Medical tech: the biggest innovations of the last ten years, and the marine physics hiding inside them

Medical technology has had a remarkable decade. Continuous glucose monitoring moved from clinic to wrist. Transcatheter valves made open-heart surgery optional for many patients. Ultrasound shrank to a probe that plugs into a phone. Hyperbaric medicine quietly became an evidence-based discipline. What interests an engineer about all of these is how much of the underlying physics is shared with the instruments we lower into the Severn. Here are five examples.

Ultrasonic phase-shift blood-flow sensors and the ADCP

The biggest practical innovation in vascular medicine of the last ten years is arguably not a drug or a device but the collapse in the cost and size of Doppler ultrasound. A handheld probe can now measure blood velocity in a carotid artery by sending a pulse of ultrasound, receiving the echo from the moving red cells, and measuring the phase shift between transmitted and received signal. Velocity is proportional to that phase shift divided by the time between pulses, and the sign tells you the direction.

An acoustic Doppler current profiler, the ADCP that we mount on the estuary bed, does precisely the same thing with the same equation. It sends pulses from three or four angled transducers, receives echoes from the suspended sediment and plankton that move with the water, measures the phase shift in each beam, and resolves the three components of current velocity in a stack of range-gated cells up through the water column. The frequency is lower, around 300 to 1200 kilohertz instead of several megahertz, because seawater absorbs sound more with depth and we are looking through 30 metres rather than 3 centimetres. The scatterers are silt instead of blood cells. The aliasing problem, where a velocity above the instrument's ambiguity limit folds back and reads as a flow in the wrong direction, is identical, and so is the fix, which is to use a pair of pulse spacings and unwrap the phase. The constituents on our main sheet were reduced from exactly these phase-shift velocity records, plus a pressure gauge for the level.

Hyperbaric diving medicine and fish passage through a turbine

Hyperbaric oxygen therapy, long a treatment for decompression illness, has in the last decade accumulated solid evidence for diabetic foot ulcers, radiation injury and carbon monoxide poisoning, and the chambers have moved from naval bases into ordinary hospitals. The underlying physics is the solubility of gases in tissue as a function of pressure, and the rate at which a dissolved gas comes out of solution as the pressure falls. Decompression tables are a model of that rate, and the central rule is that it is the rate of pressure change, not the total change, that does the harm.

A fish passing through a low-head bulb turbine experiences the same problem in a tenth of a second. The pressure on the suction side of a runner blade can drop below atmospheric and recover within a few blade chords, and a fish with a gas-filled swim bladder undergoes a decompression that no diving table would permit. The injury is barotrauma, the same word the hyperbaric physicians use. The fish-friendly runner we describe on our main sheet exists because turbine designers borrowed the nadir-pressure and rate-of-change limits from diving medicine, measured them for salmonids in a pressure chamber, and redesigned the blade gap and the draft tube so that the pressure trajectory stays inside the survivable envelope. The chamber used for those fish tests is, in its working principle, a hyperbaric treatment chamber run in reverse.

Flow-diverting stents and the draft tube problem

The transcatheter heart valve and the flow-diverting stent for cerebral aneurysms are two of the decade's defining devices, and both are flow-control problems. A flow diverter is a fine mesh tube placed across the neck of an aneurysm so that the blood in the sac slows, stagnates and clots harmlessly, while the main vessel keeps its flow. The design question is how much of the flow to divert without starving the side branches, and the answer comes from computational fluid dynamics on patient-specific geometry, run against the pressure and shear thresholds that the vessel wall can tolerate.

A turbine draft tube is a diffuser that recovers kinetic energy from the water leaving the runner by slowing it down gradually. Slow it too fast and the flow separates from the wall, the pressure recovery is lost and a swirling, pulsing rope forms along the axis that shakes the machine and erodes the lining. The CFD methods used to design a draft tube that avoids separation are, with different boundary conditions, the methods used to design a stent that avoids starving a branch artery. We have had conversations with vascular modellers about meshing strategy that would be unremarkable in either discipline. The common enemy is adverse pressure gradient, and the common tool is a well-resolved wall layer and an honest turbulence model.

Wearable pressure and glucose sensing and the bed-mounted gauge

Continuous glucose monitoring is the most used medical sensor innovation of the decade, and the engineering achievement underneath it is not the chemistry but the signal processing: a small, drifting, noisy electrochemical signal, sampled every few minutes and turned into a trustworthy reading through calibration against a known reference and a model of the sensor's lag and drift. Wearable blood pressure cuffs and implantable pulmonary artery pressure sensors have the same shape of problem, with the added difficulty that the reference is hard to obtain.

A bed-mounted tide gauge is a pressure sensor left in salt water for a year. It drifts with temperature, with biofouling, and with the slow creep of its own diaphragm. It is calibrated before deployment against a known head and checked on recovery against a surveyed level, and the data between those two points is corrected with a drift model that looks very much like the one inside a glucose monitor. Both instruments are telling you about a slowly varying quantity through a sensor whose own behaviour is changing, and both depend on the discipline of recording the calibration rather than the discipline of the sensor.

Computational flow imaging

Four-dimensional flow magnetic resonance imaging, which measures the full three-dimensional velocity field of blood in the heart and great vessels over a cardiac cycle, has moved from research to clinical use in the last decade. It produces a time-resolved velocity dataset that is, in form, exactly what a moored ADCP array produces for an estuary cross-section over a tidal cycle. The analysis tools have converged too: the same vortex identification criteria, the same energy loss calculations, the same streamline visualisations. We have used a wall-shear-stress plotting routine written for aortic flow to show scour risk at a caisson corner. It needed only the units changing.

The common lesson

The last ten years of medical technology have been, in large part, a decade of fluid mechanics instruments becoming small, cheap and trustworthy. The decade before that was when the same thing happened to marine instruments. The lesson that carries both ways is that the sensor is never the hard part. The hard part is the calibration, the datum and the honesty about what the model does not know. We quote tidal levels to Chart Datum for the same reason a cardiologist quotes a pressure gradient against a stated reference: so that the next person can check the work.

Our own checkable work is on the home sheet. The feasibility desk shows the constituents, the schedule and the yield calculator that these instruments feed, and it is built to be argued with.