3D-Printed Acoustic Masks

Clinicians pick the probe to suit the depth. Linear arrays for the near field, convex probes for the far field. In both cases the transmit is focused with electronic delays, and that focusing caps the frame rate at roughly 50 fps, because every focal zone needs its own transmit event.

Plane wave imaging removes the cap by firing every element at once, but it gives up the focused transmit beam and the lateral resolution that comes with it. For a wearable system the trade is worse still: delays mean transmit events, transmit events mean time, and time on a moving body means motion artefact.

There is a second problem underneath the first. A rigid probe strapped to a curved surface makes imperfect edge contact, and the hydrogel filling the gap dehydrates within hours. Elastomer couplants last longer but damp the deep-tissue return. Continuous imaging with a conventional couplant is measured in hours, not days.

The mask is a rectangular polylactic acid shell, 60 mm by 25 mm by 31 mm, printed on an Ultimaker 2+ by fused deposition modelling with a 0.40 mm nozzle at 240 °C. PLA was chosen because it is non-reactive with the gel, rigid enough to hold the boundary of the lens, and biocompatible enough for prototyping against skin.

Its inner face is not flat. It carries a concave arc of radius 53.88 mm, a value taken from the range of human neck diameters. That arc is the whole optical design: with the fluid faster than tissue, a concave interface converges the transmitted plane wave instead of diverging it.

The curved cavity is coated with a thin latex sheet, then filled with the non-Newtonian gel using a squeeze bottle and levelled with a glass rod to the brim. Volumetric measuring cylinders keep the fill quantity uniform across lenses, which is what makes lenses of different radii comparable to each other at all.

The filling is a saturated solution of polymerised sodium polyacrylate, a non-Newtonian fluid. It was chosen against three competing requirements that a solid lens cannot satisfy at once.

It has to be faster than tissue, or the concave interface diverges rather than converges. Measured by the direct speed-of-sound method at 7.6 MHz and 20 °C, it runs at 1595 m/s against the 1540 m/s assumed for tissue. Solid lenses reach a far higher speed of sound, up to 2680 m/s in reported designs, but they pay for it in impedance mismatch.

It has to be quiet. Attenuation of 0.238 dB/cm-MHz and an acoustic impedance of 2.0798 MRayls sit closer to tissue than the solid lenses in the literature, so less energy is lost reflecting off the lens boundary on the way in and on the way back.

It has to stay put. Its rheology holds the shape of the cavity, holds contact at the probe edges on a curved surface, and resists drying. That last property is what turns a benchtop trick into something wearable.

The convergence was checked first in simulation with k-Wave, recording maximum beam pressure over the axial and lateral planes for a 128-element array with and without the mask. The pre-focal region develops the tapered high-pressure zone that a converging wavefront produces, and it is absent without the lens.

Note what the reflected path does. The insonified wave converges, but the echoes returning through the gel-to-medium interface diverge, which is the transmit case run in reverse. The receive beamformer has to account for the refraction at the arc, so the two-way travel time is computed as three segments: through the lens at c_l, out to the scatterer at c_m, and back.

The experimental chain used an L11–5v linear probe on a Verasonics Vantage 128-channel system at 7.6 MHz centre frequency. Transmit was 51 plane waves steered from −18° to +18° in 0.72° steps, reconstructed with delay-and-sum. The targets were a wire phantom in water, a CIRS General Purpose Ultrasound Phantom Model 054GS, and in vivo the human carotid artery, upper arm and index finger.

No delay adjustment was applied to compensate for the known speed of sound inside the lens region. The authors note this would improve image quality further, which means the numbers below are a floor rather than a ceiling.

Three masks were fabricated at 53.88, 28.80 and 19.20 mm radius, printed in distinct PLA colours purely so they could be told apart on the bench. Across the sweep, clutter falls and intensity rises as the radius shrinks, which is the behaviour the geometry predicts and a useful handle for focused ultrasound applications. Contrast-to-noise ratio peaks at R = 19.20 mm and 7.60 MHz.

A lens that images well for an hour is a lens, not a wearable. The dehydration test put 5.00 g of the non-Newtonian gel and 5.00 g of conventional acoustic gel in identical printed masks, held at 20 °C with minimal airflow, and weighed both daily for ten days.

After ten days the fluid lens had lost 13.60 % of its mass. The acoustic gel had lost 50.40 %. In the parallel imaging series the difference shows up as a hard stop: B-mode images could be acquired with conventional gel only through day four, and from day five the gel had dried past usability. Lateral resolution and contrast-to-noise ratio for the non-Newtonian gel stayed essentially flat across the whole ten days.

The same fluid was also tested as a couplant in its own right, filling the gap between a rigid probe and a curved surface. On moulds imitating human curvature, lateral resolution improved by 35 % over conventional gel at the R 58.33 configuration, with a 3.5 % higher density that improves edge contact and makes trapped air bubbles less likely. Skin compatibility was checked over 24 hours of continuous contact, with no irritation, lesion, pigmentation or oedema, at a neutral pH of 7.

The design ignores refraction and diffraction at the spherical geometry for simplicity. That simplification has a price: paraxial error across a spherical surface produces destructive interference between reflections arriving from different depths. Paraxial approximations in the design, or a metastructure lens in place of the plain arc, are the stated routes out.

The receive path leaves something on the table too. No delay adjustment compensates for the known speed of sound inside the lens, though the value has been measured. Folding it into the beamformer is a software change with no hardware cost.

The direction of travel is the same one the flexible array work is on. A passive part that upgrades an existing probe, rather than a new probe, keeps the wiring, the power budget and the frame rate exactly where they were.

In vivo imaging in this work was performed in line with the principles of the Declaration of Helsinki, and the coupling-media tests followed National Institutes of Health procedures.