Abstract

This thesis investigates event-based sensing for hemodynamic measurements in transmission mode through the human finger. Conventional frame-based imagers are limited by exposure time, frame rate, and motion blur, which reduce sensitivity to fast optical transients. To address these bottlenecks, this thesis introduces a novel transmission-mode sensing framework using a Prophesee EVK4 HD neuromorphic sensor. By capturing asynchronous ON/OFF events generated by local intensity changes, this work presents a new approach to three tasks: intra-finger pulse transit time (PTT) and pulse wave velocity (PWV) estimation, event-based transmission mode laser speckle contrast imaging (LSCI), and Flicker Spike Arrival Time (FSAT) Delay Mapping, a novel method which allows event cameras to capture relative intensities through tissue.

The early chapters establish device-level and optical foundations, including address-event representation, logarithmic pixel response, modified Beer-Lambert behavior in tissue, and speckle decorrelation theory. The experimental chapters then present a common transmission-mode platform with near-infrared illumination and application-specific processing pipelines. For PTT/PWV, event streams are converted into longitudinal pulse waveforms, pulse feet are detected with intersecting-tangent timing, and beat-paired delays are used to estimate intra-finger transit time. For LSCI, temporal event binning with micro-seconds scale rolling window hops for variance-to-mean contrast statistics allow high virtual-frame-rate LSFI playbacks. For FSAT Delay Mapping, periodic illumination is used as a timing stimulus, and spatial latency differences in ON/OFF event responses are converted into FSAT Delay Maps that reflect optical attenuation and scattering.

Results demonstrate physiologically consistent intra-finger transit times for PTT, speckle-derived perfusion contrast within expected tissue ranges even at very low effective exposure times of 10ms, and FSAT latency maps that resolve clear vascular morphology. Measured PTT was $13.84 \pm 4.12$ ms. Computed PWV was 6.47 m/s. Event-based LSCI integration windows of 10 ms produced flow maps with relative contrast values of $0.231 \pm 0.087$. FSAT Delay Mapping provided the best structural clarity among all other event domain imaging methodologies tested and establishes a promising novel baseline warranting further investigation and improvement as an imaging technique.

These findings support event-based imaging as a single-sensor framework for combining high-temporal-resolution hemodynamic timing, speckle-derived perfusion contrast, and latency-based optical attenuation mapping. The thesis also documents implementation limits, including sparse per-pixel sampling at short integration windows, calibration dependence of latency-based inference, and the need for bias tuning for each specific use-case and experimentation modality. These challenges must be addressed before clinical translation.

Committee Chair

Shantanu Chakrabartty

Committee Members

Yiannis Kantaros, Chuan Wang, Christine O' Brien

Comments

This thesis develops a transmission-mode neuromorphic sensing framework using event cameras to measure hemodynamic signals, including intra-finger PTT/PWV, event-based LSCI/LSFI, and FSAT delay mapping, a novel methodology that exploits the intensity dependent latency as a surrogate for intensities in tunable illuminance/exposure settings.

Degree

Master of Science (MS)

Author's Department

Electrical & Systems Engineering

Author's School

McKelvey School of Engineering

Document Type

Thesis

Date of Award

Spring 2026

Language

English (en)

Author's ORCID

https://orcid.org/my-orcid?orcid=0009-0006-9999-0391

Download

Included in

Engineering Commons

Share

COinS