KOKofi Odame

Kofi Odame

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kofi Odame has a total of 20 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Inferring forced expiratory volume in 1 second (FEV1) from mobile ECG signals collected during quiet breathing

    by Maria T Nyamukuru, Alix Ashare, Kofi M Odame

    Objective. Forced expiratory volume in one second (FEV1) is an important metric for patients to track at home for their self-management of asthma and chronic obstructive pulmonary disease (COPD). Unfortunately, the state-of-the-art for measuring FEV1 at home either depends on the patient’s physical effort and motivation, or relies on bulky wearable devices that are impractical for long-term monitoring. This paper explores the feasibility of using a machine learning model to infer FEV1 from 270 seconds of a single-lead electrocardiogram (ECG) signal measured on the fingers with a mobile device. Methods. We evaluated the model’s inferred FEV1 values against the ground truth of hospital-grade spirometry tests, which were performed by twenty-five patients with obstructive respiratory disease. Results. The model-inferred FEV1 compared to the spirometry-measured FEV1 with a correlation coefficient of r = 0.73, a mean absolute percentage error of 23% and a bias of −0.08. Conclusions. These results suggest that the ECG signal contains useful information about FEV1, although a larger, richer dataset might be necessary to train a machine learning model that can extract this information with better accuracy. Significance. The benefit of a mobile ECG-based solution for measuring FEV1 is that it would require minimal effort, thus encouraging patient adherence and promoting successful self-management of asthma and COPD.

  • Analog Gated Recurrent Unit Neural Network for Detecting Chewing Events

    by Kofi Odame, Maria Nyamukuru, Mohsen Shahghasemi, Shengjie Bi, David Kotz

    We present a novel gated recurrent neural network to detect when a person is chewing on food. We implemented the neural network as a custom analog integrated circuit in a 0.18 $\mu$ m CMOS technology. The neural network was trained on 6.4 hours of data collected from a contact microphone that was mounted on volunteers' mastoid bones. When tested on 1.6 hours of previously-unseen data, the analog neural network identified chewing events at a 24-second time resolution. It achieved a recall of $91\%$ and an F1-score of $94\%$ while consuming $1.1~\mu$ W of power. A system for detecting whole eating episodes—like meals and snacks—that is based on the novel analog neural network consumes an estimated $18.8~\mu$ W of power.


  • Open Access

    Extracting Fractional Inspiratory Time from Electrocardiograms

    by Maria Nyamukuru, Kofi Odame

    Non-invasive at-home monitoring of lung and lung airways health enables the early detection and tracking of respiratory diseases like asthma and chronic obstructive pulmonary disease (COPD). Various proposed approaches estimate the respiratory rate and extract the respiratory waveform from an electrocardiogram (ECG) signal as a way to discreetly monitor lung health. Unfortunately, these approaches fail to accurately capture the respiratory cycle phase features, resulting in a non-specific, incomplete picture of lung health. This paper introduces an algorithm to extract more respiratory information from the ECG signal by framing the problem as a binary segmentation task. In addition to respiratory rate (RR), the algorithm derives the fractional inspiratory time (FIT), a direct measure of airway obstruction derived from respiratory phase information. The algorithm is based on a gated recurrent neural network that infers vital respiratory information from a single-lead ECG signal. We measure our algorithm's performance on 5 subjects from the MIMIC dataset and 5 subjects from the CEBS database. Our algorithm maintains exceptional performance in estimating the respiratory rate and outperforms current algorithms that extract the respiratory cycle phases and FIT/ inspiratory:expiratory ratio (IER). Our algorithm reports a root mean squared error (RMSE) of 0.06 in the computation of FIT (values range from 0.2-0.6) and a RMSE of 0.54 breaths per minute (bpm) for respiratory rate (values range from 8 - 28 breaths per minute (bpm)) on the MIMIC dataset, and an FIT RMSE of 0.11 and and RR RMSE of 0.66 bpm on the CEBS dataset.


  • Open Access

    Tiny Eats: Eating Detection on a Microcontroller

    by Maria T. Nyamukuru, Kofi M. Odame

    There is a growing interest in low power highly efficient wearable devices for automatic dietary monitoring (ADM) [1]. The success of deep neural networks in audio event classification problems makes them ideal for this task. Deep neural networks are, however, not only computationally intensive and energy inefficient but also require a large amount of memory. To address these challenges, we propose a shallow gated recurrent unit (GRU) architecture suitable for resource-constrained applications. This paper describes the implementation of the Tiny Eats GRU, a shallow GRU neural network, on a low power micro-controller, Arm Cortex M0+, to classify eating episodes. Tiny Eats GRU is a hybrid of the traditional GRU [2] and eGRU [3] to make it small and fast enough to fit on the Arm Cortex M0+ with comparable accuracy to the traditional GRU. The Tiny Eats GRU utilizes only 4% of the Arm Cortex M0+ memory and identifies eating or non-eating episodes with 6 ms latency and accuracy of 95.15%.

  • A Power Adaptive, 1.22-pW/Hz, 10-MHz Read-Out Front-End for Bio-Impedance Measurement

    by Mohammad Takhti, Kofi Odame

    This paper presents a read-out front-end ASIC for the measurement of tissue impedances. The 10 mm $^2$ 2-channel front-end ASIC is fabricated in a 0.18 μm CMOS technology. The measurement results show that the proposed ASIC operates over a frequency range of 100 Hz up to 10 MHz. The ASIC has 1.22 pW/Hz power performance, with an SNR over 72 dB for frequencies $\leq$ 1 MHz, and over 65 dB SNR for frequencies $\geq$ 1 MHz. The total measured power consumption of the read-out front-end is shown to range from 2.1 mW to 21.7 mW depending on the input frequency. To achieve a wide dynamic range, the IA has an adaptive gain control feature, and it employs programmable bias currents and reconfigurable structure to save power while processing low frequency signals. A dual-DAC hybrid SAR ADC is presented that reduces the power consumption of the ADC driver by 7 fold for frequencies between 4 kHz and 1 MHz.


  • Structured Design Methodology to Achieve a High SNR Electrical Impedance Tomography

    by Mohammad Takhti, Kofi Odame

    In this paper, we present a methodology for designing the main circuit building blocks of an electrical impedance tomography (EIT) system. In particular, we derive equations that map system-level EIT specifications to the performance requirements of each circuit block. We also review the circuit architectures that are best suited for meeting a given set of performance requirements. Our proposed design methodology is focused on maximizing the EIT system's signal-to-noise ratio (SNR) while minimizing total power consumption.


  • Open Access

    An Analog Front End ASIC for Cardiac Electrical Impedance Tomography

    by Arun Rao, Yueh-Ching Teng, Chris Schaef, Ethan K. Murphy, Saaid Arshad, Ryan J. Halter, Kofi Odame

    In this paper, an end-to-end CMOS application specific integrated circuit (ASIC) for readout channel in a cardiac electrical impedance tomography system is presented. The ASIC consists of an integrated current driver for current injection, an instrumentation amplifier, variable gain amplifier at the analog front end for voltage readout from electrodes, and an on-chip 10-bit successive approximation register analog to digital converter with serial peripheral interface. The ASIC is fabricated in the CMOS 0.18 $\mu$m process with a supply voltage of 3.3 V. Amplitude and phase extraction of the voltages is performed in the digital domain with a matched filter. A fully integrated solution for use in multiple electrode system is demonstrated. The readout chain in the ASIC achieves a minimum signal-to-noise ratio of 71 dB over the frequency range of 500 Hz–700 kHz, while maintaining an average accuracy of 99.7$\%$. Frame rates of 21 frames per second for a 32 electrode system is feasible, and the ASIC has an overall power consumption of 11.8 mW.


  • Open Access

    Signal-to-Noise Ratio Analysis of a Phase-Sensitive Voltmeter for Electrical Impedance Tomography

    by Ethan K. Murphy, Mohammad Takhti, Joseph Skinner, Ryan J. Halter, Kofi Odame

    In this paper, thorough analysis along with mathematical derivations of the matched filter for a voltmeter used in electrical impedance tomography systems are presented. The effect of the random noise in the system prior to the matched filter, generated by other components, are considered. Employing the presented equations allow system/circuit designers to find the maximum tolerable noise prior to the matched filter that leads to the target signal-to-noise ratio (SNR) of the voltmeter, without having to over-design internal components. A practical model was developed that should fall within 2 dB and 5 dB of the median SNR measurements of signal amplitude and phase, respectively. In order to validate our claims, simulation and experimental measurements have been performed with an analog-to-digital converter (ADC) followed by a digital matched filter, while the noise of the whole system was modeled as the input referred at the ADC input. The input signal was contaminated by a known value of additive white Gaussian noise (AWGN) noise, and the noise level was swept from 3% to 75% of the least significant bit (LSB) of the ADC. Differences between experimental and both simulated and analytical SNR values were less than 0.59 and 0.35 dB for RMS values ≥ 20% of an LSB and less than 1.45 and 2.58 dB for RMS values <; 20% of an LSB for the amplitude and phase, respectively. Overall, this study provides a practical model for circuit designers in EIT, and a more accurate error analysis that was previously missing in EIT literature.


  • A 2.5 pJ/b Binary Image Sensor as a Pathfinder for Quanta Image Sensors

    by Saleh Masoodian, Arun Rao, Jiaju Ma, Kofi Odame, Eric R. Fossum

    This paper presents a pathfinder binary image sensor for exploring low-power dissipation needed for future implementation of gigajot single-bit quanta image sensor (QIS) devices. Using a charge-transfer amplifier design in the readout signal chain and pseudostatic clock gating units for row and column addressing, the 1-Mpixel binary image sensor operating at 1000 frames/s dissipates only 20-mW total power consumption, including I/O pads. The gain and analog-to-digital converter stages together dissipate 2.5 pJ/b, successfully paving the way for future gigajot QIS sensor designs.