DSC Perfusion: AIF Detection

Routine acquisition of DSC-MRI PWI datasets highly benefits from full automated post-processing. Selection of arterial input and venous output function is a key step that ensures robustness and reliability of unsupervised processing. A novel method of AIF and VOF selection is proposed by means of tubular filtering and simple analysis of mean temporal signals. Weighting factors the favor arterial and venous signals, as well as vessel orientations are derived. As a result, robustness of AIF and VOF selection was improved.

DSC MRI provides valuable perfusion parameters that correlate with brain tumor progression, but requires a difficult to measure arterial input function (AIF). Using a modification of standard tracer kinetics applied to a tissue perfusion model allows both the AIF and residue curve to be determined for each pixel. The parameters are estimated by Bayesian probability theory using Markov chain Monte Carlo simulations to sample the joint posterior probabilities for the parameters. Here we report DSC MRI investigations on a mouse gliomas that demonstrates characteristic perfusion parameters that do not require an independent measurement of an AIF.

In bolus tracking the perfusion errors associated with bolus delay/dispersion may be minimised using a local Arterial Input Function (AIF) analysis. We improve a previously presented local AIF method by minimising the influence of the Mean Transit Time (MTT) on the local AIF selection. This is particularly important for identifying local AIFs in regions bordering normal and abnormal MTT tissue. We assess the improvement by comparing the amount of delay/dispersion remaining in the deconvolved tissue response after each local AIF approach, and compare both methods with the standard global AIF analysis.

Perfusion-weighted Magnetic Resonance (MR) Imaging is used to assess the risk of tissue infarction in acute stroke patients and tumor angiogenesis in cancer patients. We compared circular global arterial input function (AIF) and local AIF algorithms, recently proposed automated methods for MR signal deconvolution. 13 patients with 2 MR scans within 48 hours were studied. The variation between global AIF cerebral blood flow (CBF) maps from the first and second scans was 0.220 ± 0.043, and between local AIF CBF maps was 0.263 ± 0.041 (P-value = 0.0015). Superior repeatability of global AIF-based CBF maps may be important in speedy diagnosis and risk stratification.

A variety of post-processing programs and algorithms for dynamic susceptibility contrast (DSC) MR perfusion are available; however, the accuracy and reliability of these programs have not been subject to a standardized quality control. We developed digital phantom data set, to evaluate the accuracy and characteristics of quantitative values derived from DSC perfusion analysis software. By using this phantom, we could check tracer-delay dependency for CBF, CBV, MTT, and Tmax, as well as linearity of CBF and MTT against true values.

Transverse relaxation in living tissue is contributed by the dephasing of spins due to diffusion in mesoscopic magnetic fields induced, for example, by paramagnetic tracer in the blood pool. The associated correlation time is commensurate with the echo time of typical measurement sequences. Varying the echo time changes the character of dephasing from reversible for short TE to irreversible for long TE. This dependence is quantified by calculating the transverse relaxation rate in the capillary network for multiple refocusing pulses in the static dephasing regime. This yields a modified formula for the mean vessel size in the vessel size imaging.

Two widely used methods exist for computing cerebral blood volume (CBV) in DSC-MRI perfusion, using measured signals as well as deconvolved residue function. Some authors claim that these methods do not deliver identical results. We explain that the methods must deliver equivalent results and any difference in obtained values is just caused by signal post-processing errors and wrong interpretation of indicator-dilution theory and convolution theorem. Identity of the two methods is shown in time- and frequency-domains as well as by means of numerical results. Possible sources of the processing errors are discussed and solutions how to avoid those are proposed.

Bolus tracking perfusion evaluation relies on the deconvolution of a tracer concentration time-courses in an arterial and a tissue voxel following the tracer kinetic model. The object of this work is to propose a method to design a data driven Tikhonov regularization filter in the Fourier domain and to compare it to the singular value decomposition (SVD) based approaches using the mathematical equivalence of Fourier and circular SVD (oSVD).

The residue function, R(t), is fundamental for description of microcirculation. To define this function is one of the aims of DSC perfusion MRI. It is found by solving an ill-posed problem, the deconvolution, for which one of approaches is to fit a model R(t) to data. Commonly, phenomenological functional shapes are used to model R(t) respecting only its most general properties. Studies based on ASL indicate insufficiency of this approach. In this work we present a derivation the residue function from the laws of laminar flow and a model for the architecture of the vascular tree.

This study investigates the use of a rat model for the MR-guided focused ultrasound treatment of hepatocellular carcinoma. PRF-thermometry, thermal dose calculation and post-ablation imaging are used to determine the ablated liver area and compared to necropsy. Thermal dose reliably predicts the ablated area for single sonications, but care has to be taken to avoid overestimation in lesions resulting from multiple sonications.

We showed the evidence of MRI-guided focused ultrasound can suppress the ictal actvity induced by the chemical kindling of rat brain via kainic acid. This evidence demonstrate that non-invasive suppression of epilepsy may be feasible using pulsed, low-energy focused ultrasound.

MR-HIFU is emerging as a treatment modality for a variety of pathologies. Treatments near tissue interfaces can result in unwanted heating caused by an impedance mismatch. This research uses the proton resonance frequency measured by MRI to explore the changes in heating pattern and shift in sonication focus as a result of proximity to an interface in a thermal phantom. Air, acrylic, rubber and a tissue-equivalent gel pad were tested with treatment cells focused at 1, 2 and 4 cm from the interface material revealing 0.7 to 3 mm shifts depending on focal position and interface material.

Ultrasound monitoring of MR-guided transcranial ultrasound therapy could help identify control parameters to better deliver therapy to the brain. An MR-compatible PVDF hydrophone with a high sensitivity was constructed and characterized. The hydrophone was used to monitor microbubble-mediated ultrasound disruption of the blood-brain barrier in rats. Comparison of captured acoustic emissions with T1w and T2w images demonstrated that the hydrophone was able to detect differences in acoustic emissions in sonications producing different bioeffects. The results show promise for real-time monitoring of MRI-guided transcranial therapy.

Physicians are increasingly confronted with non-palpable breast lesions only visible on MRI. This study examined the visibility and palpability of focused ultrasound lesions in fatty human breast tissue. Eighteen sonications were made around the perimeter of an arbitrary prescribed “tumor” square, representing a non-palpable tumor area. Potential stiffness changes were measured using MR-ARFI showing the displacement difference between the pre and post sonication. The lesions were fully registered with images, circumscribing a tumor area in a human cadaver breast, and thus, providing a visible and palpable perimeter for a surgeon as a guide for excision during breast conservation surgery.

Focused ultrasound (FUS) along with an ultrasound contrast agent (UCA) can induce transient and local increase in the permeability of blood vessel wall or cell membrane, and the change in blood-brain barrier (BBB) permeability can be appropriately indicated by contrast-enhanced MRI. Recently, most studies have used optimum FUS parameters with intravascular injection of pre-formed micro-bubbles to produce BBB disruption with minimum damage to the neurons. However, there are no studies reporting that under biosafety regime BBB disruption could still be predicted by MR contrast enhancement. The purpose of this study was to see if the traditional T1-weighted (T1W) imaging sequences, spin echo (SE) and gradient echo (GE), can discern the difference in the BBB disruption in lower dose regime or not. A high sensitivity R1 mapping was used as a gold standard and absolutely quantification. The quantitative analysis indexing the degree of BBB disruption and the correlation against Evans blue (EB) staining were also demonstrated. Our results suggest that, in the absence of hemorrhage, contrast-enhanced T1W gradient echo and spin echo sequence were equally reliable in quantifying the BBB disruption.

MR thermometry relies on the proton resonance frequency shift with temperature, which can produce off-resonance artifacts in EPI and spiral sequences. This work analyzed the appearance of the focused ultrasound (FUS) hotspot on several EPI and Spiral trajectory designs through simulations and FUS experiments. The distortion of the FUS spot with single shot EPI or Spiral imaging can be severe for the high temperature changes used in ablation, and may lead to under-estimation of the peak temperature. Multi-shot sequences can be used to reduce the shifts/distortion to a tolerable level.

MRgHIFU is a hybrid technology which aims to offer efficient and safe thermal ablation of targeted tumors or other pathological tissues, while preserving the normal surrounding structures unaltered. Theoretically MRgHIFU has no limitation on lesion size [1]. The main challenge is to avoid near and far field heating [2]. We demonstrate here that beam reflection on bones is a major problem whenever bone is situated in the proximity of the prescribed region for sonication, even laterally from the main beam axis. This study evaluates selective de-activation of phased-array transducer’s elements as a potential strategy to reduce bone reflection.

One challenge in MRgHIFU is to provide safe and thermally neutral focusing of HIFU beam pattern using acoustic radiation force imaging (ARFI). The radiation force is localized and highly directional (along the main propagation axis of the HIFU beam) while negligible outside the focal zone. This force initiates a tissue displacement correlated to the amplitude of the acoustic field and thus a phase shift that can be encoded in the MR signal using a motion encoding gradient (MEG) [1]. In addition, ARFI also provide ‘stiffness weighted’ images that may allow one to assess for pre- versus post- therapy changes in tissue. Since HIFU also causes tissue heating, temperature elevation and RFI effects are always associated, at various degree. We propose here to obtain a precise localization of the HIFU focal point by subtracting GRE phase images from two independent acquisitions, where ARF-induced phase shift is sequentially encoded with positive and, respectively, negative monopolar MEG pulse. For illustration, the MEG was implemented here along the slice-select direction.

To improve the accuracy in temperature measurements over a wide temperature region (20 – 95 C), we designed and fabricated a test sample holder and conducted temperature measurements over the temperature range. The test sample holder comprised a reference chamber for temperature reference and a heating chamber. Both chambers, filled with water, were in well thermal insulation. Nonlinear relationship between proton resonance frequency shift and temperature was observed for the wide temperature region. Accurate information of temperature variations over a wide temperature region would be valuable to thermal therapy for a temperature region that could reach the water boiling temperature.

Temperature sensitive liposomes (TSL) incorporating both a chemotherapeutic drug, i.e. doxorubicin, and a clinically approved MRI contrast agent, [Gd(HPDO3A)(H2O)] were prepared and evaluated for MR image guided drug delivery. A gel phantom was prepared containing spots of agarose gel mixed with the liposomes. Before and after heating with High Intensity Focused Ultrasound (HIFU), a T1 map was obtained with a Look-Locker EPI-sequence. When heated above the phase transition temperature, the TSLs showed a rapid release of both the drug and contrast agent. The spots with liposomes which were heated with HIFU clearly showed a lower T1 after ultrasound application.

Dynamic beam-steering of high intensity focused ultrasound (HIFU) based on MR-guidance is a promising technology for the non-invasive ablation of pathological tissue in abdominal organs such as liver and kidney. A particular problem of this technique remains the intrinsic latency between the position measurement and the beam update, which leads to undesired energy dispersion and potentially to the destruction of non-pathological tissue. In this study, dynamic beam-steering using a robust Kalman-predictor for 3D motion anticipation is evaluated experimentally.

For certain MR thermometry applications, such as tissue property determination or total accumulated thermal dose calculations, retrospectively reconstructed temperature maps are acceptable. For such purposes, we have implemented a temporally constrained reconstruction method. The technique uses the entire dynamic imaging data set and an iterative cost function minimization algorithm to create 3-D temperature maps with high spatial resolution (~1 - 2mm3), high temporal resolution (~1 sec), and large field of view coverage (~26x16x3cm3). We present the TCR method and applications to retrospective determination of tissue thermal conductivity, ultrasound power deposition, and total accumulated thermal dose.

During HIFU treatment, the focus of ultrasound (US) is arranged to the targeting region determined in advance. In practical treatments, however, the focus might be deviated due to the complex path (tissue-bone interface or tissue-air interface) of US beams. For safety considerations, accurate localization of heating focus is important before performing HIFU treatment. In this study, a spherical model is proposed to estimate the real position of US focus. A low power pre-treatment experiment was performed on ex vivo porcine muscle. The estimated focus position was verified via magnetization transfer ratio images after a high power HIFU transmission.

An inverse parameter estimation technique that non-invasively determines ultrasound tissue properties ( speed of sound, attenuation) using MRI temperature maps of low level ultrasound heating pulses is presented. The properties determined by the new technique are compared to ultrasound tissue properties measured using the transmission-substitution technique.

The purpose of this study was to investigate the effects of unfocused ultrasound on blood-brain barrier opening across the whole brain using contract enhanced-MRI. T1-weighted FSE images of the brain were acquired in rats for several minutes after gadolinium administration and unfocused ultrasound whole brain treatment. Signal increased immediately after sonication, and continued to increase in the brain as time passed, while muscle signal decreased due to washout. Our findings demonstrate that unfocused ultrasound sonication can disrupt the blood-brain barrier across the whole brain, including cortex and deep grey matter nuclei. This can be observed using contrast-enhanced MRI.

In this study, an imaging sequence, which simultaneously monitors temperature change and magnetization transfer (MT) contrast at 2-sec temporal resolution, was applied on rabbit thigh muscle during HIFU sonicaiton to verify in vivo feasibility. The characteristics of better immunity to phase variance (in contrast to temperature mapping derived from phase images) and clear distinction between heated spot (4.29%¡Ó0.41%) and non-heated region (-0.19%¡Ó0.30%) of MT, even after turning off HIFU pulse, suggest its usefulness in long-term monitoring. In conclusion, MRI with simultaneous temperature and MT mapping is an effective technique to evaluate tissue damage for HIFU treatment.

An MR-compatible system was developed for performing focused-ultrasound exposures in preclinical models. A focused-ultrasound transducer attaches to the positioning system and is submerged within a closed water tank. Sonicating a phantom and measuring the thermal focal zone registers ultrasound and MRI coordinates. For each axis, a 5 cm travel and 0.1 mm positioning resolution was achieved. The system was constructed with non-magnetic components and operation of the focused-ultrasound system within the bore during imaging did not result in any mutual interference. This system is used to study the applications of ultrasound energy for novel therapeutic applications in preclinical animal models.

MRI is a well-suited candidate for temperature monitoring during the heating with transcranial HIFU. For this application, the body coil is usually used because of the constraints due to the large sized of the HIFU system and the stereotactic frame surrounding the patient head. This study showed the improvement of image quality, and therefore temperature sensitivity, by using a dual Flex-coil arrangement. Further improvement is possible by designing dedicated coil arrays with a larger number of coil elements and integrated EMI filters within the coil architecture to reject any interference of the HIFU shots with the MR signal.

A novel prototype for brain therapy with transcranial focused ultrasound is presented here. The first part of this study showed that this new HIFU system was fully MR-compatible. Secondly, we optimized a sequence for MR thermometry, and followed the increase in temperature in a gel heated with increasing power (from 125 to 500 Wac). Finally, we showed it is possible to heat veal brains through a human skull at a high frequency and monitor the heating process with MRI. After validation on cadaver heads, this work will open new horizons to tumor brain therapy in animals and then in humans.

The Pennes' perfusion term in the Pennes bioheat transfer equation depicts the rate at which blood flow removes heat from a point and can play an important role in tissue heat dissipation. Because tissue perfusion is known to change over the course of a thermal therapy treatment, the ability to perform multiple flow assessments to detect perfusion changes during magnetic resonance-guided high-intensity focused ultrasound treatment is of high importance. In this work, we present a method to use arterial spin labeling to determine the Pennes' perfusion term in brain tissue and evaluate performance as a function of various imaging parameters, such as flip angle , bandwidth, and resolution. The results indicate that the proposed technique could be applied in MRgHIFU to provide an efficient estimate of the Pennes' perfusion term. Although demonstrated on brain tissue, this technique could be applied to other tissue types.