Skip to main content

5th International Symposium on Focused Ultrasound

North Bethesda, MD, USA. 28 August- 1 September 2016

A1 Treatment of essential tremor and Parkinson’s disease tremor by MRI guided Focused Ultrasound: a report of 38 consecutive cases in a single center

Menashe Zaaroor, Alon Sinai, Dorit Goldsher, Ayelet Eran, Maria Nassar, Ilana Schlesinger

Rambam Health Care Campus, Haifa, Israel


Thalamotomy of the ventral intermediate nucleus (VIM) is effective in alleviating medication resistant tremor in patients with essential tremor (ET) and Parkinson’s disease (PD). MRI guided Focused Ultrasound (MRgFUS) is an innovative technology that enables non-invasive thalamotomy via thermal ablation.


Thirty eight ET and PD patients with severe medication resistant tremor underwent MRgFUS underwent unilateral VIM thalamotomy using MRgFUS. Effect was evaluated using clinical Rating Scale of Tremor (CRST) in ET patients and Unified PD Rating Scale motor part (UPDRS) in PD patients. Quality of life was assessed by Quality of life in ET Questionnaire (QUEST) and PD Questionaire (PDQ-39).


Tremor stopped in the treated hand in 37 patients immediately following the treatment. In one patients tremor was modified but not abolished. At one month post-treatment, the ET patients’ CRST score decreased from 38.6 ± 12.0 to 9.3 ± 7.7 (p < 0.001) and QUEST scores decreased from 44.8 ± 17.8 to 13.1 ± 15.9 (p < 0.001). In PD patients UPDRS-motor part decreased from 26.2 ± 8.7 to 16.3 ± 11.0 (p = 0.0087) and PDQ39 decreased from 40.8 ± 18.2 to 26.5 ± 15.1 (p = 0.027). During follow up of 1-24 months (mean 10.9 ± 8.1 months) tremor reappeared in seven of the patients, but in all but three, to a lesser degree than before the procedure.

Adverse events that transiently occurred during sonication included: Headache (n = 11), short lasting vertigo (n = 17) and dizziness (n = 4), nausea (n = 4), burning scalp sensation (n = 3), vomiting (n = 3) and lip paresthesia (n = 2). Adverse events that lasted after the procedure included gait ataxia (n = 5), unsteady feeling when walking (n = 4,) unilateral taste disturbances (n = 3) and hand ataxia (n = 3). All adverse events were transient and none lasted beyond 3 months.


MRgFUS VIM thalamotomy to relieve medication resistant tremor was safe and effective in ET, and PD. Current results emphasize its low adverse events profile and high efficacy in treating tremor. Large randomized studies are needed to assess prolonged efficacy and safety.

A2 Focused Ultrasound likely dominates deep brain stimulation and stereotactic radiosurgery for medically-refractory essential tremor: an initial decision and cost-effectiveness analysis

Jonathon Parker1, Vinod Ravikumar1, Pejman Ghanouni1, Sherman Stein2, Casey Halpern1

1Stanford University, Stanford, California, USA; 2University of Pennsylvania, Philadelphia, Pennsylvania, USA


Essential Tremor (ET) is one of the most common neurologic conditions, and conservative measures are frequently suboptimal. Recent data from a multi-institution, randomized controlled clinical trial demonstrated that Magnetic Resonance-guided Focused Ultrasound (MRgFUS) thalamotomy improves upper limb tremor in medically refractory ET. This study assesses the cost-effectiveness of this novel therapy in comparison to existing procedural options.


PubMed and Cochrane Library searches were performed for studies of MRgFUS, Deep Brain Stimulation (DBS), and Stereotactic Radiosurgery (SRS) for ET. Pre- and post-operative tremor-related disability scores were collected from 32 studies involving 83 MRgFUS, 615 DBS, and 260 SRS cases. Utility (defined as percent change in functional disability) was calculated, and Medicare reimbursements were collected as a proxy for societal cost – costs of MRgFUS for ET were derived from a combination of available costs of approved indications and SRS costs where appropriate. A decision and cost-effectiveness analysis was then constructed, implementing meta-analytic techniques.


MRgFUS thalamotomy resulted in significantly higher utility scores compared with DBS and SRS based on estimates of Medicare reimbursement (p < 0.001). MRgFUS was also the most inexpensive procedure out of the three (p < 0.001).


Preliminary experience with MRgFUS for ET suggests that this novel therapeutic may be more effective than available alternatives and potentially less costly for society. It thus will likely “dominate” DBS and SRS as a more cost-effective option for medically refractory ET. Our findings support further investigation of MRgFUS for ET and broad adoption.

A3 Tractography-based VIM identification for Focused Ultrasound thalamotomy: initial results

Vibhor Krishna, Amelia Hargrove, Punit Agrawal, Barbara Changizi, Eric Bourekas, Michael Knopp, Ali Rezai

The Ohio State University, Columbus, Ohio, USA


The ventral intermediate nucleus (VIM) is not visible on conventional Magnetic Resonance Imaging (MRI). A novel method for tractography-based VIM identification has recently been described. We report the short-term clinical results of prospective VIM targeting with tractography in a cohort of patients undergoing Focused Ultrasound thalamotomy.


All patients underwent structural and diffusion weighted imaging (60 diffusion directions, 2 mm isovoxel) with 3 Tesla MRI scanner (Philips Ingenia CX). The images were processed using streamline tractography (Stealth Viz, Medtronic Inc.). The lateral and posterior borders of VIM were defined by tracking the pyramidal tract and medial lemniscus respectively. A VIM region of interest (ROI) was placed 3 mm away from these borders (Figs. 1, 2 and 3). The structural connectivity of this VIM ROI was confirmed to the motor cortex (M1) and cerebellum. The coordinates of tractography-based VIM in relation to posterior commissure were noted for surgical targeting. The parameters analyzed include a clinical tremor scale (pre-, intraoperative, and post operative), operative time, and number of sonications.


Tractography-based VIM targeting was successful in 7 out of 8 patients. The coordinates of tractography-based VIM were significantly different from the standard coordinates (3-D distance 3.9 ± 2.4 mm). Therapeutic sonication (>55 °C temperature, 10 seconds) at the tractography target resulted in >50 % tremor improvement with intraoperative objective tremor assessment without any motor or sensory side-effects. The mean operative time was 78 ± 3.3 minutes with 12.8 ± 3.9 average sonications. Overall the tremor scores significantly improved one month after surgery (preop CRST total 62.1 ± 15.5 versus 30.3 ± 14.1, two tailed t-test p = 0.006). None of the patients experienced sensory deficits or motor weakness during follow-up.


We report that prospective tractography-based VIM targeting is safe and feasible. The short-term clinical results are satisfactory. Long-term tremor efficacy outcomes are desirable to further assess the usefulness of this technique.

Fig. 1 (abstract A3).
figure 1

Axial T1 projection showing the relation of VIM target 3 mm medial and anterior to pyramidal tract and medial lemniscus respectively

Fig. 2 (abstract A3).
figure 2

Postoperative sagittal T1 projection demonstrating the relationship between pyramidal tract and medial lemniscus in relation to thalamotomy lesion

Fig. 3 (abstract A3).
figure 3

Postoperative axial T1 projection demonstrating the relationship between pyramidal tract and medial lemniscus in relation to thalamotomy lesion

A4 Targeted delivery of brain-penetrating non-viral GDNF gene vectors to the striatum with MRI-guided Focused Ultrasound reverses neurodegeneration in a Parkinson’s disease model

Brian Mead1, Namho Kim2, Panagiotis Mastorakos2, Jung Soo Suk2, Wilson Miller1, Alexander Klibanov1, Justin Hanes2, Richard Price1

1University of Virginia, Charlottesville, Virginia, USA; 2Center for Nanomedicine/Wilmer Eye Institute, Johns Hopkins University, Baltimore, Maryland, USA


Parkinson’s disease (PD) is characterized by the degeneration of dopaminergic neurons in the motor control pathways of the brain. Gene therapy using glial cell derived neurotrophic factor (GDNF) has shown some limited promise for treating PD; however, we hypothesize that outcomes could be further improved by enhancing gene vector distribution. We previously developed a gene therapy approach that entails delivering systemically administered non-viral gene-bearing nanoparticles (BPN) across the Blood-Brain Barrier with MRI-guided Focused Ultrasound (FUS). BPN rapidly penetrate brain tissue due to a dense coat of polyethylene glycol, and this approach mediates efficient and localized transgene expression in the brain of healthy rats. Here, we tested whether the FUS-mediated delivery of GDNF plasmid-bearing BPN (GDNF-BPN) reverses neurodegeneration in the rat 6-OHDA PD model.


6-OHDA rats were ultrasonically coupled to a 1.15 MHz MRI-compatible FUS transducer. T2 and T2* pre-treatment scans were obtained to allow FUS targeting of striatum. Microbubbles (2x105/g) and 100 μg of ~50 nm non-viral GDNF plasmid-bearing BPN (polyethylene glycol/polyethylenimine) were co-injected i.v. and FUS was applied at 0.6 MPa, with a 0.5 % duty cycle, for 2 min. Contrast T1 and T2* images allowed semi-real time confirmation of BBB disruption and safety, respectively. Efficacy was assessed using an ELISA for GDNF, tyrosine hydroxylase (TH) and VMAT2 immunolabeling for neural degeneration, HPLC for dopamine, and behavioral analysis (i.e. apomorphine-induced rotational asymmetry and forepaw use bias in 6-OHDA rats).


Striatum-targeted delivery of GDNF plasmid-bearing BPN with FUS led to an ~80 % reduction in apomorphine-induced rotational asymmetry, eliminated forepaw use bias (Fig. 4a,b), and fully restored TH+ dopaminergic neuron density in both the substantia nigra pars compacta (SNpc) and striatum compared to untreated 6-OHDA rats (Fig. 4c,d). T2* MRI confirmed safety of the BBB opening approach.


FUS-mediated delivery of systemically circulating non-viral GDNF-BPN to the striatum of 6-OHDA rats confers a significant behavioral benefit as well as a restoration of TH+ cell number in the nigrostriatal pathway, indicating cessation and/or reversal of neurodegeneration. Our studies indicate that delivery of GDNF-BPN with FUS may provide a powerful, non-invasive and highly tailorable gene therapy approach to slow or stop the neurodegenerative process in PD.

Fig. 4 (abstract A4).
figure 4

Graphs of rotational bias (a) or forepaw use bias (b) following 6-OHDA injection. (c) Representative images of TH-immunlolabeled sections through the SNpc. Graphs represent TH+ cell number in SNpc (d) and staining intensity in the striatum (e). * p < 0.05

A5 Focused ultrasound facilitated gene delivery for neuro-restoration in Parkinson’s disease mice

Shutao Wang, Oluyemi Olumolade, Tara Kugelman, Vernice Jackson-Lewis, Maria Eleni (Marilena) Karakatsani, Yang Han, Serge Przedborski, Elisa Konofagou

Columbia University, New York, New York, USA


Not released for publication


Not released for publication


Not released for publication


Not released for publication

A6 MRI-g-FUS for the treatment of Alzheimer’s disease

Kullervo Hynynen1, Isabelle Aubert2

1Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; 2Sunnybrook Research Institute, Toronto, Ontario, Canada


Over the past ten years consistent effort has been put forward at the University of Toronto to develop Focused Ultrasound methods for the treatment of AD. This talk will review the progress made so far.


The first studies demonstrated safe antibody delivery in AD mouse model with significant reduction in the plaque load. A follow up studies with two-photon microscopy showed that blood vessels with plague deposits showed a different type of opening than vessels in normal brain but large molecule delivery into the brain was still possible in these animals. Another study demonstrated that plaque reduction can be achieved by just opening the BBB with microbubbles.


The histology revealed stimulation of neurogenesis. Multiple treatments of old mice resulted in memory rescue without any observable side-effects. A follow up study demonstrated that this neurogenesis was not induced with exposures that did not cause observable BBB opening even with the presence of the microbubbles. An ongoing study in large animals has shown that half-brain BBB opening can be safely and repeatable performed indicating the feasibility of clinical translation.

A7 Scanning Focused Ultrasound disruption of the blood-brain barrier as an Alzheimer’s disease therapy

Gerhard Leinenga1, Rebecca Nisbet2, Robert Hatch1, Anneke Van der Jeugd3, Harrison Evans2, Jürgen Götz2

1Clem Jones Centre for Ageing Dementia Research, Queensland Brain Institute, The University of Queensland, St Lucia, Queensland, Australia; 2The University of Queensland, Brisbane (St Lucia Campus), QLD, Australia; 3University of Leuven, Leuven, Belgium, Belgium


Alzheimer’s disease is the most common form of dementia. Pathological abnormalities in the Alzheimer’s disease brain includes the presence of amyloid-beta plaques, hyperphosphorylation and intracellular aggregation of tau and synaptic degeneration. Focused ultrasound combined with intravenous injection of microbubbles has been shown to reversibly open the blood-brain-barrier (BBB). By moving the focus in scanning mode we are able to open the BBB throughout the brain of a mouse. Here we tested the effects of repeated scanning ultrasound (SUS) in APP23 amyloid plaque-bearing mice, pR5 tau mice and wild-type mice to determine the effects of SUS on amyloid, tau and dendritic spines.


The device used was the Therapy Imaging Probe System (TIPS, Philips Research), which has an eight-element annular array transducer with a focal length of 80 mm a radius of curvature of 80 mm, a 33 mm central opening, and a motorized 3D positioning system. The focus 6 dB size was 1.5 mm x 1.5 mm x 12 mm at 1 MHz. Settings that were applied were 1 MHz centre frequency, 0.7 MPa peak rarefactional pressure applied outside the skull, 10 Hz pulse repetition frequency, 10 ms pulse length and 10 % duty cycle immediately after retroorbital injection of in-house made microbubbles. APP23 mice that accumulate amyloid beta, pR5 mice that overexpress FTD-mutant tau, and wild-type C57Bl/6 mice were treated weekly by scanning ultrasound (SUS) for periods of 4 to 7 weeks.


In APP23 mice we used repeated scanning ultrasound (SUS) treatments of the mouse brain to remove amyloid-beta. Spinning disk confocal microscopy revealed extensive internalization of Abeta into the lysosomes of activated microglia in mouse brains subjected to SUS. Plaque burden was reduced in SUS-treated AD mice compared to sham-treated animals. Treated AD mice also displayed improved performance on three memory tasks.

In PR5 mice we investigated the efficacy of a novel tau isoform-specific single chain antibody fragment, RNX, delivered by passive immunization in the P301L human tau transgenic pR5 mouse model. When administration of RNX was combined with scanning ultrasound (SUS), RNX delivery into the brain and uptake by neurons were markedly increased, as were reductions in tau phosphorylation and anxiety-like behavior.

In wild-type mice we investigated the effects of SUS on neuronal excitability and morphology. We performed patch-clamp recordings from hippocampal CA1 pyramidal neurons in wild-type mice 2 and 24 hours after a single SUS treatment, and one-week and three months after six weekly SUS treatments. No change in CA1 neuronal excitability was observed compared to sham-treated neurons at any time-point. Multiple SUS treatments had the effect of preventing the loss of CA1 synapses that occurred in sham-treated neurons.


We show that scanning Focused Ultrasound disruption of the BBB has multiple biological effects in the brain which make it an attractive candidate for an Alzheimer disease therapy. SUS reduced plaque burden and amyloid-beta levels in APP23 amyloid mice, through activation of microglia, and improved performance on tests of memory function. In pR5 tau mice SUS alone reduced hyperphosphorylation of tau, and enhanced the delivery of anti-tau antibodies resulting in improved reductions in pathology and behavioral abnormalities. In wild-type mice SUS was shown to have no effect on the firing of hippocampal neurons or their morphology, but prevented spine loss at 3 months after six weekly SUS treatments. If these effects on Abeta and tau pathology, and dendritic morphology are recapitulated in human patients SUS may emerge as a promising AD therapy.

A8 Scanning ultrasound as a treatment tool of proteinopathies including Alzheimer’s disease

Jürgen Götz1, Rebecca Nisbet1, Ann Van der Jeugd1, Harrison Evans1, Gerhard Leinenga2

1The University of Queensland, Brisbane (St Lucia Campus), QLD, Australia; 2Clem Jones Centre for Ageing Dementia Research, Queensland Brain Institute, The University of Queensland, St Lucia, Queensland, Australia


Neurological disorders constitute a substantial social and economic burden, as they cause considerable ill health but few direct deaths. Treatment strategies for neurodegenerative diseases are hampered by the fact that the Blood-Brain Barrier (BBB) establishes an efficient barrier for therapeutic agents (Leinenga et al., Nature Reviews Neurology 2016). We have recently shown that scanning ultrasound (SUS) allows microglial-mediated clearance of extracellularly deposited amyloid-beta in APP mutant APP23 mice and restores memory functions in three cognitive tests to wild-type levels, in the absence of overt damage to the brain (Leinenga and Götz, Science Translational Medicine 2015) However, it had not been determined whether SUS treatment reduces the intracellular tau pathology that together with amyloid deposition characterizes Alzheimer’s disease.


We investigated the efficacy of a novel tau-specific single chain antibody fragment, delivered by passive immunization in the human tau transgenic pR5 mouse model, a model of the tau pathology of Alzheimer’s disease (Götz et al., Science 2001). To further assess the efficacy and drug-delivering ability of SUS, we established four experimental groups, using the novel anti-tau antibody that was injected weekly over four weeks, either on its own, or together with SUS. A third group used SUS only, and a fourth was the anaesthesia control group. The mice were analysed on the elevated plus maze, histologically and biochemically. Furthermore, uptake of the antibody by the brain was determine using fluorescently labelled single-chain antibody fragments.


A histological and biochemical analysis of the pR5 tau transgenic mice revealed that SUS as well as the employed antibody ameliorated the tau pathology that characterizes the pR5 mice. In addition, the anxiety-like behaviour that characterizes pR5 mice was significantly reduced. We furthermore found enhanced delivery of the antibody using SUS yielding a synergistic therapeutic effect as determined by histology and using the elevated plus maze.


Our study suggests that SUS is a method that benefits diseases with protein aggregates more generally, whether they are intra- or extracellular. The therapeutic delivery combined with SUS could offer significant clinical benefits for the treatment of patients with Alzheimer’s disease and related tauopathies. Considering that the yearly costs of passive immunotherapy for AD is expected to exceed $25,000 per patient, combining SUS with antibody delivery could drastically reduce these costs.

A9 Enhancement of FUS mediated delivery of stem cells to the brain

Paul Fishman1, Paul Yarowsky2, Victor Frenkel3, Shen Wei-Bin3, Ben Nguyen3

1University of Maryland School of Medicine/Baltimore VAMC, Baltimore, Maryland, USA; 2Research & Development Service, VA Maryland Healthcare System and University of Maryland School of Medicine, Baltimore, Maryland, USA; 3University of Maryland School of Medicine, Baltimore, Maryland, USA


FUS mediated Blood-Brain Barrier disruption (BBBD) can enable even large therapeutics such as stem cells to enter brain from the bloodstream and could be a major advance in cell delivery over current invasive methods of brain injection. The efficiency of cellular entry after FUS mediated BBBD alone however is low. We hypothesized that this process could be enhanced by combining it with a complementary strategy termed magnetic targeting. Stem cells can be safely loaded with super-paramagnetic iron oxide nanoparticles (SPION) in culture, allowing cells to be attracted by an external magnet. Our previous study showed SPION loaded stem cells to have enhanced brain retention near a magnet on the skull in a rat model of traumatic brain injury, where BBBD also occurs. The goal of our current project was to determine if magnetic attraction of SPION loaded stem cells would also enhance their delivery to brain after FUS mediated BBBD.


With a small animal MRI guided FUS device (Image Guided Therapy, IGT and 7 T Bruker MRI), we sonicated young adult rats (~120 g) with both radiologic (enhancement of the target region with gadolinium on post-sonication TI MRI), and histologic (staining with Evans’ blue dye) evidence of BBBD, without tissue damage or hemorrhage. Confirmation of the cells within brain as those injected was performed by staining with Perl’s reagent for iron and by immuno-histochemistry with a human specific antigen. The procedure was then combined with the application of a powerful magnet to the head directly after IV injection of hNPCs.


With BBBD alone human neuro-progenitor cells (hNPCs) loaded with SPION were observed in rat brain after intravenous (IV) injection directly after sonication only within the treated regions. To demonstrate the effect of magnetic attraction, we injected equal numbers of SPION and non-SPION labeled cells, where each cell type was labeled with a different fluorophore. In animals that had FUS mediated BBBD followed by a magnet applied to the head, significantly greater numbers of SPION labeled cells were observed compared to the non-labeled cells. This result was most pronounced in regions of the brain close to the skull (cerebral cortex) and magnet surface. More powerful magnets including magnetic arrays resulted in more effective retention of SPION labeled cells in even deeper brain regions such as the striatum. There, 90 % of hNPCs observed contained SPIONs compared to 60-70 % with a less powerful magnet.


These results demonstrate that the use of magnetic attraction can substantially enhance delivery of stem cells after BBBD. In prior published work, stem cells were delivered to brain after FUS mediated BBBD using cells injected directly into the carotid artery. In an effort to accomplish this goal in a safer and less invasive manner, our study utilized IV cell injection (tail vein), supporting the view that the combination of FUS mediated BBBD and magnetic attraction can allow stem cells to enter brain with a minimally invasive strategy.

A10 Fluorescent lipid microbubbles for targeted brain drug delivery through the Focused Ultrasound-induced blood-brain barrier opening in vivo

Carlos Sierra Sanchez, Camilo Acosta, Cherry Chen, Shih-Ying Wu, Maria Eleni (Marilena) Karakatsani, Elisa Konofagou

Columbia University, New York, New York, USA


Focused ultrasound (FUS) in the presence of lipid microbubbles can induce non-invasive, transient and reversible Blood-Brain Barrier (BBB) opening. This study entailed assessment of the feasibility of fluorescently loaded microbubbles, labeled with the fluorophore 5-dodecanoylaminfluorescein (C-12), as a vector for targeted brain drug delivery. Compared to prior studies by our group, where fluorescently-labeled dextrans were co-administered with microbubbles, this new methodology improves the safety and allows a more targeted drug delivery with potentially lower toxicity, avoiding systemic exposure. The main objective was thus to determine feasibility and safety of using the loaded microbubbles as carriers towards targeted brain drug delivery with simultaneous cavitation monitoring.


A spherical, single-element, FUS transducer (center frequency 1.5 MHz) was used. A pulse-echo transducer (center frequency 10 MHz), confocally mounted at the center of the FUS transducer, was utilized for passive cavitation detection (PCD). FUS (pulse length 10,000 cycles; pulse repetition frequency 5 Hz; duration 5 minutes; acoustic pressure 450-750 kPa) targeted mouse brains in vivo, in combination with fluorescent microbubbles for C-12 delivery, which was evaluated by in vivo transcranial PCD, through the quantification of inertial (ICD) and stable harmonic (SCDh) and ultraharmonic (SCDu) cavitation doses at 30, 60 and 300 s; together with ex vivo fluorescence imaging. The BBB opening was verified using in vivo T1-w Magnetic Resonance Imaging (MRI). The safety of this technique was assessed through ex vivo hematoxylin & eosin staining for microhemorrhage detection and immunohistochemistry (Iba-1 for microglial activation) together with in vivo T2-w MRI for edema assessment.


Successful targeted C-12 delivery was achieved at 600 and 750 kPa in six out of 14 cases (Fig. 5). Comparison of ICD, SCDh and SCDu between successful and unsuccessful cases yielded a statistically significant linear relationship between the successful targeted drug delivery and CD and specific thresholds for efficient delivery were identified.

No edema was detected in mice sacrificed on Day 0 but edema appeared on Day 1 on mice sacrificed on Day 7. In all cases cases (except one) it was repaired within a week. Microhemorrhages were observed after sonication in some cases but were also cleared within the first week. However, a higher number of cell nuclei was observed in the sonicated region compared to the unsonicated side in some mice survived up to one week after opening. Iba-1 immunohistochemistry also showed microglial activation.


FUS was applied in conjunction with fluorescent microbubbles and, for the first time, the existence of CD thresholds for assessing successful drug delivery was defined. For CD above these thresholds, significant fluorescent enhancement was observed, demonstrating C-12 targeted delivery. One week after sonication, edema was cleared out but microglial activation was observed in certain cases. Therefore, this study indicates the feasibility and safety of a new methodology of FUS-induced BBB opening for targeted albeit potentially riskier brain drug delivery and provides a platform for predicting successful delivery via PCD.

Fig. 5 (abstract A10).
figure 5

BBB opening and fluorescence delivery: T1-w MRI showing BBB opening at pressures (a) 450, (b) 600 and (c) 750 kPa. Fluorescence delivery (green) and DAPI (blue) in two horizontal sections of mouse brains sonicated at (d) 600 and (e) 750 kPa

A11 Ultrasound-mediated delivery of gadolinium and fluorescent–labelled liposomes through the Blood-Brain Barrier

Muna Aryal1,2, Iason T Papademetriou3, Yong-Zhi Zhang2, Chanikarn Power1,2, Nathan McDannold1,2, Tyrone Porter3

1Brigham and Women’s Hospital, Boston, Massachusetts, USA; 2Harvard Medical School, Boston, Massachusetts, USA; 3Boston University, Boston, Massachusetts, USA


The main objectives of this study were: 1) to examine whether gadolinium and fluorescent labelled liposomes can extravasate into the brain parenchyma after ultrasound mediated Blood-Brain Barrier disruptions; and 2) to test whether extravasated liposomes were size dependent or not. The liposomes were labelled with gadolinium (Gd) and fluorophore, thus enabling detection of extravasated liposomes via MRI in vivo and fluorescence methods in tissue, respectively.


Liposomes labelled with gadolinium and fluorophore were prepared using lipid film hydration and extrusion to two different sizes; ~70-85 nm and ~ 130-150 nm. Animals were divided into two different groups based on the use of particle sizes; group A (~70-85 nm) and group B (~130-150 nm). Focused ultrasound mediated Blood-Brain Barrier disruption (BBBD) was produced in one hemisphere in 15 mice. Particles were injected before sonication. Sonications (0.69 MHz at 0.42 MPa) were performed in two locations combine with Definity (10 μl/kg). Acoustic emissions were recorded during FUS. T1-weighted contrast enhanced and T2*-weighted MRI were used to confirm Gd leakage and damage detection respectively. Mice were euthanized 5-24 hours after FUS and post-process for fluorescence measurement.


In T1-weighted contrast enhanced MRI, gadolinium-leakage was able to detect on sonicated area at 5-24 after FUS but not on non-sonicated area (control). Detection of fluorescence signal from brain tissue homogenates confirm the liposomal particles extravasation on sonicated locations. On group A, gadolinium and fluorescence signal intensities on sonicated locations were increased by 26 % and 62 % respectively as compared with control and signal enhancement were statistically significant compared with control ( p = 0.017 and p = 0.02 respectively, two-tailed, paired ttest). On group B, gadolinium and fluorescence signal intensities on sonicated locations were increased by 24 % and 40 % respectively as compared with control. Comparison of fluorescence signal intensities between two groups on sonicated location was statically significant whereas it was not significant on controls (p < 0.05 and p = 0.07 respectively, one-tailed unpaired ttest).


Overall, this work demonstrates that ultrasound can deliver upto ~ 150 nm liposomes that labelled with gadolinium and fluorophore through the Blood-Brain Barrier. The results indicate that the extravasation of liposomes were size dependent.

A12 Sterile inflammatory response (SIR) in the brain following exposure to low intensity pulsed Focused Ultrasound and microbubble infusion

Zsofia Kovacs1, Saejeong Kim1, Neekita Jikaria1, Farhan Qureshi2, Michele Bresler2, Joseph Frank2

1National Institutes of Health, Bethesda, Maryland, USA; 2National Institutes of Health Clinical Center, Bethesda, Maryland, USA


Magnetic Resonance Imaging (MRI)-guided pulsed Focused Ultrasound (pFUS) in combination with systemic injection of microbubbles (MB) is being advocated to increase drug or gene delivery by causing localized Blood-Brain Barrier (BBB) disruption (D). The objective of this study is to investigate the molecular and cellular responses following pFUS + MB associated with BBBD in the rat brain.


Female Sprague Dawley rats (<200 g) were sonicated at 0.3 MPa acoustic pressures with 10 ms burst length and 1 % duty cycle (9 focal points, 120 sec/9 focal points) using a single-element spherical FUS transducer (589.636 kHz; FUS Instruments). 100 μl Optison™ MB (GE Healthcare) was administered intravenously. Gadofosveset-enhanced T1w images were obtained with a 3.0 T MRI (Phillips). Proteomic and mRNA expression in the brain following pFUS + MB were analyzed with ELISA, Western blot, quantitative real-time PCR or immunofluorescent staining (Fig. 6a). Proteomics were normalized to sham and statistical analysis was performed by one-way ANOVA corrected for multiple comparisons. Rats were also injected with 8 mg/kg Rhodamine encapsulated magnetic polymers (MicroTRACK™; BioPal) 3 days prior to sonication to label splenic macrophages (CD68) to monitor tropism to the brain.


Post contrast T1w MRI and histology showed open BBB without evidence of microhemorrhage. Within 5 minutes following sonication, increased expression of pro-inflammatory and anti-inflammatory cytokines, chemokines and trophic factors (CCTF) was detected in the parenchyma lasting up to 24 hours (Fig. 6b). Increases in heat shock protein 70 (HSP70), tumor necrosis alpha (TNFa), and interleukin (IL) 1a, 1b and 18 consistent with damage associated molecular patterns (DMAP)1 and activation of nuclear factor kappa-light-chain-enhancer of activated B cells (NFkB) inflammatory pathways were observed with SIR to injury2 (Fig. 6b). NFkB pathway-related gene activation along with anti-apoptotic genes, immune cell chemoattractants, selectins, and cell adhesion molecules showed significant (>2fold) increases in mRNA expression (Fig. 6c). Histological analysis showed significant increases (p < 0.05) in the following: number of TUNEL positive cells within 6 hrs, GFAP and Iba1 staining for activated astrocytes and microglia (1-24 hrs) and increase of ICAM up to 24 hrs post sonication. We also detected a >4 fold greater (p < 0.05) CD68 positive cells on day 6 post sonication containing intracellular fluorescent beads within the pFUS + MB treated hemisphere compared to contralateral hemisphere.


The temporal molecular response to pFUS + MB is indicative of SIR2 originating from the parenchyma. The pattern of cytokines immediately after pFUS + MB is initiated by cellular release of DAMPs and TNFa observed with mild cerebral trauma or ischemia [1,2]. Increases in monocyte chemoattractant protein (MCP-1), vascular endothelial growth factors (VEGF), stromal derived factor 1 (SDF-1), erythropoietin (EPO) and brain derived neurotropic factor (BDNF) are associated with BBBD stimulating angiogenesis, neurogenesis and stem cell migration observed with ischemia and trauma. These results indicate that pFUS + MB rapidly affects the cerebral vasculature as evident by BBBD in addition to the shockwave from MB collapse that induces mild stress within various cellular elements in the parenchyma inducing a SIR.


1. Chen, G. Y., et al. 2010 Nat Rev Immunol 10(12):826-37.

2. Gadani, S. P., et al. 2015 Neuron 87(1):47-62.

Fig. 6 (abstract A12).
figure 6

pFUS + MB elicits a transient microenvironmental response in the brain that reflects a sterile inflammation

A13 Volumetric MR thermometry in a clinical transcranial MR guided Focused Ultrasound system

Henrik Odéen1, George Chiou2,3, John Snell4, Nick Todd2, Bruno Madore2,3, Dennis Parker1

1University of Utah, Salt Lake City, Utah, USA; 2Brigham and Women’s Hospital, Boston, Massachusetts, USA; 3Harvard Medical School, Boston, Massachusetts, USA; 4Focused Ultrasound Foundation, Charlottesville, Virginia, USA


Transcranial MR-guided Focused Ultrasound (tcMRgFUS) applied within a small central brain volume has achieved excellent outcomes in treatment of movement disorders (Elias NEJM 2013). Although transducers used in tcMRgFUS have been designed with large apertures to spread the beam energy over as much skull as possible to reduce skull and cortex heating, currently used MR temperature imaging (MRTI) methods cannot monitor the temperature increase over the entire insonified brain volume. Instead, temperatures are typically measured in one 2D slice, leaving the majority of the insonified brain volume unmonitored. We have previously developed and published methods to achieve fully 3D MRTI covering the entire insonified brain with good spatial and temporal resolution (Todd MRM 2009/2010) but the techniques have not been evaluated on clinical tcMRgFUS systems. In this work-in-progress study we demonstrate the value of volumetric MRTI with two different pulse sequences applied during heating on a clinical tcMRgFUS system.


PRF MRTI was performed with a product 3D gradient recalled echo (GRE) pulse sequence and a custom-implemented 3D GRE segmented echo planar imaging (EPI) pulse sequence on a 3 T MRI (Discovery 750 T, GEMS). tcFUS heatings were performed in an ex vivo human skull filled with tissue-mimicking gel (ATS Laboratories) in a clinical tcMRgFUS system (ExAblate Neuro, Insightec).

Pulse sequences parameters are listed in Table 1. All data were zero filled interpolated (ZFI) to 1-mm isotropic spacing (GRE data additionally ZFI to 0.5-mm spacing). The skull was physically positioned in the FUS system, and the focus electronically steered, to target deep brain-structures located outside the normal treatment envelope. FUS sonications were applied at 940 W for 30/60 s while imaging with the GRE/EPI sequences, respectively. The EPI sequence used an echotrain length of 16 with bi-polar readout gradient, sampling 192 phase encodings in each direction, so that full “positive only” and “negative only” images could be reconstructed.

k-space data were retrospectively down sampled by a factor of R = 4 and 8 (split into multiple time-frames without throwing any data away) for the GRE and EPI experiments, respectively, giving acquisitions times of 3.6 and 7.8 s, and reconstructed with a temporally constrained reconstruction algorithm (Todd MRM 2009).


Figure 7 shows three orthogonal views of temperature maps overlaid on magnitude images in the GRE experiment. Attempting to focus this far outside the normal treatment envelope results in severe near- and far-field heating. Heating on both the cortex and in the far-field along the petrous bone is visible. In Fig. 8 the temperature evolution at the focal spot is compared to that near the petrous bone. The gel near the bone shows a delayed and greater maximum temperature compared to the focus.

Three orthogonal views of the larger FOV in the segmented EPI experiment are shown in Fig. 9. Heating along large parts of the cortex can be seen. In the underlying magnitude images it can be seen that the EPI sequence experiences more artifacts than the GRE sequence.

The focal spot position (evaluated as the temperature center-of-mass) was tracked as a function of time during the heating in the GRE data ZFI to 0.5 mm (data not shown). The focal spot did not experience any shift to within the finer ZFI spacing.


This study shows that volumetric thermometry over a FOV covering the focal spot and the skull base can be achieved with readily available pulse sequences. With custom implemented pulse sequences the fully insonified FOV (from skull cap to skull base) can be covered. Even though the reconstruction is done retrospectively, the described methods are valuable as research tools in e.g. treatment envelope evaluations. By utilizing 3D imaging ZFI can be performed in all directions to minimize partial volume effects, and very accurate dynamic focal spot localization can be performed.

Future studies will compare the accuracy and precision of the described methods to standard 2D MR thermometry. Experiments comparing 3D MRTI with fiber optic probe measurements at potential target positions outside the currently available treatment envelope will also be performed (Monteith JNS 2016).

Table 1 (abstract A13). MR scan parameters. TR – Repetition time, TE – Echo time, FA – Flip angle, BW – Bandwidth (readout), FOV – Field of view, Res – Resolution, Tacq – Acquisition time (before subsampling)
Fig. 7 (abstract A13).
figure 7

Three orthogonal views of GRE temperature maps overlaid on magnitude image

Fig. 8 (abstract A13).
figure 8

Temporal evolution of heating comparing focal spot to far-field next to petrous bone

Fig. 9 (abstract A13).
figure 9

Three orthogonal views of EPI temperature maps overlaid on magnitude image

A14 Multi-echo MR thermometry compared to single echo MR thermometry in the treatment of essential tremor

Kim Butts Pauly, Mike Marx, Pejman Ghanouni

Stanford University, Stanford, California, USA


The choice of receive bandwidth in MR thermometry acquisitions needs to balance two competing choices. Low bandwidths improve SNR, while high bandwidths reduce spatial shift of the hotspot due to the temperature off-resonance. The low bandwidth MR thermometry sequence in use in essential tremor MRgFUS treatments required repeated swapping of phase and frequency directions since the location of the hotspot could only be trusted in the phase encode direction.

A solution is to use multiple high bandwidth acquisitions, which when averaged, regain much of the SNR of the lower bandwidth image. Such a multiecho sequence has recently become available for clinical use. The purpose of this work was to compare the performance of the multi-echo (ME) MR thermometry to the single echo (SE) MR thermometry in clinical treatments of essential tremor.


Fifteen patients were treated for essential tremor, 12 with only single echo thermometry (TE and BW/pix = 12.8 and 44), 2 with only multi-echo thermometry (TE and BW/pix = 3, 8, 13, 18, 22 and 278), and 1 with both sequences. All other image parameters remained the same, including TR = 100. All thermometry images were processed offline in Matlab with a single baseline subtraction (α = -0.00909), followed by referenceless processing for constant and linear terms. The multi-echo thermometry was further processed with a phase unwrapping algorithm in the TE dimension. The multiple echoes were then combined with a weighting by the square of the temperature SNR of each image.


The decrease in sampling time for the ME thermometry dictates a theoretical decrease in temperature SNR of only 11 % due to the decrease in sampling time alone. In the one patient that had both sequences in the same scan planes for two sonications, there was a 9 % decrease in SNR in the ME thermometry as measured in the frame used for the referenceless processing, comparing well with theory.

A review of all sonications with ME thermometry when phase encoding was S/I revealed much reduced artifacts over SE thermometry with phase encoding in the S/I direction. Example images are provided in Fig. 10. The region of interest measurement indicated a 30 % improvement in temperature SNR for the ME over the single echo, due to this artifact reduction.

The spatial shifts should theoretically be reduced by a factor of 6.3 in the ME thermometry, as compared to the SE thermometry, due to the increase in receive BW. For example, a 23 °C temperature rise would result in a 0.6 mm shift with the single echo, and only a 0.1 mm shift with the multi-echo.


When phase encoding is S/I, multi-echo thermometry is superior to single echo thermometry due to a reduction in ghosting artifacts and spatial shifts. Alignment of the focal spot should not require swapping phase and frequency directions. Future work will include a prospective study to verify this. In the other directions, ME thermometry is comparable to SE thermometry with the reduction of spatial shifts coming with a loss of temperature SNR of about 10 %.

Fig. 10 (abstract A14).
figure 10

Comparison of MR thermometry images when phase encoding is S/I. The single echo thermometry (a,b) demonstrates numerous ghosting artifacts (yellow arrows) that are not seen in ME thermometry (c,d). (a,b,d) are from the same patient

A15 High-resolution whole-brain MR thermometry with a 3D EPI stack-of-stars pulse sequence

Sumeeth Jonathan, William Grissom

Vanderbilt University, Nashville, Tennessee, USA


Real-time whole-brain MR thermometry is needed for transcranial MR-guided Focused Ultrasound to accurately track rapid heating at the focus and to monitor for unsafe heating in the near- and far-field. Previous efforts to meet this need have been based on 3D segmented echo planar imaging (EPI) and spiral MR readouts (Fielden et al., 2014). Here, we propose a 3D EPI stack-of-stars temperature mapping pulse sequence that enables greater volume coverage than has been reported with previous approaches. The sequence allows flexible adjustment of the scan acceleration factor and does not require a high density of receive coils for high frame rate thermometry.


Readout Trajectory: Fig. 11 shows the proposed readout trajectory. Each TR comprises a 2D EPI plane that is frequency-encoded in the axial (x-y) and phase-encoded in the slice (z) dimensions. The plane is rotated between TRs by 111.25 degrees to sample a golden angle stack-of-stars k-space. The sequence was implemented on a Philips Achieva 3 Tesla scanner with parameters: field-of-view = 28.0 x 28.0 x 11.9 cm, 43 slices, in-plane resolution = 1.50 x 1.50 x 2.75 mm, 17 ms TE/45 ms TR, 14.5° flip angle, spectrally-selective fat suppression.

In vivo Experiment: To confirm the achievable brain volume coverage and image quality, with local IRB approval, brain images in a healthy volunteer were acquired without heating using the pulse sequence with one, two, and three interleaved shots and with an 8-channel receive coil array. Fully-sampled images were reconstructed using the density-compensated conjugate gradient method.

Phantom Heating Experiment: A tissue-mimicking gel phantom was sonicated using a clinical MR-HIFU system (Philips Sonalleve) operated at 1.2 MHz and 80 W for 12 s, while scanning with the proposed sequence with one shot and with 5 abdominal receive coils. Temperature maps were reconstructed in one second increments using the k-space hybrid method at three scan acceleration factors.


In vivo Results: Distortions and signal dropouts caused by off resonance are visible in the sagittal and lower axial single-shot brain images in Fig. 12 (white arrows). The distortions can be reduced by increasing the number of EPI shots per angle, which increases the pixel bandwidth in z. An axial slice positioned for monitoring Focused Ultrasound thalamotomy (red arrow) is also shown, which contains no visible distortions at any multishot factor.

Phantom Heating Results: Fig. 13a shows reconstructed sagittal and axial temperature maps through the middle of the phantom at peak heat. Significant temperature aliasing does not appear until the scan is accelerated by a factor of 14.1 or a 1 second window width. The hotspot temperature curve plots in Fig. 13b show that the reconstructed temperatures all coincide, though the 14.1x curve appears noisier. Importantly, at all acceleration factors, there is agreement with a hotspot measurement obtained by a 2D multislice Cartesian EPI temperature mapping pulse sequence (7 slices, same in-plane resolution) which was used as a reference standard.

A video of the entire sonication monitored using this pulse sequence can be viewed at


A 3D EPI stack-of-stars temperature mapping pulse sequence was proposed and validated against a 2D multislice Cartesian EPI temperature mapping pulse sequence. The sequence enables fine spatiotemporal resolution with large volume coverage, without requiring temporally-regularized reconstruction or a large number of receive coils.

Fig. 11 (abstract A15).
figure 11

Illustration of the 3D k-space trajectory. A 2D EPI plane is scanned each TR and rotated by 111.25 degrees between TRs to sample a 3D volume. Scan acceleration is achieved by using a small number of consecutive rotated planes/TRs for reconstruction

Fig. 12 (abstract A15).
figure 12

Brain images acquired with the proposed sequence. The sagittal and lower axial slice images contain distortions and dropout in regions with large frequency offsets (white arrows), which are diminished when multiple shots are used to shorten each readout

Fig. 13 (abstract A15).
figure 13

Phantom temperature map reconstructions (a) and hot spot temperature curves (b) at three acceleration factors. Significant aliasing does not appear until the scan is accelerated by 14.1x. There is good reference standard agreement at all accelerations

A16 Ultrafast and sensitive volumetric passive acoustic mapping

Costas Arvanitis1, Nathan McDannold2,3, Gregory Clement4

1Georgia Institute of Technology, Atlanta, Georgia, USA; 2Brigham and Women’s Hospital, Boston, Massachusetts, USA; 3Harvard Medical School, Boston, Massachusetts, USA; 4Cleveland Clinic, Cleveland, Ohio, USA


Not released for publication


Not released for publication


Not released for publication


Not released for publication

A17 Tissue stiffness imaging with interleaved multiple-point MR-ARFI

Dennis Parker, Joshua de Bever, Henrik Odéen, Allison Payne, Douglas Christensen

University of Utah, Salt Lake City, Utah, USA


Although Focused Ultrasound (FUS) has the potential to treat a number of pathologies the methods used to guide treatment are still limited in the ability to identify effective treatment endpoints. Changes in MRI parameters such as proton density and T1 and T2 relaxation times may result from reversible edema instead of tissue death. While dynamic contrast enhanced and late Gd-enhanced MRI are more specific, Gd contrast cannot be administered multiple times during a procedure to monitor treatment progression. MRI acoustic radiation force impulse imaging (MR-ARFI) can monitor tissue stiffness, but the acquisition is relatively slow, and must be repeated to measure tissue displacement at more than a single point [1]. The purpose of this work was to develop an efficient method to use MR-ARFI to measure tissue displacement volumetrically with an array of points as a first step towards monitoring tissue stiffness changes during MRgFUS procedures.


A 3D segmented gradient echo (GRE) echo-planar pulse sequence, designed to simultaneously measure tissue displacement with MR-ARFI and the corresponding tissue heating [2–5], was further modified to allow interleaved acquisition of multiple point MR-ARFI (mpMR-ARFI) volumes. Phase change due to displacement was separated from that due to temperature by complex subtraction of an interleaved volume acquired with no FUS applied. Temperature was obtained using the proton resonance frequency method with a referenceless background phase subtraction [6].

Experiments were performed in a gelatin phantom and excised pig brain on 3 T MRI scanners (Siemens Tim Trio and PrismaFit). Pulse sequence parameters were: TR/TE = 73/43 ms, FA = 30°, readout bandwidth = 752 Hz/pixel, echotrain length = 7, FOV = 160x114x55mm, matrix = 128x91x22 giving voxel dimensions of 1.25x1.25x2.5 mm before zero filled-interpolation. Bipolar motion-encoding gradients (MEG) of 15 ms duration (each bipolar lobe) were applied prior to signal readout. Ultrasound pulses of 50 acoustic watts were applied during the second 15 ms bipolar MEG lobe. Navigator echoes were used to measure and compensate for B0 field drift due to gradient heating caused by high MEG strength and duty cycle. The same navigators echoes can be used for respiratory correction in vivo [7].


Results from a13-point mpMR-ARFI acquisition in a gelatin phantom using the Siemens Tim Trio and MEG = 20mT/m are given in Fig. 14. An example cropped single slice from 10 of the 13 acquired displacement volumes is shown in Fig. 14a with a cropped slice of the mpMR-ARFI composite of all 13 displacement volumes shown in Fig. 14b. The corresponding temperature increase measured during the mpMR-ARFI acquisition is shown in Fig. 14c and was relatively low (<3 °C). (For all mpMR-ARFI images, point separation is 5 mm).

Figure 15 shows the improved image quality achieved by the phase-navigator correction in mpMR-ARFI measurements obtained using a PrismaFit 3 T MRI scanner with MEG = 40 mT/m. The ghosting artifact in Fig. 15a due to B0 field drift caused by high amplitude (40mT/m) and duty cycle (40 %) MEG gradients during mpMR-ARFI acquisition. After correction (Fig. 15b) mpMR-ARFI displacement images have negligible artifact.

Results of the displacement measured in excised pig brain using the Siemens PrismaFit 3 T MRI scanner and MEG = 40 are shown in Fig. 16 before (Fig. 16a) and after (Fig. 16c) FUS ablation. An estimate of the cumulative thermal dose is shown in Fig. 16b. Note the decreased displacement at central point.


Measuring tissue stiffness (displacement as a function of ultrasound intensity) and changes in displacement before, during, and after the procedure provides the unique potential to remotely palpate the ablated volume to monitor the formation of lesions created with MRgFUS. It will also provide a crucially needed assessment of tissue change in regions of fat where conventional MR thermometry fails.



2. de Bever J, Farrer A, Odeen H, Parker DL. A 3D multi-contrast pulse sequence for acquisition of MR acoustic radiation force imaging concurrently with proton resonance shift thermometry. ISTU; 2014; Las Vegas, Nevada.

3. de Bever J, Farrer A, Odeen H, Parker DL. Simultaneous Acquisition of MR Acoustic Radiation Force Imaging and Proton Resonance Shift Thermometry with 3D Multi-Contrast Pulse Sequence. ISMRM; 2014; Milan, Italy.

4. de Bever J, Odeen H, Parker DL. Measurement of dynamic tissue response to focused ultrasound using 3D MR-ARFI. ISTU; 2014 April 17, 2015; Utrecht, The Netherlands. p Abstract: 2174218

5. de Bever J, Odeen H, Parker DL. Simultaneous Acquisition of Acoustic Radiation Force Imaging and Proton Resonance Frequency Shift Thermometry Using Interleaved Acquisition with Temporally Constrained Reconstruction for Increased Temporal Resolution. ISMRM; 2016; Singapore.

6. Rieke V, Vigen KK, Sommer G, Daniel BL, Pauly JM, Butts K. Referenceless PRF shift thermometry. Magn Reson Med 2004;51(6):1223-1231.

7. Svedin BT, Payne A, Parker DL. Respiration artifact correction in three-dimensional proton resonance frequency MR thermometry using phase navigators. Magn Reson Med 2015; Early View; PMCID: PMC4752934

Fig. 14 (abstract A17).
figure 14

a Individual images, b Composite displacement, c simultaneous heating

Fig. 15 (abstract A17).
figure 15

Ghosting (a) is corrected by navigator echoes (b) allowing clean mpMR-ARFI displacement measurement (c,d)

Fig. 16 (abstract A17).
figure 16

mpMR-ARFI displacement measurements in an excised pig brain before (a) and after (c) MRgFUS ablation with estimated cumulative dose in CEM (b)

A18 Binary localization of cavitation activity based on harmonic content for transcranial brain therapy

Guillaume Maimbourg1, Mathieu David Santin2, Alexandre Houdouin1, Stéphane Lehericy2, Mickael Tanter1, Jean Francois Aubry1

1Institut Langevin Ondes et Images, ESPCI ParisTech, CNRS UMR 7587, INSERM U979, Paris, France; 2CENIR, ICM, CNRS U7225, INSERM U975, Paris, France


Cavitation activity may occur during BBB disruption (due to UCA injections) and also during HIFU treatment due to nucleation under high negative pressure. The corresponding microbubble activity has to be monitored to assess the safety and efficiency of brain treatments by ultrasound. The purpose of this study is to binary discriminate the position of microbubbles, inside or outside the skull, in order to know whether cavitation occurs in the brain of the patient or not.

Binary localization is achieved here by taking advantage of the attenuation properties of the skull. The skull acts indeed as a low pass filter for acoustic signals. Thus we hypothesize that the harmonic content of signals from cavitation events could be used as a binary indicator of their localization: inside or outside the brain case.


A wideband Passive Cavitation Detector (PCD) recorded the acoustic signals from microbubbles activity.

An in vitro setup (Fig. 17) mimics a BBB opening configuration (contrast agent flow with 0.6 MPa sonication during 10 ms at focus) or a thermal ablation (calf brain sample at focus with 3-4 MPa sonication during 0.3 s at focus). Experiments were performed either with no skull in place, or with human (6 samples) or monkey (1 sample) skull in front of the PCD.

In vivo BBB opening were conducted on macaque (900 kHz, 0.6 MPa at focus) with the same PCD mounted on the monkey head.

The spectra are computed from data recorded by the PCD and reveal harmonics, subharmonic and ultraharmonics of the excitation frequency (900 kHz in all cases).

The ratio of a high frequency harmonics over a lower frequency harmonic was then calculated. This ratio is expected to decrease when the acoustic signal from cavitation activity crosses through the skull. In order to achieve binary localization, two types of ratio have been computed: ultraharmonic ratios (e.g. 5/2 over 1/2) and harmonic ratios (e.g. 4 over 2).


The ultraharmonics ratio obtained during in vitro thermal-like ablation is plotted in Fig. 18. This ratio significantly decreases when acoustic signal crosses through the skull. Thus a threshold can be introduced to binary localize cavitation inside/outside the skull.

Table 2 summarises the results for harmonic ratio 4 over 2 and ultraharmonic ratio 5/2 over 1/2 obtained in vitro and in vivo during BBB opening and thermal necrosis. For BBB opening, neither the sub- nor the ultraharmonics appears in vivo, probably due to the confinment of the largest microbubbles by the vasculature. Thankfully in vitro experiments point out that the harmonic ratio remains relevant to binary localize microbubbles in the human case. Indeed the harmonic ratio 4 over 2 exhibits a -30 ± 3 dB decrease when crossing through human skull.

In order to ensure the repeatability of this method, each configuration was statistically investigated by plotting receiver operating characteristics (ROC) curves. These plots (Fig. 19) illustrate the performance of this approach. Except for in vivo BBB opening on monkey, ROC curves demonstrate that a sensitivity and a specificity which are simultaneously nearly 100 % can be obtained for the binary localization of cavitation activity.


This preliminary study, mainly done in vitro, shows that a low-cost and easy-to-use PCD can binary localize cavitation activity inside/outside the skull using the filtrating effect of the skull on the harmonic content of the spectra. The statistical study shows that the method is able to localize microbubble activity inside or outside the skull with high sensibility and high sensitivity. We look forward to testing extensively this technique in vivo for both BBB opening and thermal necrosis.

Fig. 17 (abstract A18).
figure 17

In vitro experimental setup for mimicking BBB opening

Fig. 18 (abstract A18).
figure 18

Ultraharmonics ratio (3/2 to 15/2 over 1/2) recorded during in vitro thermal ablation

Table 2 (abstract A18). ultraharmonic (5/2 over 1/2) and harmonic (4 over 2) ratio during thermal necrosis and BBB opening in dB scale
Fig. 19 (abstract A18).
figure 19

Receiver operating characteristics for ultraharmonic (5/2 over 1/2) ratio and harmonic ratio (4 over 2) during thermal ablation and BBB opening

A19 Inflection of temperature vs. power curve in tcMRgFUS: correlation with lesion location

Kim Butts Pauly1; Christian Federau1; Beat Werner2; Casey Halpern1; Pejman Ghanouni1

1Stanford University, Stanford, California, USA; 2University Children’s Hospital Zurich, Zurich, Switzerland


During tcMRgFUS, power levels are progressively increased to align, verify treatment location, and create a durable ablation. It is predicted that the measured temperature at the focus for a consistent timepoint will increase monotonically with power. However, the temperature rise is often observed to actually decrease with increasing power, creating an inflection in the peak temperature vs. power graph (Fig. 20). The purpose of this work was to show this behavior in the temperature rise is correlated with a lack of alignment between the thermal lesion and the monitored scan plane.


Fifteen ET datasets were included in this study. Post-treatment 3D T2-weighted FSE images (FOV 24 cm, matrix 320x320, slice thickness 1 mm) were thresholded at the signal intensity of zone 1 (Fig. 20). Measurements from the targeted ACPC plane to the top and bottom edges of zone 2 were made in the sagittal plane.

MR thermometry (TE/TR = 12.8/100 (n = 13); TE/TR = 3,8,13,18,22/100, (n = 2)) was processed with a single baseline subtraction (α = -0.00909), followed by referenceless processing for constant and linear terms. The maximum temperature of the third time point (k-space center 8.8 s after sonication initiation) in the axial scan plane was plotted vs. power. This time point was chosen because all sonications were at least 10s in duration. The number of sonications after any inflection was noted.

In 12 cases, temperature-power curves from early sonications in the axial scan plane were extrapolated to higher power levels. The amount that the measured temperature was below the estimated temperature was measured.


The results are shown in Fig. 21. While the distance from the treatment plane to the inferior aspect of the lesion remains essentially constant, the distance from the treatment plane to the superior aspect of the lesion increases with number of sonications after inflection. While it is expected that multiple sonications at the same location may increase lesion size, this data demonstrates that lesion size increases preferentially in the direction of the transducer as we increase sonications after inflection.

The movement of the lesion superiorly from the AC-PC plane is correlated with the difference between the actual and estimated temperature (✰). Although this was quantitated only in the axial scan plane, the inflection was seen in all three scan planes. Prior work demonstrated that the lesion is double oblique (the superior aspect of the lesion is posterior and medial); therefore, movement of the focal spot towards the effective transducer aperture moves it also out of the monitored sagittal and coronal scan planes.


One interpretation of this data is that an increase in the acoustic absorption effectively shields the focal spot, with subsequent hotspots located closer to the transducer. A second explanation is that the acoustic properties of the skull may be changing during treatment.

An additional implication of this work is for comparison of simulation with thermometry. To simplify this comparison without the complication of the alignment of the thermometry plane, a sonication early in the treatment should be used for comparison.

Fig. 20 (abstract A19).
figure 20

a Temperature in the third temperature image (*) b does not always rise monotonically with power. c The distance from ACPC plane to the top and bottom edges of zone 2 were measured on images thresholded at the zone 1 signal intensity

Fig. 21 (abstract A19).