A Biomedical Imaging System Based on an Integrated Array of Magnetoresistive Sensors

Abstract

Magnetic tunnel junctions (MTJs) provide a valuable platform for biomedical imaging applications especially if the sensors are built on top of a CMOS integrated circuit. In this paper, we devise a platform for readout of MTJs that rely on analog circuitry with relaxed specifications, while part of the system complexity is leveraged to digital post-processing. The proposed system comprises a high-density array of 64 × 64 MTJs that are readout by 4 acquisition channels which comprehend mismatch correction, amplification, filtering and data conversion. In order to handle the expected large spreads that result from fabrication mismatches in high-density MTJ arrays, we introduce a pre-calibration procedure that enhances the signal processing dynamic range, enabling the usage of medium performance electronics while leading to high-quality images even in the presence of weak magnetic fields. Finally, we discuss how a proper sampling strategy may lead to improved imaging noise performance, especially in regards to low-frequency (flicker) noise with virtually no drawbacks. A mathematical model is used to estimate the noise performance, which is contrasted to extensive transient noise simulations in a 180 nm commercial CMOS process and enable the assessment of the proposed techniques. We show that the proposed system is able to acquire each pixel with an input-referred noise of only 55 µV, while the input signal range in our application is only 2.34 mV.