Quantification of Endogenous Brain Tissue Displacement Imaging by Radiofrequency Ultrasound

Abstract

The purpose of this paper is a quantification of displacement parameters used in the imaging of brain tissue endogenous motion using ultrasonic radiofrequency (RF) signals. In a preclinical study, an ultrasonic diagnostic system with RF output was equipped with dedicated signal processing software and subject head-ultrasonic transducer stabilization. This allowed the use of RF scanning frames for the calculation of micrometer-range displacements, excluding sonographer-induced motions. Analysis of quantitative displacement estimates in dynamical phantom experiments showed that displacements of 55 µm down to 2 µm were quantified as confident according to Pearson correlation between signal fragments (minimum p ≤ 0.001). The same algorithm and scanning hardware were used in experiments and clinical imaging which allows translating phantom results to Alzheimer's disease patients and healthy elderly subjects as examples. The confident quantitative displacement waveforms of six in vivo heart-cycle episodes ranged from 8 µm up to 263 µm (Pearson correlation p ≤ 0.01). Displacement time sequences showed promising possibilities to evaluate the morphology of endogenous displacement signals at each point of the scanning plane, while displacement maps-regional distribution of displacement parameters-were essential for tissue characterization.

Keywords: brain; radiofrequency ultrasound; tissue displacements; transcranial sonography.

Figure 1

Structures of phantoms used in the evaluation of displacement detection system: (A) static structure (Model 532A); (B) structure with a harmonic motion phantom with displacement field (concentric dotted lines) excited in tissue-mimicking material (TMM).

Figure 2

Results of the detection algorithm evaluation of targeted radiofrequency (RF) signals from the static phantom (Model 532A): (A) B-scan of the phantom structure with targets of interest; (B) the detected positions of static targets presented in the form of inter-frame displacements; (C) the calculated variability of static target positions or displacements relative to the transducer. Mean and standard deviation calculated from n = 21, error bar illustrates standard deviation up to 0.2 µm.

Figure 3

Results of the detection algorithm evaluation on targeted RF signals from the harmonic motion phantom: (A) B-mode ultrasound image of the phantom with the TMM; (B) map of motion: root mean square of the accumulated displacement signal d_rms[p] represented in color scale. asterisks indicates the points in which weak (green asterisk) and intense (black asterisk) displacement signals were estimated; (C,E) displacement signals at weak and intense movements locations represented in the form of inter-frame displacements; (D,F) displacement signals at weak and intense movement locations represented in the form of accumulated displacements.

Figure 3

Results of the detection algorithm evaluation on targeted RF signals from the harmonic motion phantom: (A) B-mode ultrasound image of the phantom with the TMM; (B) map of motion: root mean square of the accumulated displacement signal d_rms[p] represented in color scale. asterisks indicates the points in which weak (green asterisk) and intense (black asterisk) displacement signals were estimated; (C,E) displacement signals at weak and intense movements locations represented in the form of inter-frame displacements; (D,F) displacement signals at weak and intense movement locations represented in the form of accumulated displacements.

Figure 4

Oscillations of accumulated displacements. at dominant frequency of 2 Hz in dynamic phantom: (A) areas where 2 Hz frequency was determined as the peak frequency in spectral analysis; (B) epochs (grey rectangles) are time windows of similar waveforms in the average signal of points with dominant frequency; similarity detected by reference-template matching technique; here six epochs were identified.

Figure 5

Confidence mapping of displacements in harmonic excited TMM: locations of confident (p ≤ 0.001) displacements root mean square in color scale. Asterisks in the graph indicates the points in which weak (green asterisk) and intense (black asterisk) displacement signals were estimated.

Figure 6

(A) The detected displacement signals for the 2 Hz (upper), 4 Hz (middle), 8 Hz (lowest) solenoid excitation frequencies; (B) detected displacements resulting from proportional changes in excitation amplitude of the solenoid. The dashed line represents the ideal theoretical case of detection. Excitation peak-to-peak voltages were 0.5 V, 1 V, 1.5 V, 2 V, 2.5 V, and 3 V at frequencies of 2 Hz and 8 Hz.

Figure 7

Assessment of endogenous motion in the coronal plane of cranium: (A,C) B-mode ultrasound image together with the outlined cross-sectional region of the hippocampus; (B,D) map of endogenous motion in the coronal plane of cranium: root mean square of displacement signal d_rms[p] represented in color scale. Hippocampus region outlined with the solid line. Parts (A,B) correspond to the subject with Alzheimer’s disease, parts (C,D) to the healthy elderly subject.

Figure 8

Hippocampus region-averaged displacements signals: (A,B) inter-frame displacements of the hippocampus; (C,D) accumulated displacements of the hippocampus. Parts (A,C) correspond to the subject with Alzheimer’s disease, parts (B,D) to the healthy elderly subject.

Figure 9

Displacement confidence evaluation by the Pearson correlation of accumulated displacements for a patient with Alzheimer’s disease. (A) Dominant frequency (here 71 BPM) determined as the most frequent value of peak frequency from points with spectral peak frequency within a 40–125 BPM interval; (B) epochs are time windows of similar waveforms in the average signal of points with dominant frequency; similarity was detected by reference-template matching technique, here seven epochs were identified; (C) Pearson correlation was calculated between each pair of signal segments from epochs that belonged to the same spatial point; here 7 × 6:2 = 21 pairs/correlations for every point. The minimum of Pearson correlation r (D) and maximum p-value (E) were used as the confidence criteria—the point was confident if minimum r > 0 and maximum p-value ≤ 0.01 from all 21 correlations for a particular point. In D and E the cross-sectional region of the hippocampus outlined with black solid line.

Figure 10

Confidence mapping of displacements in the coronal plane of the brain (upper images for a patient with Alzheimer’s disease, lower images for healthy elderly participant): (left) locations of confident displacements with annotated hippocampus ROI; (middle) spatial averaged accumulated displacements in the ROI, similar segments are in grey rectangles with red lines; (right) superimposed similar segments (tiny colorful lines) and their average (bold black line).