Selective imaging of adherent targeted ultrasound contrast agents

Abstract

The goal of ultrasonic molecular imaging is the detection of targeted contrast agents bound to receptors on endothelial cells. We propose imaging methods that can distinguish adherent microbubbles from tissue and from freely circulating microbubbles, each of which would otherwise obscure signal from molecularly targeted adherent agents. The methods are based on a harmonic signal model of the returned echoes over a train of pulses. The first method utilizes an 'image-push-image' pulse sequence where adhesion of contrast agents is rapidly promoted by acoustic radiation force and the presence of adherent agents is detected by the signal change due to targeted microbubble adhesion. The second method rejects tissue echoes using a spectral high-pass filter. Free agent signal is suppressed by a pulse-to-pulse low-pass filter in both methods. An overlay of the adherent and/or flowing contrast agents on B-mode images can be readily created for anatomical reference. Contrast-to-tissue ratios from adherent microbubbles exceeding 30 dB and 20 dB were achieved for the two methods proposed, respectively. The performance of these algorithms is compared, emphasizing the significance and potential applications in ultrasonic molecular imaging.

Figure 1

Flow chart for the image–push–image method detailing the process for detection of targeted contrast agents. I (input), original data contain echoes from free agents plus tissue; A, free agent signal is suppressed by a slow-time low-pass filter (s-LP) and an estimate of tissue signal is formed as in (11); B, following the radiation force pulse, the acquired data set includes tissue, free and bound agents as in (12); C, following the radiation force pulse and a slow-time low-pass filter, the data set represents an estimate of echoes from tissue plus bound agents as in (13); O (output), subtracting A and C yields an estimate of the bound agents as in (14).

Figure 2

Flow chart for the harmonic method detailing signal-processing steps. In the input (I), echoes from adherent targeted agents overlap those of tissue and circulating agents. Combination of filtering in fast time and slow time can provide additional information about: A, free and adherent agents in harmonics band; B, anatomical reference in the fundamental band and C, freely flowing agents only in harmonic band. Output data contain echoes from adherent agents only as in (18).

Figure 3

Echo correlation from simulations and experiments for flowing agents in a 200 μm vessel with a 60° beam–flow angle. The transmission centre frequency was either 2 MHz or 4 MHz and PNP was 210 kPa.

Figure 4

Echo correlation from experiments for adherent agents. The transmission centre frequency was either 2 MHz or 4 MHz, and PNP was 130, 210 or 290 kPa.

Figure 5

Images of a pair of tubes containing targeted contrast agents surrounded by tissue using the image–push–image method with PNP of 210 kPa and a transmission centre frequency of 4 MHz (a) and of 2 MHz (b). The CTR is 37 dB and 34 dB for the targeted tube, and 10 dB and 16 dB for the control tube in (a) and (b), respectively.

Figure 6

Images of a pair of tubes containing targeted contrast agents surrounded by tissue using the harmonic method with PNP of 210 kPa and a transmission centre frequency of 4 MHz. Images are shown for: (a) all contrast agents; (b) tissue background; (c) flowing agents only and (d) adherent only, as labelled A, B, C and O in figure 2, respectively.

Figure 7

Same as figure 6, except tubes have been rinsed to remove free agents.

Figure 8

Received harmonic signal power for free and adherent agents. (a) data for 2 MHz and 4 MHz, at PNP of 210 and 290 kPa and 10 mm s⁻¹. (b) data for 4 MHz, 210 kPa, and flow rates of 2.5, 5 and 10 mm s⁻¹.