Experimental Design: Milestone 1

Milestone 1: Implementation of a Real-time, user-controlled, 128 Channel ICE based ARFI system

Our current implementation of ICE ARFI utilizes the 64-channel Siemens AcuNav array, offers no real-time user controls and requires off-line processing of echo signals to form ARFI images. We propose to create an operator adjustable, real-time 128 channel ARFI system.

Scanner Modification for 128 Channel ICE Catheter

The new 128-element intra-cardiac transducer will require modifications of the scanner firmware to form real-time ARFI images in a clinical setting with conventional operator controls. We will develop and implement a new mapping for the transmit/receive channels to enable proper beam forming. We will modify the probe files and other scanner parameters with the appropriate transducer geometries. Finally, we will optimiz e the gain management controls in order to improve image quality and maximize SNR.

Once the new transducer has been fully integrated into the system, the scanner will be programmed to transmit radiation force pulses. The scanner settings will be saved and merged into custom beam sequences that properly execute an ARFI image acquisition, such that the reference, excitation, and tracking lines are all transmitted and recorded at the appropriate times. These sequences will utilize all of the advances in ARFI imaging including ECG-gating, parallel-tracking, and multiplexed imaging and will be generated for various depths and radiation force pulse durations.

Once generated, theses equences will be t ested on phantoms and excised tissue, where the performance of the new array will be evaluated and compared with the commercially available 64-element catheter. All of the acoustic intensity and thermal safety measurements will be remade for the new array to determine suitable limits on frame rates and excitation pulse lengths. If transducer heating remains a concern, we will work with Siemens to mount thermistors on the new arrays to ensure compliance with FDA guidelines.

Design and Implementation of User Interface for ARFI Imaging

In the environment of a beating heart, an ARFI imaging system must be flexible enough to provide some user control such that appropriate sequences with application specific parameters can be quickly generated. Under other NIH funding, we are currently working with Siemens Medical Systems to implement a realtime ARFI imaging platform on the Siemens Antares scanner. Once completed in early 2007, the scanner will be able to generate, execute, process, and display ARFI sequences with simultaneous B-mode imaging. The user will be able to control excitation pulse length, excitation pulse amplitude, depth of focus, field of view, display time, motion filter time threshold, and the display dynamic range. Side by side B-mode and ARFI images will be displayed on the Antares monitor in real time.

We will develop three methods of ARFI image acquisition: single acquisition, continuous acquisition, and ECG-gated continuous acquisition. Single acquisition will allow the user to optimize imaging parameters without needlessly transmitting high-energy radiation force pulses into the patient. Continuous acquisition will be used to visualize the changes in myocardial stiffness as the heart beats. This will also likely be the best method to localize the regions of interest and determine proper imaging plane alignment. As ECG-gating allows repeated image acquisition through a specific cross section of the heart, we will use ECG-gated acquisition to image the developing lesion as it forms during the ablation. For all modes, we will measure FDA- described safety related parameters (MI, TI, probe temperature, ISPTA).

Design, Development and Implementations of Motion Filters

Motion filters are necessary to remove any motion not imparted to the tissue by radiation force pulses. This motion includes bulk cardiac motion and the rebound motion of the catheter as the pulses are emitted. We model these motion sources with a much slower accelerations than those resulting from radiation force excitation and have devised linear, quadratic and more sophisticated filters to eliminate the artifacts they impose on ARFI images (ref).

We propose to design, implement and experimentally evaluate improved “motion filters” for ICE ARFI imaging. The filters utilize local measurements of cardiac motion before the ARFI “pushing pulse” is applied and/or after the tissue has recovered to form a constant velocity, constant acceleration, or higher-order model of local cardiac motion. The predictions of the model are used to subtract physiologic motion occurring during ARFI-induced motion from the measured displacements. We propose to utilize pulse sequences with zero-voltage, ARFI pushing pulses during our animal studies to evaluate alternative motion filtering methods. The extent to which a filter accurately predicts motion during the pseudo-ARFI displacement in these studies will be used to assess its performance. Since the physiologic motion can be measured by conventional methods, a gold standard is available for these filter design studies. Raw echo data will be used to evaluate filters in off-line studies and promising candidates will be coded into the Antares scanner.

Implementation of Cardiac Synchronous Acquisition with Decreased Scan Time

Our realization of real-time intra-cardiac ARFI uses ECG-gating of the acquisitions to minimiz e the effect of cardiac motion. A single complete ARFI image is formed during the diastolic phase of each heartbeat. With a typical heart rate, the frame rate is 1-2 Hz. To minimize motion artifact and limit jitter due to cardiac trigger instability we will decrease the single frame acquisition time from 220 ms t o 50 ms. This improvement in acquisition time will be implemented by parallel receive processing.

We propose to implement parallel receive imaging to track displacements simultaneously along four beamlines centered about the excitation pulse. As a result, the acquisition time and the required number of excitations pulses to interrogate a specific region of interest will be reduced by a factor of four.

Currently, the scanner samples raw radiofrequency data at a frequency of 40 MHz, whereas quadrature demodulated data can be sampled without aliasing at 6 MHz. With the scanner's sampling frequency far exceeding the Nyquist criteria, we will sequentially address four laterally adjacent B-mode lines for each transmitted pulse. This realization of parallel receive imaging will allow us to track displacements simultaneously along four beam-lines centered about the excitation pulse. As a result, the acquisition time and the required number of excitation pulses to interrogate a specific region of interest will be reduced by a factor of four.

In multiplexed ARFI imaging, radiation force pulses are fired consecutively along three beams spaced one third of the lateral field of view apart. Multiplexed tracking lines, consisting of four parallel receive lines centered around each of the excitation pulses are then recorded consecutively. As the received lines are multiplexed across three different locations, the temporal sampling resolutions at a single location will be reduced by a factor of three. However, as the primary application of these sequences is real-time imaging, where displacements at a single instant of time are displayed, the loss of temporal resolution was deemed to be acceptable.

The paper in Appendix C describes parallel receive beam-forming and multiplexing techniques in ARFI imaging in detail (ref Dahl).