K. Michael Sekins

Characterizing an Agar/Gelatin Phantom for Image Guided Dosing and Feedback Control of High-Intensity Focused Ultrasound

The temperature dependence of an agar/gelatin phantom was evaluated. The purpose was to predict the material property response to high-intensity focused ultrasound (HIFU) for developing ultrasound guided dosing and targeting feedback. Changes in attenuation, sound speed, shear modulus and thermal properties with temperature were examined from 20°C to 70°C for 3 weeks post-manufacture. The attenuation decreased with temperature by a power factor of 0.15. Thermal conductivity, diffusivity and specific heat all increased linearly with temperature for a total change of approximately 16%, 10% and 6%, respectively. Sound speed had a parabolic dependence on temperature similar to that of water. Initially, the shear modulus irreversibly declined with even a slight increase in temperature. Over time, the gel maintained its room temperature shear modulus with moderate heating. A stable phantom was achieved within 2 weeks post-manufacture that possessed quasi-reversible material properties up to nearly 55°C.

Acoustic hemostasis of porcine superficial femoral artery: Simulation and in-vivo experimental studies

In-vivo focused ultrasound studies were computationally simulated and conducted experimentally with the aim of occluding porcine superficial femoral arteries (SFA) via thermal coagulation. A multi-array HIFU applicator was used which electronically scanned multiple beam foci around the target point. The spatio-temporally averaged acoustic and temperature fields were simulated in a fluid dynamics and acousto-thermal finite element model with representative tissue fields, including muscle, vessel and blood. Simulations showed that with an acoustic power of 200W and a dose time of 60s, perivascular tissue reached 91°C; and yet blood reached a maximum 59°C, below the coagulation objective for this dose regime (75°C). Per simulations, acoustic-streaming induced velocity in blood reached 6.1cm/s. In in-vivo experiments, several arteries were treated. As simulated, thermal lesions were observed in muscle surrounding SFA in all cases. In dosing limited to 30 to 60 seconds, it required 257W to provide occlusion (o...

Acoustic neural network thermometry in monitoring of thermal therapies.

The potential benefits of monitoring tissue temperatures and lesioning by “acoustic” methods are evident. Such methods are typically challenging, and limited to interrogating (acoustically) only one or two physical processes during dosing (e.g., sound speed and thermal expansion), which may narrow valid temperature ranges. Due to the complexity of tissue behavior and acoustic interactions during heating, a “gestalt” of acoustic parameters was enabled to be “learned” during dosing by a recurrent neural network (RNN). Using this method, real‐time three‐dimensional (3‐D) acoustic thermometry was demonstrated and (from RNN coefficients) insights were gained into specific acoustic parameter contributions accompanying temperature changes during and after dosing. The RNN method is described for two applications. First, for radiofrequency ablation on excised bovine livers, coregistered B‐mode and ARFI elasticity images were acquired. RNN thermometry images were then calculated from rf data, thermal doses calculat...

A method for automated detection of high intensity focused ultrasound (HIFU) beams in 3D space

In HIFU therapy, it is critical to determine the HIFU beam path and its focus before the therapy-level dose is administered into tissue. In this work, we adapted echo-strain methods to construct the HIFU beam information and developed an automated detection method using 3D data segmentation, beam trajectory estimation, and focal point localization. A series of 3D volume ultrasound data acquisitions was interleaved in time with a testing pulse with low intensity (Isppa ∼200 W/cm2) and small duty cycle (4.7%). 3D echo-strain was derived from tissue apparent displacement and was segmented based on a volume growing technique. Our methods were tested on tissue mimicking phantom materials with acoustic properties similar to human muscle. Testing was done to show repeatability and accuracy for HIFU beams focused at varying depths and steering angles. The error from thirteen (13) different focal positions was 5.7 ± 2.3 mm (mean ± std) in 3D space, with error defined as the Euclidian distance between the estimated focal location and the experimentally determined focal location. Experimental location was determined via a thermocouple embedded in the phantom, and the location of the thermocouple was determined by the highest temperature rise location when moving the HIFU 2D array transducer through a 3D volume using a short (1 second) and low power (Isppa ∼40W/cm2) focused beam sonication. It was concluded that an automated HIFU beam focal point detection method was developed and provides accurate localization of the HIFU beam focus.

Real‐time Tissue Thermometry Using an Acoustic Neural Network Method during HIFU Treatment

A recurrent neural network solution was proposed to map changes in multiple acoustic parameters, such as apparent displacement, echo backscattered intensity change, echo signal correlation coefficient and tissue elasticity, to temperature rise during High Intensity Focused Ultrasound (HIFU) dosing. The training and testing of the neural network was based on data from experiments on HIFU treated ex‐vivo porcine and bovine liver. The experiments included acquiring 3D echo signals along with directly measuring temperature with thermocouples in the region of the HIFU focal zone (treatment target) before and during the HIFU dosing event. The results demonstrate that the proposed method may provide temperature rise estimation (thermometry) over practical temperature ranges for the therapy, and may reveal the individual acoustic parameters’ correlations with temperature rise.

Acoustic hemostasis device for automated treatment of bleeding in limbs

A research prototype automated image-guided acoustic hemostasis system for treatment of deep bleeding was developed and tested in limb phantoms. The system incorporated a flexible, conformal acoustic applicator cuff. Electronically steered and focused therapeutic arrays (Tx) populated the cuff to enable dosing from multiple Txs simultaneously. Similarly, multiple imaging arrays (Ix) were deployed on the cuff to enable 3D compounded images for targeting and treatment monitoring. To affect a lightweight cuff, highly integrated Tx electrical circuitry was implemented, fabric and lightweight structural materials were used, and components were minimized. Novel cuff and Ix and Tx mechanical registration approaches were used to insure targeting accuracy. Two-step automation was implemented: 1) targeting (3D image volume acquisition and stitching, Power and Pulsed Wave Doppler automated bleeder detection, identification of bone, followed by closed-loop iterative Tx beam targeting), and 2) automated dosing (auto-selection of arrays and Tx dosing parameters, power initiation and then monitoring by acoustic thermometry for power shut-off). In final testing the device automatically detected 65% of all bleeders (with various bleeder flow rates). Accurate targeting was achieved in HIFU phantoms with end-dose (30 sec) temperature rise reaching the desired 33-58°C. Automated closed-loop targeting and treatment was demonstrated in separate phantoms.A research prototype automated image-guided acoustic hemostasis system for treatment of deep bleeding was developed and tested in limb phantoms. The system incorporated a flexible, conformal acoustic applicator cuff. Electronically steered and focused therapeutic arrays (Tx) populated the cuff to enable dosing from multiple Txs simultaneously. Similarly, multiple imaging arrays (Ix) were deployed on the cuff to enable 3D compounded images for targeting and treatment monitoring. To affect a lightweight cuff, highly integrated Tx electrical circuitry was implemented, fabric and lightweight structural materials were used, and components were minimized. Novel cuff and Ix and Tx mechanical registration approaches were used to insure targeting accuracy. Two-step automation was implemented: 1) targeting (3D image volume acquisition and stitching, Power and Pulsed Wave Doppler automated bleeder detection, identification of bone, followed by closed-loop iterative Tx beam targeting), and 2) automated dosing (auto-...

High intensity focused ultrasound characterization of deep bleeder acoustic coagulation cuffs.

Research prototype therapeutic ultrasound “cuffs” have been recently constructed to demonstrate the feasibility of cauterizing bleeding (acoustic hemostasis) of deep limb wounds, such as those from military combat or other penetrating trauma. The deep bleeder acoustic coagulation (DBAC) cuffs required a variety of HIFU power and acoustic beam assessments to assure safety and efficacy. The cuff, capable of > 1000 W acoustic power, comprised multiple two‐dimensional (2‐D) HIFU arrays that surrounded the limb and could dose with multiple arrays simultaneously. Each array was capable of high power (>170 W), significant electronic steering (solid angle = 57×34 deg), and depth focusing (5–20 cm). Device characterization required both individual array power measurements, and “integrated” whole cuff multiarray dosing assessment. The single array characterizations included (a) low‐power beam plots, (b) high power focused power assessment in “apertured” radiation force balance tests, and (c) Schlieren beam pattern ...

Real-time RNN-based acoustic thermometry with feedback control

A major obstacle to the widespread adoption of HIFU therapy is the development of a suitable method of monitoring the a blation therapy in real-time. While MR-thermometry has emerged as a promising method for HIFU therapy monitoring, acoustic guidance has continuously been sought for reasons of cost and practicality. We have previously demonstrated the potential of acoustic thermometry, by using a recurrent neural network (RNN) to estimate changes in tissue temperature during HIFU ablation therapies. A limitation of this method is that an excessive therapeutic dose can cause multiple, non-linear changes within the ultrasound data, resulting in unreliable temperature estimates from the RNN. Accordingly, we propose a revised method of dosing wherein closed loop feedback is used to provide a controlled and specific dose; not only to ensure an efficacious lesion, but also to preserve the integrity of the ultrasound image, thereby producing accurate temperature estimates from the RNN. This investigation of con...

Understanding changes in tissue phantom material properties with temperature.

Phantoms used for high intensity focused ultrasound (HIFU) applications require rigorous evaluation of material properties since, locally, the material experiences extreme changes in temperature and stresses with the HIFU treatment. Here we present the testing of an agar/gelatin phantom intended for both acoustic radiation force imaging (ARFI) and HIFU applications. The phantom shear modulus, speed of sound, attenuation, and thermal properties were all evaluated over the range of room temperature to 80 °C. With the exception of the thermal properties, all measurements were taken during both heating and cool down. Cavitation threshold and melting point were also tested. The change in material sound speed and thermal properties with temperature were quasireversible and similar to that of water. Material attenuation showed a slight decrease with temperature, but appeared to also be reversible. Shear modulus decreased significantly with temperature, going to near zero. The response was not reversible, returni...

Simulation aided dosing control of a 2D array therapeutic ultrasound transducer

Setting proper input power and duty cycle to produce a desired thermal dose is a non-trivial task in ultrasound thermal therapy. This paper presents an approach to estimate these parameters based on computer simulations. An in-house custom 2D therapeutic phased array was used for the studies. Acoustic power measurements were used to provide input values and validation for the pressure and temperature field simulations. Simulated pressure, intensity and temperature fields were in turn used to optimize the dosing control parameters. Simulation-projected dosing parameters were verified in a gellum-gum based tissue mimicking phantom (TMM) with 75μm butt-welded thermocouples (TC) embedded inside. The time averaged input powers were varied from 12W to 25W. A 30s dose was applied at each setting, and the temperatures at the focus and a location 2mm away from the focus in array azimuth direction were acquired. Results showed that the peak end-of-dose temperature rise was approximately linearly proportional to the input acoustic power. Simulations predicted a similar linear relationship with less than 10% error from the experimental results. Ex-vivo bovine liver studies also showed good correlation between the simulation-predicted lesion and the experimental lesion; similar patterns were observed. The results demonstrated that by combining simulations with force balance measurement, dosing instructions for any desired focal temperature can be quickly and easily produced.