Michael Burcher

Capon beamforming in medical ultrasound imaging with focused beams

Medical ultrasound imaging is conventionally done by insonifying the imaged medium with focused beams. The backscattered echoes are beamformed using delay-and-sum operations that cannot completely eliminate the contribution of signals backscattered by structures off the imaging beam to the beamsum. It leads to images with limited resolution and contrast. This paper presents an adaptation of the Capon beam- former algorithm to ultrasound medical imaging with focused beams. The strategy is to apply data-dependent weight functions to the imaging aperture. These weights act as lateral spatial filters that filter out off-axis signals. The weights are computed for each point in the imaged medium, from the statistical analysis of the signals backscattered by that point to the different elements of the imaging probe when insonifying it with different focused beams. Phantom and in vivo images are presented to illustrate the benefits of the Capon algorithm over the conventional delay-and-sum approach. On heart sector images, the clutter in the heart chambers is decreased. The endocardium border is better defined. On abdominal linear array images, significant contrast and resolution enhancement are observed.

Time reversal operator decomposition with focused transmission and robustness to speckle noise : Application to microcalcification detection

The decomposition of the time reversal operator (DORT) is a detection and focusing technique using an array of transmit-receive transducers. In the absence of noise and under certain conditions, the eigenvectors of the time reversal operator contain the focal laws to focus ideally on well-resolved scatterers even in the presence of strong aberration. This paper describes a new algorithm, FDORT, which uses focused transmission schemes to acquire the operator. It can be performed from medical scanner data. A mathematical derivation of this algorithm is given and it is compared with the conventional algorithm, both theoretically and with numerical experiments. In the presence of strong speckle signals, the DORT method usually fails. The influence of the speckle noise is explained and a solution based on FDORT is presented, that enables detection of targets in complex media. Finally, an algorithm for microcalcification detection is proposed. In-vivo results show the potential of these techniques.

Real-time photoacoustic data acquisition with Philips iU22 ultrasound scanner

A one of a kind photoacoustic system has been built around a Philips iU22 ultrasound scanner. The modified channel board architecture allows access to the raw per-channel photoacoustic data, while keeping all of the imaging capabilities of an actual commercial ultrasound scanner. A captured photoacoustic data frame is Fourier beamformed to generate a single laser shot photoacoustic image. In addition to the photoacoustic data, the system supplies the beamformed ultrasound data, providing a truly dual-modality imaging capability. A tunable OPO laser system (700-900nm), pumped by an Nd:YAG solid state laser, is used as an illumination source with 5ns long pulses. An FPGA-based electronic board synchronizes the iU22 start of frame with the laser firing, currently permitting photoacoustic imaging at a rate of 10 Hz (laser repetition rate limit). At that imaging frame rate the photoacoustic system, consisting of a PC modified with 32 Gbytes of acquisition memory and an FPGA array, is able to store several minutes of continuously captured data, enabling monitoring and off-line analysis of dynamic photoacoustic events and/or fast scanning for performing pseudo-3D imaging. The system can use all of the standard iU22 array transducers both for photoacoustic imaging, and in all of the ultrasound imaging modes.

Acoustic fingerprints of dye-labeled protein submicrosphere photoacoustic contrast agents

Dye-labeled protein microspheres, submicron in size and capable of producing thermoelastically generated ultrasound in response to laser stimulation, are presented as contrast agents for photoacoustic imaging. Incident laser energy absorbed by fluorescein isothiocyanate (FITC)-labeled elastin submicrospheres results in thermoelastically generated sound production. Plotted A-line graphs reveal a distinctive morphology and a greater than two orders of magnitude increase in signal amplitude subsequent to converting FITC elastin into submicrospheres (despite a four orders of magnitude decrease in concentration). Evidence of nonlinearity and enhancement of ultrasound backscatter indicate a potential use in contrast-enhanced harmonic imaging. Photoacoustic and ultrasound imaging of FITC-elastin submicrospheres in a water-filled phantom vessel shows enhanced contrast at low concentration and clear delineation of the phantom vessel wall.

Determination of temporal bone Isoplanatic Patch sizes for transcranial phase aberration correction

Phase aberration is a leading cause of transcranial ultrasound image degradation. In order to realign aberrated wavefronts, a delay map corresponding to the aberration can be computed from signals backscattered from a region of interest (ROI) in the medium, and used to correct the beamforming delays. However, such a map is only effective for correcting the aberration in a limited area called the isoplanatic patch (IP) around the ROI. This fundamentally limits the effectiveness of transcranial aberration correction to restore image quality. In this paper, IP sizes are measured in vitro for aberration correction with an X7-2 2D array (Philips Healthcare, Andover, MA) through 12 ex vivo human temporal bone samples. The angular IP size is found to be 36deg plusmn 18deg. An in vivo experiment confirms that the IP is limited angularly (~30deg) but large in depth (~15 cm). Small IP sizes and high refocusing effectiveness within the IP are correlated with high gradients in the measured phase aberration maps. This study indicates that phase aberration correction with a single delay map is only effective for transcranial ultrasound applications with a small angular field of view.

A novel phase aberration measurement technique derived from the DORT method: comparison with correlation-based method on simulated and in-vivo data

Clinical ultrasound image quality is degraded by tissue velocity inhomogeneities that distort the phase of the traveling ultrasonic wavefronts. degrading image resolution and contrast. This paper investigates an entirely new approach for aberration measurement that uses a modified DORT (French acronym for decomposition of the time reversal operator) method, which uses imaging transmission schemes to acquire the time reversal operator. In this novel scheme, the aberration profile is directly estimated from the unwrapped phase of the first DORT eigenvectcr. We term this method focused DORT (FDORT). We compared the FDORT method with a correlation-based, least-mean-squares (LMS) method using 1D arrays on: simulated data; experimental phantom data with a thin-film rubber aberrator; and in-vivo breast data. Measurements were carried out in regions of speckle and point-like scatterers, and corrections were performed using a near-field phase screen aberration model. In simulation, the residual rms error (between applied and estimated aberrator) was 12.5ns with FDORT and 19.5ns with LMS; the point target brightness improvement was 122% (/spl plusmn/29%) for FDORT and 132% (/spl plusmn/20%) for LMS. The phantom experiments showed 46% (/spl plusmn/25%) and 48% (/spl plusmn/22%) improvements in point target brightness for FDORT and LMS, respectively. In clinical data, microcalcifications were identified and used to estimate the aberrations. FDORT measured an average 65ns rms, 6.5mm FWHM aberrator with an average brightness improvement of 58% (n=3). LMS measured an average 50.2ns rms, 5.7mm FWHM aberrator with an average improvement of 61% (n=3). FDORT is observed to follow wavefront variations, even where LMS does not perform well due to low correlation values. This is particularly evident in the breast data where coherent wavefronts do not extend across the entire aperture. The initial results presented here indicate that FDORT is a useful method to estimate aberration profiles for adaptive imaging. LMS breaks down in the presence of interfering wavefronts from off-axis and multiple strong scatterers. MORT is able to identify this situation since the individual wavefronts are associated with different eigenvectors. We see the potential for FDORT to cope with these conditions and thereby estimate aberrations with greater robustness.

In Vivo Photoacoustic Imaging of Nude Mice Vasculature Using a Photoacoustic Imaging System Based on a Commercial Ultrasound Scanner

In-vivo photoacoustic/ultrasound (PA/US) imaging of nude mice was investigated using a photoacoustic imaging system based on a commercial ultrasound scanner HDI-5000. Raw per-channel data was captured and beamformed to generate each individual photoacoustic image with a single laser shot. An ultra-broadband CL15-7 linear array with a center frequency of 8 MHz, combined with a Schott Glass fiber bundle, was used as a compact high resolution imaging probe, with lateral and axial PA resolutions of about 300µm and 200µm, respectively. The imaging system worked in a dual PA-US mode, with the ultrasound outlining the tissue structure and the photoacoustic image showing the blood vessels. PA signals were generated by exposing mice to ultra-short optical pulses from a Nd:YAG-pumped OPO laser operating in a wavelength range of 700-950nm. The corresponding ultrasound images were generated in the regular B-mode with standard delay-and-sum beamforming algorithm. The system resolution was sufficiently high to identify and clearly distinguish the dorsal artery and the two lateral veins in the mouse tail. Both the saphena artery and the ischiatic vein on the cross-section of the mouse leg were clearly outlined in the PA images and correctly overlaid on the ultrasound image of the tissue structure. Similarly, cross-section PA images of the mouse abdomen revealed mesenteric vasculatures located below the abdominal wall. Finally, a successful PA imaging of the mouse thoracic cavity unveiled the ascending and descending aorta. These initial results demonstrate a great potential for a dual photoacoustic/ultrasound imaging modality implemented on a commercial ultrasound imaging scanner.

Aberration estimation using FDORT: insights and improved method for speckle signals

Clinical ultrasound imaging is degraded by tissue velocity inhomogeneities that reduce resolution and contrast. The Focused DORT (French acronym for decomposition of the time reversal operator) method, here denoted FDORT, can be used as an aberration estimation method. FDORT uses per-channel received RF data obtained from several focused transmissions. An aberration profile is estimated using a singular value decomposition method. The advantage of FDORT over a crosscorrelation based method is its ability to identify individual wavefronts in complex RF regions. It also allows frequencydependent estimation. FDORT exhibits good results with point scatterers but significant residual rms errors (between applied and estimated aberrator) in speckle regions (typically 15ns for a 45ns aberration). This study aims to explain the behavior of FDORT in speckle by interpreting the FDORT matrix as a crossspectrum matrix of the backscattered signal, similar to the one studied by Måsøy et al. [JASA 117 (1) 2005]. This explanation is then used to improve the method; reducing both the bias and variance of the estimation. Originally, the first singular vector was used for the estimation: it contained the amplitude and delay law that maximized the speckle brightness. A bias results from the fact that the transmit itself is aberrated. We propose a method to reduce the bias to a linear shift by using a combination of all eigenvectors. A new scheme is introduced to reduce the variance, typically by a factor of 2. Finally, we introduce a non-biased estimator by forming a tensor from a full synthetic aperture data set. Its first singular vector maximizes the speckle brightness by correcting both transmit and receive this is equivalent to an FDORT estimation when the aberration is completely corrected in transmit and provides the lowest error. We performed Field-II (Jensen) simulations using a 45ns, 4mm FWHM near-field phase screen on a speckle phantom with point scatterers and cysts. Cyst contrast improvements and rms residual errors are computed for the different methods. In-vivo aberration estimation results are also presented. The theoretical understanding of the speckle behavior in FDORT has led to improved performance and further insights into using statistical-based approaches for aberration measurement. Keywords-aberration, medical, correlation matrix

Preserving speckle statistics in minimum-variance beamformed images: the effectiveness of spatial compounding

Data-dependent apodization techniques such as the minimum-variance beamformer (MVB) can beat the diffraction limit of the conventional Delay-and-Sum (DAS) beamformer, yielding enhanced resolution and contrast. However, the MVB algorithm strongly increases the speckle variance. This needs to be compensated in order for the MVB algorithm to be clinically successful. This paper shows that combining the MVB with spatial compounding yields unprecedented image quality with significantly increased lateral resolution and decreased clutter compared to conventional DAS imaging, without the drawback of increased speckle variance.

A Modified Commercial Ultrasound Scanner used for In-Vivo Photoacoustic Imaging of Nude Mice Injected with Non-Targeted Contrast Agents

Photoacoustic (PA) experiments were performed using a modified commercial ultrasound scanner equipped with an array transducer and a Nd:YAG pumped OPO laser. The contrast agent SIDAG (Bayer Schering Pharma AG, Germany), used to enhance the optical absorption, demonstrated an expected pharmacokinetic behavior of the dye in the tumor and in the bladder of the nude mice. A typical behavior in the tumor consisted of an initial linear increase in PA signal followed by an exponential decay. PA signal approached the pre-injection level after about one hour following the dye injection, which was consistent with the behavior for such contrast agents when used in other imaging modalities, such as fluorescence imaging. The in-vivo spectral PA data from the mouse bladder, conducted 1.5 hours after the dye injection, clearly demonstrated presence of the dye. The multi-spectral PA data was obtained at 760nm, 784nm and 850nm laser excitations. The PA intensities obtained at these three wavelengths accurately matched the dye absorption spectrum. In addition, in the kidney, a clearance organ for this contrast agent, both in-vivo and ex-vivo results demonstrated a significant increase (~ 40%) in the ratio of PA signal at 760nm (the peak of the dye absorption) relative to the signal at 850nm (<1% absorption), indicating significant amounts of the dye in this organ. Our initial results confirm the desired photoacoustic properties of the contrast agent, indicating its great potential to be used for imaging with a commercial array-based ultrasound scanner.