A microsystem for in vivo wireless monitoring of plastic biliary stents using magnetoelastic sensors

Biliary strictures due to benign disease or malignancy can obstruct bile flow, leading to serious complications including cholangitis and sepsis. Plastic biliary stents placed via ERCP are widely used to relieve obstructions but often occlude within months due to biofilm and sludge accumulation. Current clinical monitoring relies on indirect, lagging indicators such as elevated liver enzymes and imaging findings of biliary dilation, which do not directly assess stent patency and may only be performed after symptom onset. At occlusion onset, proteins and bacteria adhere to stent surfaces forming biofilm and sludge, increasing viscous mass. This work investigates a direct, wireless method to monitor stent patency using a miniaturized magnetoelastic resonant sensor integrated into the stent, where changes in resonant frequency and quality factor indicate mass accumulation. The study addresses challenges in sensor miniaturization and placement compatible with ERCP, achieving adequate wireless range while mitigating transmit-to-receive feedthrough, and ensuring clinical usability in vivo.

Magnetoelastic materials (amorphous alloys) convert magnetic fields to elastic strain and re-emit magnetic fields, underpinning passive wireless sensing and widely used in retail anti-theft tags. Prior sensing modalities for implanted environments include acoustic and optical detection, but optical methods are infeasible due to tissue occlusion and acoustic methods have limited range. Electromagnetic excitation and detection with inductive coils are therefore preferred. The authors previously demonstrated benchtop monitoring of sludge accumulation in stents using ribbon-shaped magnetoelastic resonators, correlating decreases in resonant frequency and quality factor with added mass. Prior attempts without robust packaging risked sensor damage during deployment. This work advances prior art by suppressing the frequency increase from miniaturization via structural mass loading using micromachined bias magnets, integrating a flexible 3D printed package for ERCP compatibility, and implementing time-domain decoupling to reduce drive feedthrough while maintaining sensitivity and range.

Sensor and package: The magnetoelastic sensor consists of a rectangular amorphous alloy foil (length 8.25 mm, width 1 mm, thickness 28 µm). Two permanent magnets (1 mm × 1 mm × 60 µm) are attached on the foil to provide DC magnetic bias and add structural mass to counteract the resonant frequency increase from miniaturization. The assembly is integrated into a customized 3D-printed polymeric package designed to protect the sensor during ERCP deployment while accommodating sharp longitudinal curvature as the stent passes through the endoscope elevator and guidewire introducer. The package restores the sensor to its natural shape upon deployment in the bile duct and fits within a standard plastic biliary stent. Interrogation subsystem: The transmit side includes a sinewave signal generator, power amplifier, and transmit coils arranged in a wearable belt configuration. The receive side comprises a receive coil, low-noise amplifier, overvoltage protection, and data acquisition with digital signal processing. To mitigate transmit-to-receive feedthrough, time-domain decoupling is employed: the sensor is briefly driven, then the transmitter is switched off and the ringdown response is sampled. This requires rapid transmit/receive switching and high sensitivity as the ringdown amplitude decays quickly. Digital signal processing steps are used to maximize SNR. Benchtop characterization: Prior to in vivo tests, the system was characterized for sensor orientation and position relative to the coil axes. Misalignment of the sensor axis by 45° reduced signal strength to 12%, and a 67.5° incline reduced it to 1%. Axial misplacement of the sensor 7.6 cm from coil center reduced signal strength to 1.5%. Stent-embedded sensors were also tested in air and water to establish baselines for resonant frequency and quality factor versus medium viscosity. In vivo procedure: Animal experiments were approved by the University of Michigan IACUC (protocol #6901). A 25 kg female domestic swine under general anesthesia underwent ERCP using a clinical side-viewing duodenoscope (Olympus TJF Type Q180V). The bile duct was cannulated with guidewire and catheter, and a cholangiogram obtained. The sensor-enabled plastic stent was deployed over the guidewire using a standard system (Oasis system, Cook Endoscopy, Winston-Salem, NC). After implantation, transmit and receive belt coils were wrapped around the abdomen for interrogation. Typical interrogation time T was 336 s (some tests used 340 s). Coil geometries used included elliptical belts with major×minor diameters of 10.75"×6.25" (27.3×15.9 cm), 12"×10.5" (30.5×26.7 cm), and 16.5"×13.5" (41.9×34.3 cm). Additional tests introduced an axial offset of 5 cm between coil center and sensor to emulate practical misalignment and used an inflated airbag to mimic larger abdominal circumference. Fluoroscopy confirmed stent placement.

- First successful wireless interrogation of a passive magnetoelastic sensor implanted in a live animal (swine). Range approximately 17 cm with interrogation time 336 s. - Reported SNR values: abstract notes ~10^6 at ~17 cm; discussion cites SNR ~10^7 with range exceeding 17 cm; a specific in vivo measurement in Fig. 2b reports SNR ~10% (reported as percentage). An offset-coil in vivo test reported SNR ~4×10^7. - Medium effect on resonance: resonant frequency decreased from 179.3 kHz (water) to 178.2 kHz (in vivo in bile); quality factor decreased from 178 (water) to 127 (in vivo), consistent with higher viscosity of bile. - Coil size variation (to mimic larger abdomen): increasing coil from 10.75"×6.25" to 12"×10.5" reduced SNR to 25% of baseline with ≤0.4 kHz change in resonant frequency and ≤2 change in Q. Further increase to 16.5"×13.5" reduced SNR to 15% of baseline with ~0.4 kHz frequency shift and Q change of ~15. Interrogation time 336 s. - Axial offset (5 cm): resonant frequency changed by only 0.3 kHz (0.17%) with a Q change of 30; SNR ~4×10^7; interrogation time 340 s. - Sensitivity to orientation/position (benchtop): 45° sensor-coil misalignment reduced signal to 12%; 67.5° to 1%. Axial displacement 7.6 cm from coil center reduced signal to 1.5%. - Time-domain decoupling effectively mitigated transmit feedthrough, enabling detection of the low-amplitude ringdown signal.

The study demonstrates a clinically relevant configuration that wirelessly interrogates a stent-integrated magnetoelastic sensor in vivo, overcoming challenges of miniaturization, placement, and electromagnetic feedthrough. Time-domain decoupling combined with carefully designed hardware and DSP achieved high SNR at ranges ≥17 cm while keeping the resonant frequency robust to coil size and alignment variations (≤0.17% shift). The observed reductions in resonant frequency and quality factor in vivo versus water validate sensitivity to the medium’s viscosity and, by extension, to mass/viscous loading associated with sludge accumulation. Practical tests simulating larger body habitus and coil misalignment showed significant SNR reductions but minimal impact on the measured resonant frequency, supporting reliability of patency assessment despite positioning variability. Lessons from earlier unsuccessful attempts highlighted the need for robust packaging to survive ERCP and a sophisticated interrogation subsystem for adequate sensitivity and switching performance. Overall, the system addresses clinical needs by enabling direct, noninvasive monitoring of stent patency, potentially allowing earlier, targeted interventions to prevent cholangitis and emergency procedures.

This work reports the first in vivo wireless interrogation of a passive magnetoelastic sensor integrated into a plastic biliary stent, achieving high SNR at clinically relevant ranges using a miniaturized sensor with integrated magnetic bias and a protective 3D-printed package. Time-domain decoupling and optimized hardware/DSP mitigated feedthrough and extended wireless range. Results establish that resonant frequency and quality factor shifts can be measured reliably in vivo and remain stable under coil placement variations. Future directions include: further sensor miniaturization; arrays of sensors with distinct resonant frequencies distributed along the stent to localize sludge; integrating sensors within stent walls to eliminate separate packages; and developing an inexpensive, unified, portable interrogation module to support expanded animal studies and clinical translation. The approach may generalize to other stent types (e.g., urethral, arterial) and could reduce reliance on invasive diagnostics, lowering costs and improving outcomes.

- Proof-of-concept demonstrated in a single swine specimen; broader validation is needed for generalizability. - Current interrogation relies on multiple research-grade hardware components and relatively long acquisition times (≈336–340 s), limiting portability and clinical workflow integration. - SNR degrades with increased coil size and misalignment, indicating sensitivity to positioning; benchtop tests showed strong dependence on orientation and axial placement relative to the coil. - Packaging is required to protect the sensor during ERCP; moving to arrays would multiply packaging complexity unless sensors are integrated into the stent wall. - Some earlier attempts failed due to packaging robustness and stent sizing, indicating procedural and design sensitivities that must be addressed for reliable deployment.