Cardiovascular diseases (CVDs) claimed 17.9 million lives annually (2019), representing 32% of total deaths, with 85% attributed to heart attacks and strokes worldwide. Atrial fibrillation (AF) is the most frequent type of cardiac arrhythmia that has a worldwide prevalence of 46 million individuals worldwide. The absence of P-waves in the electrocardiogram and irregular heartbeat characterise AF. Although the awareness and diagnostic methods of AF have improved over the past years, precision and timely diagnosis are required for effective prevention. Electrophysiological recording serves as the gold standard for diagnosing cardiac abnormalities. There currently exist two major fundamental approaches for monitoring the heart from an electrophysiological point of view. The most common approach is the measurement of the ECG (electrocardiogram), which is recorded on the surface of the human body at specific points. A more recent approach (but less commonly used) is the measurement of the AEG (atrial electrogram), which is recorded on the surface of the human heart with higher spatial resolution than ECG. ECG is a technique used to monitor the heart’s electrical activity non-invasively, whereas AEG is used to obtain more profound and detailed insights into the electrical conduction pathways of the heart invasively. AEGs are used to localize the origin of AF for appropriate treatment. Chapter 2 elaborates on the application, the associated technological challenges, and the existing solutions in the literature concerning cardiac signal acquisition. The first approach (most commonly used) for diagnosing cardiac abnormalities is recording the ECG, with a high signal-to-noise-and-interference ratio and sufficient bandwidth using a single-channel analog front-end. One of the main challenges with acquiring ECG is baseline wandering (BW), which refers to low-frequency variations affecting the baseline of the recorded waveform. BW occurs due to body movement or poor contact between the body surface and electrodes. The impact of BW on the ECG signal quality can be twofold: (a) Saturation of the CMOS analog front-end leading to loss of information, and (b) Distortion of the ECG waveform leading to misdiagnosis. In order to prevent or suppress the impact of BW on the ECG waveform, the analog front-end requires a high-pass filter with high linearity and accuracy to minimize distortion. The information in an ECG signal lies between 0.5-200 Hz. Implementing a filter cutoff in the sub-Hz region while maintaining high linearity and accuracy is challenging. Sub-Hz filter implementation translates to large area occupation on silicon and is thus expensive. Although there are several techniques to realize large time constants on-chip, there does not exist a highly linear high-pass transfer characteristic for Sigma-Delta (ΣΔ) analog-to-digital converters (ADCs). The first part of the thesis addresses the implementation of large time constants with high linearity and accuracy. Chapter 3 of the thesis focuses on the design and analysis of a high-pass ΣΔ (HPΣΔ) analog-to-digital converter (ADC) that aims to achieve high linearity for the high-pass (HP) filter cut-off. State-space topologies, commonly used for optimizing filters, are proposed for developing the HPΣΔ modulator topologies. A ΣΔ modulator can be represented as a linear model that consists of integrators and a quantizer, with the quantization noise modeled as additive white noise. By employing the state-space synthesis method, it becomes possible to develop arbitrary transfer functions for the signal and quantization noise of the state-space ΣΔ topologies. The dynamic range of the ΣΔ topology is optimized by signal and noise scaling. Through sensitivity analysis, the impact of coefficient variations in the different integrators on the overall performance and stability of the modulator is determined. Comparative analysis reveals that the orthonormal ΣΔ modulator outperforms the observable canonical statespace- based topology. The experimental results demonstrate that the orthonormal HPΣΔ ADC achieves a figure of merit (FoM) of 5.35 pJ/conv while occupying an area of 0.126 mm2. The orthonormal HPΣΔ ADC also achieves a peak SNDR of 69.8 dB, corresponding to 11.3 bits of ENOB for a signal bandwidth of 3 kHz. The second approach (and less commonly used) for diagnosing cardiac abnormalities is recording the AEG, targeting high spatial resolution using a multi-channel analog frontend. The primary difficulties encountered when acquiring multiple inputs are associated with a proportional increase in area, power consumption, and the number of outgoing wires. In order to mitigate the resource requirements and develop a compact solution for the multi-input system, channel-sharing techniques such as time-division multiplexing, frequency-division multiplexing, and code-division multiplexing (CDM) can be effectively incorporated. These techniques enable the efficient utilization of shared resources, optimizing the overall performance and reducing the system’s complexity. CDM offers enhanced capacity by enabling multiple users to effectively share the same frequency band. It improves signal quality by effectively suppressing interference and band-limited noise, resulting in higher fidelity and more reliable transmission. The second part of this thesis (Chapter 4) focuses on acquiring multi-input AEG signals using CDM. It presents a systematic classification of modulation strategies based on their degrees of freedom to identify suitable techniques for analog signal acquisition. This work also introduces a design method for creating efficient spread-spectrum analog front-ends. Based on the proposed design strategy, spread-spectrum codes can be selected for a given application requirement (high or low-resolution acquisition). The modulation frequency and code length can be determined for optimal performance for the total number of inputs. A 4- input spread-spectrum recording system fabricated in a 0.18 𝜇m CMOS process validates the proposed design method. With a 7-bit Gold code generator (𝐿 = 127), the maximum achievable crosstalk performance is -40 dB, and the thermal noise density of the system is 224 nV/ √(Hz). The system includes shared components such as an amplifier, an analogto- digital converter (ADC), and an on-chip Gold code generator, with a compact footprint and low power consumption per channel input of 0.067mm2 and 23𝜇A, respectively. Recording AEGs using a high-density array of flexible electrodes leads to generating large amounts of data. A significant amount of area (for storage) and power (for data transmission) are required to handle the data. A custom-fabricated flexible multi-electrode array that contains 192 electrode sites is used to acquire the AEGs. Nine such sections are required to cover the entire area of an average adult heart. Chapter 5 focuses on the compressibility of AEGs. Standard compressed sensing is typically used to reduce the data using sub-Nyquist domain sampling on-chip and reconstruction using optimal algorithms off-chip. Signal statistics are not taken into account in standard CS. Both standard and rakeness-based compressed-sensing are applied on atrial electrograms for data compression in this thesis. Rakeness-based compressed-sensing (CS), which uses the signal statistics of the known input signal, is proposed for compressing AEGs. Incorporating signal statistics in the random matrix in rakeness CS leads to better reconstruction performance at higher compression ratios. A similarity analysis is also conducted to quantitatively assess the quality of the reconstructed AEGs during sinus rhythm (SR) and atrial fibrillation. Also, a team of clinicians has visually inspected the reconstructed waveforms to check their suitability for generating an activation map (to map the wavefront of the signal that propagates through the heart). From the survey, it is found that for SR AEG signals up to a compression ratio (CR) of 4.26, the signal is considered clean. However, at CR = 5.1, it becomes noisier but remains suitable for specific applications. In contrast, AF AEG signals exhibit a faster decline in signal quality at higher CRs, attributed to increased activity and distinctive features of AF signals. Even at CR = 5.1, the signal can be used to estimate local activation times but lacks detail for comprehensive feature computation. In summary, this thesis addresses numerous pivotal challenges, namely enhancing the linearity and accuracy of the HP filter cut-off for single-channel cardiac signal acquisition and maximizing resource efficiency for multi-channel cardiac acquisition. We have taken steps towards designing compact application-specific integrated circuits (ASICs) for recording and diagnosing cardiac abnormalities, such as AF. Understanding the underlying biological mechanisms through recording AEGs is crucial for early detection and monitoring of such conditions, which can significantly affect the course of disease progression. @en

Achieving an accurate sub-Hz high-pass (HP) cutoff frequency and simultaneously a high accuracy of the transfer function is a challenge in the implementation of analog-to-digital converters for biomedical ExG signals. A structured electronic design approach based on state-space forms is proposed to develop HPΣΔ modulators targeting high accuracy of the HP cutoff frequency and good linearity. Intermediate transfer functions are mathematically evaluated to compare the proposed HPΣΔ topologies with respect to dynamic range. Finally, to illustrate the design method, an orthonormal HPΣΔ modulator is designed to be implemented in 0.18 μm technology which achieves a linearity of 12-bits. @en

This paper presents a novel multichannel level-crossing (MLC) ADC architecture aimed at recording atrial electrograms from multiple channels. The proposed architecture combines synchronous sampling with level-crossing (LC) quantisation to achieve activity dependent operation while recording from multiple channels simultaneously. In the proposed architecture the number of comparisons performed by the quantiser to reach a decision is dependent on the activity of the input signal and is 2-3.3 times lower than that in a conventional SAR ADC. The architecture uses one comparator and one reference level instead of two comparators and two reference levels as in conventional LC ADCs. The proposed architecture is modeled in VerilogA and is designed to be implemented in a standard 0.18 um CMOS process. The MLC ADC converts signals from 4 channels simultaneously and achieves an SFDR of 53.33 dB and an SNDR of 48.96 dB while consuming 9.32 μW of power from a 1.8 V power supply.@en

Low-voltage and low-power front-end design is required for the safe and long-term monitoring of cardiac signals. To address the low-voltage challenge, this paper presents a subthreshold source-coupled logic (STSCL) based time-domain comparator designed in 180 nm CMOS process technology. At a low supply voltage of 0.8 V, the STSCL time-domain comparator consumes 2.3 μW at 1 MHz. Using 4 stages, the input referred noise and the offset of the comparator are 32 μVrms and 1.8 mV, respectively.@en

Atrial electrograms (AEGs) acquired with a high spatio-temporal resolution are a promising approach for early detection of atrial fibrillation. Due to the high data rate, transmission of AEG signals is expensive in terms of power consumption and resources, making its adoption a challenge for low-power wireless devices. In this paper, we investigate the feasibility of using compressed sensing (CS) for the acquisition of AEGs while reducing redundant data without losing information. We apply two CS approaches, standard CS and rakeness-based CS (rak-CS) on a data set, composed of real medical recordings. We find that the AEGs are compressible in time, and, more interestingly, in the spatial domain. The performance of rak-CS is better than standard CS, especially at higher compression ratios (CR), both during sinus rhythm (SR) and atrial fibrillation (AF). The difference in the achieved average reconstruction signal-to-noise (ARSNR) in rak-CS and standard CS, for CR = 4.26, in the time domain is 7.7 dB and 2.6 dB for AF and SR, respectively. Multi-channel data is modeled as a multiple-measurement-vector problem and the mixed norm is used to exploit the group structure of the signals in the spatial domain to obtain improved reconstruction performance over $l_{1}$ norm minimization. Using the mixed-norm recovery approach, for CR = 4.26, the difference in achieved ARSNR performance between rak-CS and standard CS is 5 dB and 2 dB for AF and SR, respectively.@en

Atrial electrogram (AEG) acquired with a high spatio-temporal resolution is a promising approach for early detection of atrial fibrillation. Due to the high data rate, transmission of AEG signals requires considerable energy, making its adoption a challenge for low-power wireless devices. In this paper, we investigate the feasibility of using compressed sensing (CS) for the acquisition of AEGs while reducing redundant data without losing information. We apply two CS approaches, standard CS and rakeness-based CS (rak-CS) on real medical recordings. We find that the AEGs are compressible in time, and, more interestingly, in the spatial domain. The performance of rak-CS is better than standard CS, especially at higher compression ratios (CR), both during sinus rhythm (SR) and atrial fibrillation (AF). More specifically, the difference in the achieved average reconstruction signal-to-noise (ARSNR) in rak-CS and standard CS, for CR = 4.26, in the time domain is 7.7 dB and 2.6 dB for AF and SR, respectively. Multi-channel data is modeled as a multiple-measurement-vector problem and a suitable mixed norm is used to exploit the group structure of the signals in the spatial domain to obtain improved reconstruction performance over l1 norm minimization. Using the mixed-norm recovery approach, for CR = 4.26, the difference in achieved ARSNR performance between rak-CS and standard CS is 5 dB and 2 dB for AF and SR, respectively.@en