|University of Illinois at Urbana-Champaign | Department of Electrical and Computer Engineering | Department of Bioengineering
Department of Statistics | Coordinated Science Laboratory | Beckman Institute | Food Science and Human Nutrition | Division of Nutritional Sciences | College of Engineering
|Friday, August 23rd, 2019|
Superresolution Imaging and Beamforming
Abundant research demonstrates that early detection of cancer leads to improved patient prognosis. By detecting cancer earlier, when tumors are in their primary stages, treatment can be applied before metastases have occurred. For example, in breast cancer the five-year survival rate of patients treated for cancer when the tumor was confined to the breast was 98.6% compared to 23.3% for patients whose cancer had metastasized to other tissue regions.
The presence of microcalcifications (MCs) is indicative of malignancy in the breast and improving the ability to detect MCs with modern imaging technology remains an open question. The presence of MCs is associated with presence of cancer in the breast, i.e., 30-50% of all nonpalpable breast cancers detected using mammograms are based on identifying the presence of MCs [Gul03, Mor05]. Therefore, improving the sensitivity of imaging techniques to detect MCs in the breast will provide an important role for the early detection and diagnosis of breast cancer.
Robust detection of MCs with diagnostic ultrasound imaging has remained a goal for many years because ultrasound is portable, real time and its lack of ionizing radiation means that ultrasound is considered safe for multiple imaging sessions. Two reasons that ultrasound has not historically performed well at detecting MCs is the loss of contrast due to speckle surrounding the MCs and the poor spatial resolution, i.e., MCs are typically much smaller than the beam from an ultrasonic transducer. Recently developed ultrasonic algorithms, i.e., MicroPure, intended for the detection of MCs in the breast did not provide improved detection over mammography. Therefore, there has been reduced enthusiasm for pursuing ultrasonic approaches to MC detection.
Recently, we developed a novel nonlinear beamforming technology for ultrasonic arrays that provides super resolution of ultrasonic images (up to 100 times improvements in lateral resolution). The beamforming technique, called null subtraction imaging (NSI), utilizes nulls in the beam pattern to create images using ultrasound. Because the nulls fall off much faster than the main lobe of a beam for an ultrasonic transducer, the lateral resolution achieved through imaging with the nulls is much narrower than imaging with the beam main lobes. The width of the beam main lobes is diffraction limited and, therefore, imaging with these beams limits the lateral resolution. By imaging with the nulls we beat the diffraction limit. Furthermore, the lateral resolution gains provided by NSI are accompanied by a reduction in side lobes present in all beam patterns and increases in the signal-to-noise ratio (SNR).
The tradeoff associated with these improvements is a suppression of speckle compared to singular targets in the field. While this means that anechoic and hyperechoic targets may lose contrast, small bright targets, such as MCs, and interfaces between tissues will become more highlighted. Therefore, we hypothesize that the NSI approach will provide superior performance for the specific imaging task of detecting small specular scatterers in the field. For the imaging tasks of identifying and quantifying MCs, NSI appears to be ideally suited. Figure 1. Provides a simple diagram showing how we implement NSI using apodization functions with zero mean (causing a null at broadside) and with a slight DC bias causing a small bridge across the null. By subtracting out the null you get an NSI beam as shown in Fig 2. Figure 3 shows some experimental images of targets using conventional delay and sum beamforming and our NSI approach, which resulted in as much as a 100 fold improvement in lateral resolution. Figure 4 shows an image of two wire scatterers spaced closely together. NSI can differentiate their locations whereas conventional beamforming does not. The final figure (Fig 5) shows a contrast target using both beamforming techniques.
This work was support be a grant from the NIH (R21EB024133)
(Left) Rectangluar apodization and (right) apodizations of zero mean and zero means with dc bias.
Beam patterns for rectangular apodization, NSI and MV.
B-mode Images from wire targers in ATS phantom using rectangular apodization (left) and NSI (right).
Two target resolution criteris using rectangular apodization and NSI.
Anechoic targets imaged using rectangular apodization and NSI.
BRL Projects >>
|Bioacoustics Research Lab.|