Digital subtraction pulmonary arteriography versus multidetector CT in the detection of pulmonary arteriovenous malformations.

PURPOSE: To compare digital subtraction pulmonary arteriography (PA) with 16-detector row computed tomography (CT) in the detection of suspected pulmonary arteriovenous malformations (PAVMs) in patients with hereditary hemorrhagic telangiectasia (HHT). MATERIALS AND METHODS: Eighteen nonconsecutive patients (median age, 47.5 years; range, 26-78 y) with a total of 42 PAVMs were included over a period of 2.75 years. At the authors' institution, all patients with HHT and their family members undergo contrast echocardiography. Positive contrast echocardiography findings prompt multidetector CT (MDCT) scanning, which, in the case of positive findings, is then followed by digital subtraction PA and embolotherapy as appropriate. Catheter-based PA was performed in the study group drawn from the group that underwent MDCT and PA. Evaluation of PAVM presence, location, and type in PA studies was conducted by three blinded interventional radiology physician reviewers and compared with the readings of MDCT studies by three blinded MDCT physician reviewers. Consensus review was performed after blinded readings were complete. RESULTS: Whole-lung analysis (ie, correct identification of a lesion anywhere in the lung) showed MDCT readings to have a mean sensitivity of 83% and specificity of 78% and PA readings to have a mean sensitivity of 70% and specificity of 100%. Lobar analysis (ie, correct identification of a lesion in a given lobe) showed MDCT readings to have a mean sensitivity of 72% and specificity of 93% and PA readings to have a mean sensitivity of 68% and specificity of 100%. CONCLUSIONS: According to the definitions in this study, MDCT provides greater sensitivity in the detection of PAVM than digital subtraction PA, but does so with a loss in specificity, and the differences depend on the level analyzed (ie, lung vs lobe).

Imaging techniques for detection and management of endoleaks after endovascular aortic aneurysm repair.

Endovascular aortic aneurysm repair (EVAR) is evolving into a viable alternative to open surgical repair for many patients with abdominal and thoracic aortic aneurysms. Endoleak development is a complication of EVAR and represents one of the limitations of this procedure. Endoleaks represent blood flow outside the stent-graft lumen but within the aneurysm sac. Lifelong imaging surveillance of patients after EVAR is critical to detect endoleaks for the patient's benefit and to determine the long-term performance of the stent-graft. Although computed tomographic angiography is the most commonly used examination for imaging surveillance, magnetic resonance angiography, ultrasonography, and digital subtraction angiography all have a role in endoleak detection and management. This review will focus on imaging techniques used for endoleak detection and the role imaging surveillance plays in the overall care of the post-EVAR patient.

Method for reduced SAR T1rho-weighted MRI.

A reduced specific absorption rate (SAR) version of the T(1rho)-weighted MR pulse sequence was designed and implemented. The reduced SAR method employs a partial k-space acquisition approach in which a full power spin-lock pulse is applied to only the central phase-encode lines of k-space, while the remainder of k-space receives a low-power spin-lock pulse. Acquisition of high- and low-power phase-encode lines are interspersed chronologically to minimize average power deposition. In this way, the majority of signal energy in the central portion of k-space receives full T(1rho)-weighting, while the average SAR of the overall acquisition can be reduced, thereby lowering the minimum safely allowable TR. The pulse sequence was used to create T(1rho) maps of a phantom, an in vivo mouse brain, and the brain of a human volunteer. In the images of the human brain, SAR was reduced by 40% while the measurements of T(1rho) differed by only 2%. The reduced SAR sequence enables T(1rho)-weighted MRI in a clinical setting, even at high field strengths.

In vivo measurement of T1rho dispersion in the human brain at 1.5 tesla.

PURPOSE: To measure T1rho relaxation times and T1rho dispersion in the human brain in vivo. MATERIALS AND METHODS: Magnetic resonance imaging (MRI) was performed on a 1.5-T GE Signa clinical scanner using the standard GE head coil. A fast spin-echo (FSE)-based T1rho-weighted MR pulse sequence was employed to obtain images from five healthy male volunteers. Optimal imaging parameters were determined while considering both the objective of the study and the guarantee that radio-frequency (RF) power deposition during MR did not exceed Food and Drug Administration (FDA)-mandated safety levels. RESULTS: T1rho-weighted MR images showed excellent contrast between different brain tissues. These images were less blurred than corresponding T2-weighted images obtained with similar contrast, especially in regions between brain parenchyma and cerebrospinal fluid (CSF). Average T1rho values for white matter (WM), gray matter (GM), and CSF were 85 +/- 3, 99 +/- 1, and 637 +/- 78 msec, respectively, at a spin-locking field of 500 Hz. T1rho is 30% higher in the parenchyma and 78% higher in CSF compared to the corresponding T2 values. T1rho dispersion was observed between spin-locking frequencies 0 and 500 Hz. CONCLUSION: T1rho-weighted MRI provides images of the brain with superb contrast and detail. T1rho values measured in the different brain tissues will serve as useful baseline values for analysis of T1rho changes associated with pathology.

T1rho-weighted MRI using a surface coil to transmit spin-lock pulses.

T1rho-weighted MRI is a novel basis for generating tissue contrast. However, it suffers from sensitivity to B1 inhomogeneity. First, excitation with a spatially varying B1 causes flip-angle artifacts and second, spin locking with an inhomogeneous B1 results in non-uniform T1rho contrast. In this study, we overcome the former complication with a specially designed spin-locking pulse sequence and we successfully obtain T1rho-weighted images with a surface coil. In this pulse sequence, the spin-lock pulse was divided into segments of equal duration and alternating phase. This "self-compensating" T1rho-preparatory pulse sequence was analyzed and the effect of an inhomogeneous B1 field was simulated using the Bloch equations. T1rho-weighted MR images of a phantom and a human knee joint in vivo were obtained on a clinical scanner with a surface coil to demonstrate the utility of the pulse sequence. The self-compensating T1rho-prepared pulses sequence resulted in substantially reduced image artifacts compared to the conventional, single-phase spin-lock pulse.

Pulse sequence for multislice T1rho-weighted MRI.

A 2D multislice spin-lock (MS-SL) MR pulse sequence is presented for rapid volumetric T1rho-weighted imaging. Image quality is compared with T1rho-weighted data collected using a single-slice (SS) SL sequence and T2-weighted data from a standard MS spin-echo (SE) sequence. Saturation of longitudinal magnetization by the application of nonselective SL pulses is experimentally measured and theoretically modeled as T2rho decay. The saturation data is used to correct the image data as a function of the SL pulse duration to make quantitative measurements of T1rho. Measurements of T1rho using the saturation-corrected MS-SL data are nearly identical to those measured using an SS-SL sequence. The MS-SL sequence produces quantitative T1rho maps of an entire sample volume with the high-SNR advantages conferred by SE-based sequences.

Three-dimensional T1rho-weighted MRI at 1.5 Tesla.

PURPOSE: To design and implement a magnetic resonance imaging (MRI) pulse sequence capable of performing three-dimensional T(1rho)-weighted MRI on a 1.5-T clinical scanner, and determine the optimal sequence parameters, both theoretically and experimentally, so that the energy deposition by the radiofrequency pulses in the sequence, measured as the specific absorption rate (SAR), does not exceed safety guidelines for imaging human subjects. MATERIALS AND METHODS: A three-pulse cluster was pre-encoded to a three-dimensional gradient-echo imaging sequence to create a three-dimensional, T(1rho)-weighted MRI pulse sequence. Imaging experiments were performed on a GE clinical scanner with a custom-built knee-coil. We validated the performance of this sequence by imaging articular cartilage of a bovine patella and comparing T(1rho) values measured by this sequence to those obtained with a previously tested two-dimensional imaging sequence. Using a previously developed model for SAR calculation, the imaging parameters were adjusted such that the energy deposition by the radiofrequency pulses in the sequence did not exceed safety guidelines for imaging human subjects. The actual temperature increase due to the sequence was measured in a phantom by a MRI-based temperature mapping technique. Following these experiments, the performance of this sequence was demonstrated in vivo by obtaining T(1rho)-weighted images of the knee joint of a healthy individual. RESULTS: Calculated T(1rho) of articular cartilage in the specimen was similar for both and three-dimensional and two-dimensional methods (84 +/- 2 msec and 80 +/- 3 msec, respectively). The temperature increase in the phantom resulting from the sequence was 0.015 degrees C, which is well below the established safety guidelines. Images of the human knee joint in vivo demonstrate a clear delineation of cartilage from surrounding tissues. CONCLUSION: We developed and implemented a three-dimensional T(1rho)-weighted pulse sequence on a 1.5-T clinical scanner.

Artifacts in T(1rho)-weighted imaging: correction with a self-compensating spin-locking pulse.

Significant artifacts arise in T(1rho)-weighted imaging when nutation angles suffer small deviations from their expected values. These artifacts vary with spin-locking time and amplitude, severely limiting attempts to perform quantitative imaging or measurement of T(1rho) relaxation times. A theoretical model explaining the origin of these artifacts is presented in the context of a T(1rho)-prepared fast spin-echo imaging sequence. Experimentally obtained artifacts are compared to those predicted by theory and related to B(1) inhomogeneity. Finally, a "self-compensating" spin-locking preparatory pulse cluster is presented, in which the second half of the spin-locking pulse is phase-shifted by 180 degrees. Use of this pulse sequence maintains relatively uniform signal intensity despite large variations in flip angle, greatly reducing artifacts in T(1rho)-weighted imaging.

Indirect 17(O)-magnetic resonance imaging of cerebral blood flow in the rat.

Proton T(1rho)-dispersion MRI is demonstrated for indirect, in vivo detection of (17)O in the brain. This technique, which may be readily implemented on any clinical MRI scanner, is applied towards high-resolution, quantitative mapping of cerebral blood flow (CBF) in the rat by monitoring the clearance of (17)O-enriched water. Strategies are derived and employed for 1) quantitation of absolute H(2) (17)O tracer concentration from a ratio of high- and low-frequency spin-locked T(1rho) images, and 2) mapping CBF without having to transform the T(1rho) signal to H(2) (17)O tracer concentration. Absolute regional blood flow was mapped in a single 3-mm brain slice at an in-plane resolution of 0.4 x 0.8 mm within a 5-min tracer washout time; these data are consistent with the less localized CBF measurements reported in the literature. T(1rho)-weighted imaging yields excellent signal-to-noise ratios, spatiotemporal resolution, and anatomical contrast for mapping CBF.