Real-Time On-Chip Time-of-Flight Detection for US Time Reversal Localization of Implants

Abstract

Time-reversal (TR) localization using ultrasound (US) has emerged as a promising method to localize mm-sized deep-tissue implantable medical devices (IMDs) without requiring precise prior positional knowledge. While in-vitro studies using commercial off-the-shelf (COTS) systems have validated its efficacy, practical deployment demands significant system miniaturization to enable low-power, wearable implementations. To this end, this paper presents a proof-of-concept ASIC architecture comprising 45 channels, each featuring an integrated receive and transmit chain. A key component is a novel peak-instance-detection (PID) circuit that detects the time of flight (ToF) of incoming US signals in real time via dynamic thresholding and that simultaneously digitizes the result, eliminating the need to store full waveforms. This enables TR with minimal power and memory overhead. The ASIC is implemented in the TSMC 130-nm BCD+ technology. Electrical measurements confirm accurate ToF detection from recorded US backscatter signals. The PID reduces the per-channel receive power to 100 μW – achieving a 2000X reduction over COTS alternatives. Memory usage is also minimized, requiring only 540 bits to store ToFs across all channels – 746X lower than COTS approaches. Furthermore, the on-chip ToF digitization simplifies the TR computation, reducing it to a digital subtraction operation, in contrast to the complex convolutions typically employed. Finally, in-vitro experiments in heterogeneous scattering media using a 3D-printed human rib phantom demonstrate a 22X power efficiency improvement over unfocused US transmission, while matching the localization performance of MATLAB-based COTS implementations. These results highlight the feasibility of the proposed architecture as a miniaturized platform for TR-based IMD localization.