arXiv subjects

Marwan Besrour

Publications and source records attributed to Marwan Besrour.

5 recordsLinked to original sources

arXiv2026

Learning to See via Epiretinal Implant Stimulation in silico with Model-Based Deep Reinforcement Learning

Objective: Diseases such as age-related macular degeneration and retinitis pigmentosa cause the degradation of the photoreceptor layer. One approach to restore vision is to electrically stimulate the surviving retinal ganglion cells with a microelectrode array such as epiretinal implants. Epiretinal implants are known to generate visible anisotropic shapes elongated along the axon fascicles of neighboring retinal ganglion cells. Recent work has demonstrated that to obtain isotropic pixel-like shapes, it is possible to map axon fascicles and avoid stimulating them by inactivating electrodes or lowering stimulation current levels. Avoiding axon fascicle stimulation aims to remove brushstroke-like shapes in favor of a more reduced set of pixel-like shapes. Approach: In this study, we propose the use of isotropic and anisotropic shapes to render intelligible images on the retina of a virtual patient in a reinforcement learning environment named rlretina. The environment formalizes the task as using brushstrokes in a stroke-based rendering task. Main Results: We train a deep reinforcement learning agent that learns to assemble isotropic and anisotropic shapes to form an image. We investigate which error-based or perception-based metrics is adequate to reward the agent. The agent is trained in a model-based data generation fashion using the psychophysically validated axon map model to render images as perceived by different virtual patients. We show that the agent can generate more intelligible images compared to the naive method in different virtual patients. Significance: This work shares a new way to address epiretinal stimulation that constitutes a first step towards improving visual acuity in artificially-restored vision using anisotropic phosphenes.

cs.LG↗

arXiv2026

A 1.6-fJ/Spike Subthreshold Analog Spiking Neuron in 28 nm CMOS

The computational complexity of deep learning algorithms has given rise to significant speed and memory challenges for the execution hardware. In energy-limited portable devices, highly efficient processing platforms are indispensable for reproducing the prowess afforded by much bulkier processing platforms. In this work, we present a low-power Leaky Integrate-and-Fire (LIF) neuron design fabricated in TSMC's 28 nm CMOS technology as proof of concept to build an energy-efficient mixed-signal Neuromorphic System-on-Chip (NeuroSoC). The fabricated neuron consumes 1.61 fJ/spike and occupies an active area of 34 $μm^{2}$, leading to a maximum spiking frequency of 300 kHz at 250 mV power supply. These performances are used in a software model to emulate the dynamics of a Spiking Neural Network (SNN). Employing supervised backpropagation and a surrogate gradient technique, the resulting accuracy on the MNIST dataset, using 4-bit post-training quantization stands at 82.5\%. The approach underscores the potential of such ASIC implementation of quantized SNNs to deliver high-performance, energy-efficient solutions to various embedded machine-learning applications.

cs.NE↗

arXiv2025

Scalable 49-Channel Neural Recorder with an Event-Driven Ramp ADC and PCA Compression in 28 nm CMOS

Neural interfaces advance neuroscience research and therapeutic innovations by accurately measuring neuronal activity. However, recording raw data from numerous neurons results in substantial amount of data and poses challenges for wireless transmission. While conventional neural recorders consume energy to digitize and process the full neural signal, only a fraction of this data carries essential spiking information. Leveraging on this signal sparsity, this paper introduces a neural recording integrated circuit in TSMC 28nm CMOS. It features an event-driven ramp analog-to-digital converter, and a spike compression module based on principal component analysis. The circuit consists of 49 channels, each occupying an on-chip area of 50 $\times$ 60 $μ$m$^2$. The circuit measures 1370 $\times$ 1370 $μ$m$^2$ and consumes 534 $μ$W. Compression testing on a synthetic dataset demonstrated an 8.8-fold reduction compared to raw spikes and a 328-fold reduction relative to the raw signal. This compression approach maintained a spike sorting accuracy of 74.9%, compared to the 79.5% accuracy obtained with the raw signal. The paper details the architecture and performance outcomes of the neural recording circuit and its compression module.

eess.SY↗

arXiv2024

Design and Implementation of a Low-Power Low-Noise Biopotential Amplifier in 28 nm CMOS Technology with a Compact Die-Area of 2500 $μ$m$^2$

This paper presents a compact low-power, low-noise bioamplifier for multi-channel electrode arrays, aimed at recording action potentials. The design we put forth attains a notable decrease in both size and power consumption. This is achieved by incorporating an active lowpass filter that doesn't rely on bulky DC-blocking capacitors, and by utilizing the TSMC 28 nm HPC CMOS technology. This paper presents extensive simulation results of noise and results from measured performance. With a mid-band gain of 58 dB, a -3 dB bandwidth of 7 kHz (from 150 Hz to 7.1 kHz), and an input-referred noise of 15.8 $μ$V$_{\rm rms}$ corresponding to a NEF of 12. The implemented design achieves a favourable trade-off between noise, area, and power consumption, surpassing previous findings in terms of size and power. The amplifier occupies the smallest area of 2500 $μ$m$^2$ and consumes only 3.4 $μ$W from a 1.2 V power supply corresponding to a power efficiency factor of 175 and an area efficiency factor of 0.43, respectively.

q-bio.NC↗

arXiv2023

Feasibility Assessment of an Optically Powered Digital Retinal Prosthesis Architecture for Retinal Ganglion Cell Stimulation

Clinical trials previously demonstrated the notable capacity to elicit visual percepts in blind patients affected with retinal diseases by electrically stimulating the remaining neurons on the retina. However, these implants restored very limited visual acuity and required transcutaneous cables traversing the eyeball, leading to reduced reliability and complex surgery with high postoperative infection risks. To overcome the limitations imposed by cables, a retinal implant architecture in which near-infrared illumination carries both power and data through the pupil to a digital stimulation controller is presented. A high efficiency multi-junction photovoltaic cell transduces the optical power to a CMOS stimulator capable of delivering flexible interleaved sequential stimulation through a diamond microelectrode array. To demonstrate the capacity to elicit a neural response with this approach while complying with the optical irradiance limit at the pupil, fluorescence imaging with a calcium indicator is used on a degenerate rat retina. The power delivered by the laser at the permissible irradiance of 4 mW/mm2 at 850 nm is shown to be sufficient to both power the stimulator ASIC and elicit a response in retinal ganglion cells (RGCs), with the ability to generate of up to 35 000 pulses per second at the average stimulation threshold. This confirms the feasibility of generating a response in RGCs with an infrared-powered digital architecture capable of delivering complex sequential stimulation patterns at high repetition rates, albeit with some limitations.

q-bio.NC↗