Fong Ming Hooi

Hybrid beamforming and steering with reconfigurable arrays

Reconfigurable arrays offer an advantage over traditional ultrasound arrays because of their flexibility in channel selection. To improve ultrasound beamforming and coverage through beam steering, we propose a hybrid beamforming technique to elongate the depth of focus of transmit beams and a method of element selection that improves steering capabilities that take advantage of array reconflgurability using annular rings. A local minimization technique to optimize the hybrid aperture is discussed in this paper. The chosen hybrid apertures covering four focal zones result in improved range in depth of focus when compared with pure spherical beams via point spread functions (PSF) and lesion signal-tonoise ratio (LSNR) calculations. Improvements were statistically significant at focal depth extremes. Our method of beam steering utilizing a quantized phase delay selection to minimize delay errors indicated better performance by removing an artifact present with traditional ringed element selection.

Dual sided automated ultrasound system in the mammographic geometry

This paper reports design and performance of an ultrasound system for automated breast imaging from both sides of the breast in standard mammographic views. A commercial mammographic system was modified to allow compression paddles on both sides of the breast and scanning of two commercial ultrasound transducers over those paddles. The paddle surface is a thin mesh providing negligible ultrasound wave distortion and some permeability to the acoustic coupling gel or lotion. Initial studies on four volunteers show marked improvement in image quality and coverage of the breast over imaging from either side alone. Visual comparisons of various simple methods of combining and displaying the image volumes from the two sides are presented. Various artifacts present in the images without speed of sound and attenuation corrections are illustrated, and methods of combining these image volumes with those of other breast screening modes and modalities are discussed.

Effect of a Gel Retainment Dam on Automated Ultrasound Coverage in a Dual-Modality Breast Imaging System

Objective. The goal of this work was to evaluate a possible improvement in ultrasound coverage for a dual‐modality breast imaging system in the mammographic geometry. Methods. A pilot study was performed to evaluate use of a rubber dam to retain ultrasound gel and improve imaging coverage at the breast periphery on a combined imaging system consisting of an ultrasound scanner and a digital x‐ray tomosynthesis unit. Several dams were constructed to encompass the shapes of various sizes of compressed breasts. Visual tracings of the breast‐to‐paddle contact area and breast periphery were made for 8 breasts to estimate coverage area. Two readers independently reviewed the resulting images and were asked to rate the overall breast image quality. Results. The percentages of breast in contact with the paddle were greater (P < .01) and the linear dimensions of breast in contact with the paddle were larger (P < .05) with the rubber dam than without it. With the dam, the mean estimated area of the breast in contact with the paddle increased 14%, whereas the mean increase in the fraction of the total breast area in contact with paddle was 30%. The difference was due to the mean total projected area of the breast decreasing 12% as the dam was pressed against it. The image quality of automated ultrasound with the rubber dam was consistently judged to be superior to that without the dam. Conclusions. This method can enhance the absolute and percentage area of the breast in contact with the paddle, reducing noncontact gaps at the breast periphery. Gently pressing the breast periphery with the dam inserted toward the chest wall improves coverage in automated breast ultrasound scanning.

Large Area MEMS Based Ultrasound Device for Cancer Detection.

We present image results obtained using a prototype ultrasound array which demonstrates the fundamental architecture for a large area MEMS based ultrasound device for detection of breast cancer. The prototype array consists of a tiling of capacitive Micro-Machined Ultrasound Transducers (cMUTs) which have been flip-chip attached to a rigid organic substrate. The pitch on the cMUT elements is 185 um and the operating frequency is nominally 9 MHz. The spatial resolution of the new probe is comparable to production PZT probes, however the sensitivity is reduced by conditions that should be correctable. Simulated opposed-view image registration and Speed of Sound volume reconstruction results for ultrasound in the mammographic geometry are also presented.

Machine learning for noise removal on breast ultrasound images

This study assessed the utility of machine learning for isolating noise and artifacts in breast ultrasound images. Such corrupt image regions (ROIs) can be automatically excluded when registering images acquired from different angles. Artifacts included posterior acoustic shadowing and enhancement arising from cancers and cysts respectively. Images were obtained on a breast-mimicking phantom containing multiple cysts and lesions with variable speed of sound and attenuation properties. In vivo breast images of cysts and cancers were also available. Results show that the classifiers were able to identify the regions of corrupt data accurately.

Combined photoacoustic and ultrasound imaging of human breast in vivo in the mammographic geometry

This photoacoustic volume imaging (PAVI) system is designed to study breast cancer detection and diagnosis in the mammographic geometry in combination with automated 3D ultrasound (AUS). The good penetration of near-infrared (NIR) light and high receiving sensitivity of a broad bandwidth, 572 element, 2D PVDF array at a low center-frequency of 1MHz were utilized with 20 channel simultaneous acquisition. The feasibility of this system in imaging optically absorbing objects in deep breast tissues was assessed first through experiments on ex vivo whole breasts. The blood filled pseudo lesions were imaged at depths up to 49 mm in the specimens. In vivo imaging of human breasts has been conducted. 3D PAVI image stacks of human breasts were coregistered and compared with 3D ultrasound image stacks of the same breasts. Using the designed system, PAVI shows satisfactory imaging depth and sensitivity for coverage of the entire breast when imaged from both sides with mild compression in the mammographic geometry. With its unique soft tissue contrast and excellent sensitivity to the tissue hemodynamic properties of fractional blood volume and blood oxygenation, PAVI, as a complement to 3D ultrasound and digital tomosynthesis mammography, might well contribute to detection, diagnosis and prognosis for breast cancer.

Acoustic attenuation imaging of tissue bulk properties with a priori information

Attenuation of ultrasound waves traversing a medium is not only a result of absorption and scattering within a given tissue, but also of coherent scattering, including diffraction, refraction, and reflection of the acoustic wave at tissue boundaries. This leads to edge enhancement and other artifacts in most reconstruction algorithms, other than 3D wave migration with currently impractical, implementations. The presented approach accounts for energy loss at tissue boundaries by normalizing data based on variable sound speed, and potential density, of the medium using a k-space wave solver. Coupled with a priori knowledge of major sound speed distributions, physical attenuation values within broad ranges, and the assumption of homogeneity within segmented regions, an attenuation image representative of region bulk properties is constructed by solving a penalized weighted least squares optimization problem. This is in contradistinction to absorption or to conventional attenuation coefficient based on overall insertion loss with strong dependence on sound speed and impedance mismatches at tissue boundaries. This imaged property will be referred to as the bulk attenuation coefficient. The algorithm is demonstrated on an opposed array setup, with mean-squared-error improvements from 0.6269 to 0.0424 (dB/cm/MHz)2 for a cylindrical phantom, and 0.1622 to 0.0256 (dB/cm/MHz)2 for a windowed phantom.

Error analysis of speed of sound reconstruction in ultrasound limited angle transmission tomography

We have investigated limited angle transmission tomography to estimate speed of sound (SOS) distributions for breast cancer detection. That requires both accurate delineations of major tissues, in this case by segmentation of prior B-mode images, and calibration of the relative positions of the opposed transducers. Experimental sensitivity evaluation of the reconstructions with respect to segmentation and calibration errors is difficult with our current system. Therefore, parametric studies of SOS errors in our bent-ray reconstructions were simulated. They included mis-segmentation of an object of interest or a nearby object, and miscalibration of relative transducer positions in 3D. Close correspondence of reconstruction accuracy was verified in the simplest case, a cylindrical object in homogeneous background with induced segmentation and calibration inaccuracies. Simulated mis-segmentation in object size and lateral location produced maximum SOS errors of 6.3% within 10 mm diameter change and 9.1% within 5 mm shift, respectively. Modest errors in assumed transducer separation produced the maximum SOS error from miscalibrations (57.3% within 5 mm shift), still, correction of this type of error can easily be achieved in the clinic. This study should aid in designing adequate transducer mounts and calibration procedures, and in specification of B-mode image quality and segmentation algorithms for limited angle transmission tomography relying on ray tracing algorithms.

MO‐D‐220‐10: Automatic Quality Control Processing for Detection of Elements Dropout in Ultrasound Transducers

Purpose: To develop a simple and reliable method of detection of lost or reduced sensitivity elements in diagnostic ultrasound arrays through analysis of the systematic features evident in appropriately acquired cineloops Methods: An algorithm was developed to detect the systematic features produced by element/channel signal loss evident in a cineloop acquired with a random background signal: A temporal median of the cineloop is produced followed by construction of a profile defined as the mean over depth [column‐wise mean] within a horizontal strip of the median image. A modified anisotropic diffusion filter is applied to the profile to isolate outlying features from characteristic noise, and simple statistical methods are then used to identify regions of the profile which exhibit characteristics of reduced sensitivity elements such as well‐defined, symmetric troughs. With an element occluded by fishing line to simulate dropout, cineloops were acquired with a GE‐Logiq 9 scanner either by rapidly scanning a transducer over a tissue mimicking phantom or by placing the transducer in a liquid phantom composed of an agitated cornstarch and water suspension. Results: Analysis with the algorithm clearly revealed the occluded section of the array as well as previously unnoticed reduced‐sensitivity elements. The efficacy of a simple and inexpensive method of detecting lost or reduced sensitivity elements of diagnostic ultrasound arrays was demonstrated in a confined dataset. The algorithm has been developed into an ImageJ plugin which presently handles uncompressed DICOM data containing rectangular scan regions; this should soon be made available for use by the medicalultrasound community after testing by Ultrasound Subcommittee members. Supported in part by the Ultrasound Subcommittee of the AAPM Science Council andNIH Grant CA91713

TU‐E‐201C‐09: Photoacoustic Tomography for Imaging in the Mammographic Geometry

Purpose: To add spectroscopicphotoacoustictomography (S‐PAT) to an ultrasound imagingsystem in the mammographic geometry to test whether this minimally invasive and ultimately inexpensive modality can detect and interpret small vascular anomalies, providing information supplemental to that of x‐ray and ultrasound much as is done with MRI. Blood oxygenation and blood volume information from S‐PAT might replace or exceed information provided by MRI resolution, possibly without contrast agents. Method and Materials: The specimen surface is flooded with a <25 ns pulse of light from a tunable, 720–900 nm plus 1064 nm, laser. An 8 cm diameter, 570 element PVDF array in a square grid of 3.2 mm spacing serves as the ultrasound receiver of the thermoacoustic signals produced by optical absorption. 20 channels were multiplexed to acquire the 570 signals. Delay‐and‐sum beamforming produces an image of the thermoacoustic sources. The receiver bandwidth, here 0.4 to ∼3 MHz at −12 dB is critical, as the center frequency from a spherical target is inversely proportional to the object diameter. A 3 mm ID tube of blood was imbedded in porcine tissues, 24mm of fat and 28mm of loin to the 725 nm illuminated surface and 45mm of loin to the receiver. 256 pulses of 70 mJ from a preliminary laser were averaged. Results: The blood tube could be easily discerned at the resolution anticipated for the acoustic frequency from the blood tube and the available receiver aperture. Conclusion: This described example is through optical and acoustical path lengths greater than that required for imaging from each side of most breasts at modest compression. This and similar results in vitro give hope of adequate sensitivity to blood collections in the breast of appropriate spatial frequencies. S‐PAT should be compatible for integration in a combined tomosynthesis/ultrasound/optical breast imagingsystem. Supported by R01_CA91713,