Ultrasound localization microscopy to image and assess microvasculature in a rat kidney

Abstract

The recent development of ultrasound localization microscopy, where individual microbubbles (contrast agents) are detected and tracked within the vasculature, provides new opportunities for imaging the vasculature of entire organs with a spatial resolution below the diffraction limit. In stationary tissue, recent studies have demonstrated a theoretical resolution on the order of microns. In this work, single microbubbles were localized in vivo in a rat kidney using a dedicated high frame rate imaging sequence. Organ motion was tracked by assuming rigid motion (translation and rotation) and appropriate correction was applied. In contrast to previous work, coherence-based non-linear phase inversion processing was used to reject tissue echoes while maintaining echoes from very slowly moving microbubbles. Blood velocity in the small vessels was estimated by tracking microbubbles, demonstrating the potential of this technique to improve vascular characterization. Previous optical studies of microbubbles in vessels of approximately 20 microns have shown that expansion is constrained, suggesting that microbubble echoes would be difficult to detect in such regions. We therefore utilized the echoes from individual MBs as microscopic sensors of slow flow associated with such vessels and demonstrate that highly correlated, wideband echoes are detected from individual microbubbles in vessels with flow rates below 2 mm/s.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Figure 1

Flowchart of the acquisition and processing. Imaging of the rat kidney was performed under anesthesia (free breathing) using a compact clinical imaging array transmitting at 6.9 MHz (note that the imaging array and the kidney are not drawn to scale). An ultrasound research platform was used to acquire a stack of 40000 frames at 300 Hz, with a dedicated sequence outputting both regular B-mode images and microbubble specific Coherence Factor Contrast Pulse Sequencing (CFCPS) images. Processing of the stack allowed us to track and correct for the organ motion (B-mode) and to localize and follow the microbubble positions over time (CFCPS). After correction of the MB positions, combining all of the localizations reveals the microvascular network and associated flow.

Figure 2

Motion detection and classification of the frames. (a) B-mode images were used to select frames and classify them in cycles and to detect physiological motion of the kidney. (b) Selection and classification of frames was performed using frame-to-frame correlation over a region-of-interest (blue dashed line) showing specular reflection and displayed here over a small time window. Respiratory motion was detected when decorrelation was prominent and associated frames were discarded from further analysis. Frames recorded between 2 successive breaths were grouped as one cycle. Small intra-cycle changes in correlation are related to cardiac pulsation from the descending aorta. (c) The intra-cycle motion (axial and lateral translation and rotation) over the entire kidney as a function of time shows the cyclic cardiac pulsing. (d) The inter-cycle motion shows the small changes of the kidney position between breaths (cycle 50 used as the reference).

Figure 3

Image of the rat kidney microvasculature. (a), (b) Density map of the microbubbles positions without (a) and with (b) motion correction revealing the different vascular structures in the kidney (C: cortex, OM: outer medulla, IM: inner medulla). (c–f) Zoom over the distinctive cortex and outer medulla network without (c), (d) and with (e), (f) motion correction. Scale bar represents 2 mm in (a), (b) and 500 µm in (c–f).

Figure 4

Velocity and direction of the in-plane microvascular flow represented by individual trajectories of MBs identified with a velocity smaller than 2 mm/s. (a) MB position density map over a region of interest (ROI). (b) In the same ROI and for a single cycle, the positions of the MBs are displayed as a function of time (colorbar: time of arrival in s). (c) Reconstructed average trajectory and velocity from the data in (b) (colorbar: velocity in mm/s). Trajectory was reconstructed using a nearest-neighbor method. (d) All of the trajectories with a velocity smaller than 2 mm/s detected for the dataset (colorbar: velocity in mm/s). The vast majority of the trajectories were found in the medulla. (e) Average direction of the blood flow in the microvascular network. The cyclic colorbar indicates the trajectory direction in radians. Most trajectories in the medulla are found to follow the descending vasa recta (outer to inner medulla). Scalebar represents 500 µm in (a), (b) and (c) and 2 mm in (d) and (e).

Figure 5

Echoes from individual MBs can be tracked over hundreds of frames in small vessels. (a) Diagram of the microbubble motion and the echoes after alignment to remove the trend in position. (b) Density map showing the location of the regions of interest (ROI) displaying vessels with slow blood flow. (c), (d), shows a combined view of the ROI with a CFCPS frame superimposed showing an isolated MB (appearing as the PSF of the imaging system) that was tracked over numerous frames, where the scale bar represents 500 µm. (e), (f), Time traces of the tracked MB as a function of the frame number. The focus time indicates the fixed position in tissue along the MB path after correcting for tissue motion. For reference, at 6.9 MHz (transmit frequency) one period equals 0.14 µs. (g) Correlation of the echoes over a large number of frames is evidenced if a single echo is used as a reference (indicated by the red arrow). (h) Average spectrum of the traces displayed in (e) with a peak at the fundamental transmission frequency.

Figure 6

By comparison with isolated MBs, tissue background or larger vessels do not remain correlated over a large number of frames. (a) Diagram of the microbubble motion and the echoes that are shown in this figure in grey scale after alignment. (b) Density map showing the location of the regions of interest over a larger vessel (c) and outside of the kidney (d). In (c) and (d), the arrow indicates the position where time traces are visualized and the scale bar represents 500 µm. (e) Short-time correlation (over a few frames) is found in the vessels characterizing faster MB velocities and (f) noise is observed over time in the tissue. In (g), fast decorrelation of the echoes is evidenced if a single echo is used as a reference (indicated by the red arrow).