cutoff velocity is increased to reduce spurious labeling due to motion. Thus, VSASL has potential as a method for time-efficient, single-time-point, free-breathing renal perfusion measurements, despite lower tSNR than FAIR.

Multi-organ comparison of flow-based arterial spin labeling techniques: Spatially non-selective labeling for cerebral and renal perfusion imaging.

PURPOSE: Flow-based arterial spin labeling (ASL) techniques provide a transit-time insensitive alternative to the more conventional spatially selective ASL techniques. However, it is not clear which flow-based ASL technique performs best and also, how these techniques perform outside the brain (taking into account eg, flow-dynamics, field-inhomogeneity, and organ motion). In the current study we aimed to compare 4 flow-based ASL techniques (ie, velocity selective ASL, acceleration selective ASL, multiple velocity selective saturation ASL, and velocity selective inversion prepared ASL [VSI-ASL]) to the current spatially selective reference techniques in brain (ie, pseudo-continuous ASL [pCASL]) and kidney (ie, pCASL and flow alternating inversion recovery [FAIR]). METHODS: Brain (n = 5) and kidney (n = 6) scans were performed in healthy subjects at 3T. Perfusion-weighted signal (PWS) maps were generated and ASL techniques were compared based on temporal SNR (tSNR), sensitivity to perfusion changes using a visual stimulus (brain) and robustness to respiratory motion by comparing scans acquired in paced-breathing and free-breathing (kidney). RESULTS: In brain, all flow-based ASL techniques showed similar tSNR as pCASL, but only VSI-ASL showed similar sensitivity to perfusion changes. In kidney, all flow-based ASL techniques had comparable tSNR, although all lower than FAIR. In addition, VSI-ASL showed a sensitivity to B1 -inhomogeneity. All ASL techniques were relatively robust to respiratory motion. CONCLUSION: In both brain and kidney, flow-based ASL techniques provide a planning-free and transit-time insensitive alternative to spatially selective ASL techniques. VSI-ASL shows the most potential overall, showing similar performance as the golden standard pCASL in brain. However, in kidney, a reduction of B1 -sensitivity of VSI-ASL is necessary to match the performance of FAIR.

Influence of labeling parameters and respiratory motion on velocity-selective arterial spin labeling for renal perfusion imaging.

PURPOSE: Arterial transit time uncertainties and challenges during planning are potential issues for renal perfusion measurement using spatially selective arterial spin labeling techniques. To mitigate these potential issues, a spatially non-selective technique, such as velocity-selective arterial spin labeling (VSASL), could be an alternative. This article explores the influence of VSASL sequence parameters and respiratory induced motion on VS-label generation. METHODS: VSASL data were acquired in human subjects (n = 15), with both single and dual labeling, during paced-breathing, while essential sequence parameters were systematically varied; (1) cutoff velocity, (2) labeling gradient orientation and (3) post-labeling delay (PLD). Pseudo-continuous ASL was acquired as a spatially selective reference. In an additional free-breathing single VSASL experiment (n = 9) we investigated respiratory motion influence on VS-labeling. Absolute renal blood flow (RBF), perfusion weighted signal (PWS), and temporal signal-to-noise ratio (tSNR) were determined. RESULTS: (1) With decreasing cutoff velocity, tSNR and PWS increased. However, undesired tissue labeling occurred at low cutoff velocities (≤ 5.4 cm/s). (2) Labeling gradient orientation had little effect on tSNR and PWS. (3) For single VSASL high signal appeared in the kidney pedicle at PLD < 800 ms, and tSNR and PWS decreased with increasing PLD. For dual VSASL, maximum tSNR occurred at PLD = 1200 ms. Average cortical RBF measured with dual VSASL (264 ± 34 mL/min/100 g) at a cutoff velocity of 5.4 cm/s, and feet-head labeling was slightly lower than with pseudo-continuous ASL (283 ± 55 mL/min/100 g). CONCLUSION: With well-chosen sequence parameters, tissue labeling induced by respiratory motion can be minimized, allowing to obtain good quality RBF maps using planning-free labeling with dual VSASL.

Enabling free-breathing background suppressed renal pCASL using fat imaging and retrospective motion correction.

PURPOSE: For free-breathing renal perfusion imaging using arterial spin labeling (ASL), retrospective image realignment has been found essential to reduce subtraction artifacts and, independently, background suppression has been demonstrated to reduce physiologic noise. However, negative results on ASL precision and accuracy have been reported for the combination of both. In this study, the effect of background suppression -level in combination with image registration on free-breathing renal ASL signal quality, with registration either on ASL-images themselves or guided by additionally acquired fat-images, was investigated. The results from free-breathing acquisitions were compared with the reference paced-breathing motion compensation strategy. METHODS: Pseudocontinuous ASL (pCASL) data with additional fat-images were acquired from 10 subjects at 1.5T with varying background suppression levels during free-breathing and paced-breathing. Images were registered using the ASL-images themselves (ASLReg) or using their corresponding fat-images (FatReg). Temporal signal-to-noise ratio (tSNR) served to evaluate precision and perfusion weighted signal (PWS) to assess accuracy. RESULTS: In combination with image registration, background suppression significantly improved tSNR by 50% (P < .05). For heavy suppression, ASLReg and FatReg showed similar performance in terms of tSNR and PWS. Background suppression with two inversion pulses induced a small, nonsignificant (P > .05) PWS reduction, but increased PWS accuracy. When applying heavy background suppression, free-breathing acquisitions resulted in similar ASL-quality to paced-breathing acquisitions. CONCLUSION: Background suppression was found beneficial for free-breathing renal pCASL precision without compromising accuracy, despite motion challenges. In combination with ASLReg or FatReg, background suppression enabled clinically viable free-breathing renal pCASL.