Sonic Estimation of Elasticity via Resonance: A New Method of Assessing Hemostasis

Abstract

Uncontrolled bleeding threatens patients undergoing major surgery and in care for traumatic injury. This paper describes a novel method of diagnosing coagulation dysfunction by repeatedly measuring the shear modulus of a blood sample as it clots in vitro. Each measurement applies a high-energy ultrasound pulse to induce a shear wave within a rigid walled chamber, and then uses low energy ultrasound pulses to measure displacements associated with the resonance of that shear wave. Measured displacements are correlated with predictions from finite difference time domain models, with the best fit corresponding to the modulus estimate. In our current implementation each measurement requires 62.4 ms. Experimental data was analyzed using a fixed-viscosity algorithm and a free-viscosity algorithm. In experiments utilizing human blood induced to clot by exposure to kaolin, the free-viscosity algorithm quantified the shear modulus of formed clots with a worst-case precision of 2.5%. Precision was improved to 1.8% by utilizing the fixed-viscosity algorithm. Repeated measurements showed a smooth evolution from liquid blood to a firm clot with a shear modulus between 1.4 and 3.3 kPa. These results show the promise of this technique for rapid, point of care assessment of coagulation.

Keywords: Coagulation; FDTD; Resonance; Shear modulus; Shear waves; Ultrasound.

Figure A-1

FDTD model staggered grid geometry.

Figure 1

Schematic representation of the SEER method. SEER begins with data acquisition, during which a high intensity ultrasound pulse applies acoustic radiation force to induce a shear wave that resonates within the test chamber. A series of low energy sensing pulses are used to acquire ultrasound echoes from the moving sample. These echoes are processed to form motion estimates indicative of the resonating shear wave. Equations in the ‘Motion Estimation’ panel explain how time-delay estimates are converted into displacements estimates. Motion estimates are compared to numerical or analytical models to find the shear modulus (and possibly other mechanical properties) that best fits the observed motion. This process is repeated every few seconds to measure the evolving shear modulus resulting from coagulation.

Figure 2

Prototype Quantra^™ cartridge. This four-channel cartridge incorporates a warming chamber, volume control chambers to set sample volume, liquid reagents, and four independent test chambers. Ultrasound is coupled in from the opposite face.

Figure 3

Acoustic path and test chamber geometry with overlaid sensing beam (left) and forcing beam (right). The displayed sensing beam is the square of the two-way sensitivity function. The displayed forcing beam is the energy of transmitted forcing beam. Ultrasound beams were modeled using FIELD II (29) with acoustic properties for blood taken from (30).

Figure 4

Experimental test chamber geometry (left) and FDTD model geometry (right). Gray regions indicate the full width at half maximum of the modeled ultrasound forcing beam. All dimensions are in mm.

Figure 5

Representative time-displacement estimates formed over approximately 30 minutes from clotting blood. The displacements induced by shear resonance superficially resemble the over-damped response from a second order system.

Figure 6

Experimental data analyzed via the free-viscosity algorithm. The top five panels depict experimental time-displacement data (black) overlying best-fit FDTD models (gray). Working downward, the large panels depict estimated modulus, estimated viscosity, and the correlation between the best-fit model and the experimental data. The dotted line in the correlation curve is a threshold value used to reject poor quality estimates.

Figure 7

Estimated moduli across six different blood samples using the free-viscosity algorithm. Each panel depicts the modulus of each of four channels as a different colored curve. The CV was computed using a one-minute average value centered 20 minutes after the reagents first contacted the blood sample. Variations between samples are to be expected because they were taken from different donors and subjected to different handling.

Figure 8

Contour plots depicting the normalized correlation between the free-viscosity model and experimental data for measured shear moduli of 0.5, 1.75, and 3.0 kPa.

Figure 9

Experimental data analyzed via the fixed-viscosity algorithm. The top five panels depict experimental time-displacement data in black overlying best-fit FDTD models in gray. Working downward, the large panels depict estimated modulus and the correlation between the best-fit model and the experimental data. The dotted line in the correlation curve is a 0.75 threshold value used to reject poor quality estimates.

Figure 10

Estimated moduli across six different blood samples using the fixed-viscosity algorithm. Each panel depicts the modulus of each of four channels as a different colored curve. The CV was computed using a one minute average value centered 20 minutes after the reagents first contacted the blood sample. Samples labeled ‘Fresh’ were drawn into citrated vacutainers and tested within four hours. Samples labeled ‘Refrig.’ were also drawn into citrated vacutainers but were refrigerated overnight before warming and testing.

Figure 11

Estimated moduli from two blood samples using the fixed-viscosity algorithm with a modified acquisition sequence to assess the impact of acoustic radiation force on clot formation. Channels 3 and 4 were dormant until 15 minutes after mixing. These results show that ultrasound radiation force has no significant impact on clot formation.