Size-Selection for Sensitivity Improvement

Preferential selection of the microbubble size improved our ability to sensitively de- tect targeted microbubbles in USMI. We demonstrated a several-fold improvement in

contrast enhancement by tailoring MCA size distributions using centrifugation. This is significant in instances where the microbubble targeting is low and maximum sensi- tivity is required. For instance in a response to therapy study, as animals are treated over time and biomarker expression is diminished, it is imperative to have low-level sensitivity to targeted MCAs. In addition, we discovered that we could achieve greater contrast sensitivities with fewer injected microbubbles. Of course, this is advantageous for minimizing the exposure to foreign particles. Finally, size-selection has assisted in our ability to generate images by illustrating spatial distribution of targeting well above the noise floor.

While microbubble size-selection has proved to be invaluable in pre-clinical studies, there are still limitations and obstacles that must be overcome for clinical advancement. First, it must be noted that centrifugation offers a substantial improvement in produc- tion yields and MCA stability over techniques such as microfluidics. Unfortunately, only a few size distributions may be extracted by using the centrifugation technique. Thus, in applications where matching the resonant frequency to the microbubble diam- eter is important, there is limited flexibility. Though, in my experience, the scattering cross-section may be more significant than matching the resonant frequency to the microbubble diameter, which should be explored in future characterization studies. Another detriment to the centrifugation method for sorting is that the process is time- consuming and the yield (resultant number of microbubbles) is relatively low, though better than other methods. Finally, mass-production for clinical implementation may be difficult due to handling issues and maintaining a sterile environment.

Ideally, a monodisperse population of microbubbles is desired for a number of rea- sons. This includes an improvement in our ability to quantify biomarkers, better pre- dictability for in vivo MCA persistence, and exactly matching the resonant frequency of the system to the size of the MCA. Unfortunately, the resultant size distribution

with centrifugation is not monodisperse. Thus, any future desire to correlate intensity to the number of microbubbles (or biomarkers) will be challenging. However, since backscatter is related to the insonation frequency, it may be possible to determine size related information by scanning the targeted microbubbles at various frequencies and implementing a compounding technique. Finally, since maximum translation of mi- crobubbles using ARF is at the resonant frequency, a wide range of microbubble sizes makes ARF-enhanced molecular imaging less efficient and less predictable.

As mentioned previously, large microbubbles may be more susceptible to detach- ment than smaller microbubbles, due to greater in vivo shear forces. Thus, larger microbubbles may attach and subsequently detach within the time that we wait for free-flowing MCAs to clear the circulation. Currently, our USMI procedure does not account for this situation. However, real-time approaches to USMI are being evaluated to determine receptor expression over time for a more accurate evaluation.

13.2.3 3-D for Improved Quantification

In Chapter 9, we used volumetric USMI to illustrate the heterogeneity of αvβ3 in a

tumor, thus emphasizing the necessity of a 3-D approach. Furthermore, we showed that a traditional 2-D USMI study may misrepresent the angiogenic expression by as much as 28%. Moreover, 2-D USMI studies that evaluate biomarker expression over time are susceptible to error. Maintaining the exact 2-D slice position and orienta- tion are essential for these types of studies. Thus, error increases substantially when the transducer is removed from a fixed clamp and repositioned. A volumetric USMI approach, however, eliminates the possibility for this type of error.

Volumetric USMI is clearly the more effective way to quantify biomarker expression as compared to an equivalent 2-D study. The setup that was used in our study was a 1-D linear array that was positioned in a fixed clamp and scanned elevationally

to produce a stacked volume of images. This implementation was a requirement for pre-clinical studies due to the way in which we quantify the targeted MCAs at each discretized step. Unfortunately, the implementation of this type of setup in a clinical environment would be very difficult, if not impossible. Thus, due to the lack of clinical 3-D ultrasound probes with contrast detectability, clinical translation of volumetric USMI may not be feasible, which limits its integration to pre-clinical studies. As 3- D contrast-enhanced clinical probes become more prevalent, this will no longer be an impediment.

As 3-D clinical probes become available with contrast-enhanced detection schemes, it will be important to evaluate the sensitivity of 3-D USMI relative to a 2-D approach. For instance, due to the increased number of transducing elements in the elevational direction, 3-D probes have the ability to focus and steer in both the lateral and eleva- tional directions. While the resolution of the transducer in the elevational direction is greatly improved, the sensitivity in detecting larger number of microbubbles may be compromised due to the smaller beam that interrogates the contrast agents. However, due to the improvement in sensitivity by size-selecting microbubbles, this may be a non-issue. Regardless, there may be a trade-off between resolution and sensitivity to large populations of microbubbles moving from a 2-D to 3-D USMI approach, which must be explored as 3-D probes become available for USMI.