Low-power biomedical processors with embedded machine-learning accelerators for analytically-intractable physiological signals

Low-power sensing technologies have emerged for acquiring physiologically-indicative patient signals. However, to achieve high clinical value, it is critical to analyze the signals to extract specific medical information. Given the complexities of the underlying processes, high-order signal models are required for accurate signal analysis. Machine-learning offers distinct advantages, but the computations are not well supported by traditional DSP platforms; high-order models lead to energy and memory intensive computations. This thesis investigates these challenges from the levels of kernel functions, microprocessor architectures, and algorithms. To enable low-energy computations, a reformulation of a polynomial supportvector machine (SVM) kernel function is proposed that can substantially reduce the real-time computations involved. Using ECG-based arrhythmia-detection and EEGbased seizure-detection applications with clinical patient data, it is shown that the polynomial models yield performance accuracy comparable to the most powerful available transformation (i.e., the radial-basis function), and yet the proposed formulation reduces energy by over 2500× and 9.3 198× (depending on the patient), respectively. Next, an accelerator-based biomedical processor is proposed. It employs a lowpower SVM accelerator realizing various kernel functions and reformulations, spanning design points within an accuracy-versus-energy and -memory trade-off space. An active-learning accelerator enables patient-specific model customization while minimizing the modeling effort from human experts. The prototype is implemented in 130nm CMOS. Medical applications for EEG-based seizure detection and ECG-based cardiac-arrhythmia detection are demonstrated using clinical data which reduce energy by 62.4× (273μJ) and 144.7× (124μJ), respectively, compared to a CPU-alone implementation. A patient-adaptive cardiac-arrhythmia detector is also demonstrated which reduces the training data required by a factor of 20×.

[1]  Naveen Verma,et al.  A Micro-Power EEG Acquisition SoC With Integrated Feature Extraction Processor for a Chronic Seizure Detection System , 2010, IEEE Journal of Solid-State Circuits.

[2]  Naveen Verma,et al.  Data-Driven Approaches for Computation in Intelligent Biomedical Devices: A Case Study of EEG Monitoring for Chronic Seizure Detection , 2011 .

[3]  François Brémond,et al.  Gesture recognition by learning local motion signatures , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[4]  Robert E. Schapire,et al.  A Brief Introduction to Boosting , 1999, IJCAI.

[5]  Shimeng Yu,et al.  Metal–Oxide RRAM , 2012, Proceedings of the IEEE.

[6]  Refet Firat Yazicioglu,et al.  A 60 $\mu$W 60 nV/$\surd$Hz Readout Front-End for Portable Biopotential Acquisition Systems , 2007, IEEE Journal of Solid-State Circuits.

[7]  John E. Wennberg,et al.  Tracking the Care of Patients with Severe Chronic Illness: The Dartmouth Atlas of Health Care 2008 , 2008 .

[8]  Naveen Verma,et al.  A low-power microprocessor for data-driven analysis of analytically-intractable physiological signals in advanced medical sensors , 2013, 2013 Symposium on VLSI Circuits.

[9]  Philip de Chazal,et al.  Automatic classification of heartbeats using ECG morphology and heartbeat interval features , 2004, IEEE Transactions on Biomedical Engineering.

[10]  Mario Konijnenburg,et al.  A voltage-scalable biomedical signal processor running ECG using 13pJ/cycle at 1MHz and 0.4V , 2011, 2011 IEEE International Solid-State Circuits Conference.

[11]  Klaus Fabian Coco,et al.  Automatic sleep staging using a single-channel EEG modeling by Kalman Filter and HMM , 2011, ISSNIP Biosignals and Biorobotics Conference 2011.

[12]  L. Rabiner,et al.  An introduction to hidden Markov models , 1986, IEEE ASSP Magazine.

[13]  David Tak-Wai Hau,et al.  Learning Qualitative Models from Physiological Signals , 1994 .

[14]  Meir Kalech,et al.  Machine-learning-based circuit synthesis , 2012, 2012 IEEE 27th Convention of Electrical and Electronics Engineers in Israel.

[15]  Dana Angluin,et al.  Queries and concept learning , 1988, Machine Learning.

[16]  T.H. Lee,et al.  Cartesian feedback for RF power amplifier linearization , 2004, Proceedings of the 2004 American Control Conference.

[17]  Jeffrey M. Hausdorff,et al.  Physionet: Components of a New Research Resource for Complex Physiologic Signals". Circu-lation Vol , 2000 .

[18]  Sun-Yuan Kung,et al.  Low-energy Formulations of Support Vector Machine Kernel Functions for Biomedical Sensor Applications , 2012, Journal of Signal Processing Systems.

[19]  Quinn Jacobson,et al.  ERSA: error resilient system architecture for probabilistic applications , 2010, DATE 2010.

[20]  Richard B Reilly,et al.  Electrograms (ECG, EEG, EMG, EOG). , 2010, Technology and health care : official journal of the European Society for Engineering and Medicine.

[21]  R. Ward,et al.  EMG and EOG artifacts in brain computer interface systems: A survey , 2007, Clinical Neurophysiology.

[22]  Gregory Molnar,et al.  Creating support circuits for the nervous system: Considerations for “brain-machine” interfacing , 2009, 2009 Symposium on VLSI Circuits.

[23]  L. Hartley,et al.  Post Myocardial Infarction , 1983 .

[24]  Sun-Yuan Kung,et al.  Feature Selection for Genomic Signal Processing: Unsupervised, Supervised, and Self-Supervised Scenarios , 2010, J. Signal Process. Syst..

[25]  Eilon Vaadia,et al.  Kernel-ARMA for Hand Tracking and Brain-Machine interfacing During 3D Motor Control , 2008, NIPS.

[26]  William M. Pottenger,et al.  Hardware-based support vector machine classification in logarithmic number systems , 2005, 2005 IEEE International Symposium on Circuits and Systems.

[27]  B. Murmann,et al.  A 12 b 75 MS/s pipelined ADC using open-loop residue amplification , 2003, 2003 IEEE International Solid-State Circuits Conference, 2003. Digest of Technical Papers. ISSCC..

[28]  Jerald Yoo,et al.  A 1.83µJ/classification nonlinear support-vector-machine-based patient-specific seizure classification SoC , 2013, 2013 IEEE International Solid-State Circuits Conference Digest of Technical Papers.

[29]  Sun-Yuan Kung,et al.  Improving kernel-energy trade-offs for machine learning in implantable and wearable biomedical applications , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[30]  Pedro P. Irazoqui,et al.  Ultra Low-Power Algorithm Design for Implantable Devices: Application to Epilepsy Prostheses , 2011 .

[31]  Naveen Verma,et al.  An ultra low power ADC for wireless micro-sensor applications , 2005 .

[32]  Won-Seong Lee.  Future memory technologies , 2008, 2008 9th International Conference on Solid-State and Integrated-Circuit Technology.

[33]  P.R. Kinget.  Device mismatch and tradeoffs in the design of analog circuits , 2005, IEEE Journal of Solid-State Circuits.

[34]  S. Walther.  A unified algorithm for elementary functions , 1899 .

[35]  Brian Litt,et al.  Flexible, Foldable, Actively Multiplexed, High-Density Electrode Array for Mapping Brain Activity in vivo , 2011, Nature Neuroscience.

[36]  Paul A. Viola,et al.  Rapid object detection using a boosted cascade of simple features , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.

[37]  Anantha Chandrakasan,et al.  An 8-channel scalable EEG acquisition SoC with fully integrated patient-specific seizure classification and recording processor , 2012, 2012 IEEE International Solid-State Circuits Conference.

[38]  Naresh Shanbhag.  What is stochastic computation? , 2010, SIGD.

[39]  Hui Cao,et al.  Approximate RBF Kernel SVM and Its Applications in Pedestrian Classification , 2008 .

[40]  Anantha Chandrakasan,et al.  A Low-Voltage 1 Mb FRAM in 0.13 $\mu$m CMOS Featuring Time-to-Digital Sensing for Expanded Operating Margin , 2012, IEEE Journal of Solid-State Circuits.

[41]  Mark Horowitz,et al.  Scaling, Power and the Future of CMOS , 2007, 20th International Conference on VLSI Design held jointly with 6th International Conference on Embedded Systems (VLSID'07).

[42]  Naveen Verma,et al.  A 1.2–0.55V general-purpose biomedical processor with configurable machine-learning accelerators for high-order, patient-adaptive monitoring , 2012, 2012 Proceedings of the ESSCIRC (ESSCIRC).

[43]  Elif Derya Übeyli.  ECG beats classification using multiclass support vector machines with error correcting output codes , 2007, Digit. Signal Process..

[44]  Naveen Verma,et al.  A data-driven modeling approach to stochastic computation for low-energy biomedical devices , 2011, 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[45]  E.L. Glassman,et al.  Reducing the Number of Channels for an Ambulatory Patient-Specific EEG-based Epileptic Seizure Detector by Applying Recursive Feature Elimination , 2006, 2006 International Conference of the IEEE Engineering in Medicine and Biology Society.

[46]  P. Tresco,et al.  Response of brain tissue to chronically implanted neural electrodes , 2005, Journal of Neuroscience Methods.

[47]  Chandan Chakraborty,et al.  Automated Screening of Arrhythmia Using Wavelet Based Machine Learning Techniques , 2012, Journal of Medical Systems.

[48]  John Guttag,et al.  Reducing energy consumption of multi-channel mobile medical monitoring algorithms , 2008, HealthNet '08.

[49]  J. Kwong,et al.  An Energy-Efficient Biomedical Signal Processing Platform , 2010, IEEE Journal of Solid-State Circuits.

[50]  A.P. Chandrakasan,et al.  Nanometer MOSFET Variation in Minimum Energy Subthreshold Circuits , 2008, IEEE Transactions on Electron Devices.

[51]  Zeeshan Syed,et al.  Scalable Personalization of Long-Term Physiological Monitoring: Active Learning Methodologies for Epileptic Seizure Onset Detection , 2012, AISTATS.

[52]  Boris Murmann,et al.  A/D converter trends: Power dissipation, scaling and digitally assisted architectures , 2008, 2008 IEEE Custom Integrated Circuits Conference.

[53]  Valeria Bertacco,et al.  Machine learning-based anomaly detection for post-silicon bug diagnosis , 2013, 2013 Design, Automation & Test in Europe Conference & Exhibition (DATE).

[54]  Naveen Verma,et al.  Scalable customization of atrial fibrillation detection in cardiac monitoring devices: Increasing detection accuracy through personalized monitoring in large patient populations , 2011, 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[55]  S. Tehrani.  Status and Outlook of MRAM Memory Technology (Invited) , 2006, 2006 International Electron Devices Meeting.

[56]  Joël Hartmann,et al.  Towards a New Nanoelectronic Cosmology , 2007, 2007 IEEE International Solid-State Circuits Conference. Digest of Technical Papers.

[57]  Trevor Mudge,et al.  Razor: a low-power pipeline based on circuit-level timing speculation , 2003, Proceedings. 36th Annual IEEE/ACM International Symposium on Microarchitecture, 2003. MICRO-36..

[58]  Kush Gulati,et al.  A low-power reconfigurable analog-to-digital converter , 2001, 2001 IEEE International Solid-State Circuits Conference. Digest of Technical Papers. ISSCC (Cat. No.01CH37177).

[59]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[60]  Ali H. Shoeb,et al.  Application of Machine Learning To Epileptic Seizure Detection , 2010, ICML.

[61]  David A. Johns,et al.  Analog Integrated Circuit Design , 1996 .

[62]  Deog-Kyoon Jeong,et al.  A low-noise differential front-end and its controller for capacitive touch screen panels , 2012, 2012 Proceedings of the ESSCIRC (ESSCIRC).

[63]  John A. Rogers,et al.  Inorganic semiconductor nanomaterials for flexible and stretchable bio-integrated electronics , 2012 .

[64]  F. L. D. Silva,et al.  Dynamics of non-convulsive epileptic phenomena modeled by a bistable neuronal network , 2004, Neuroscience.

[65]  Rahul Sarpeshkar,et al.  An ultra-low-power programmable analog bionic ear processor , 2005, IEEE Transactions on Biomedical Engineering.

[66]  David Blaauw,et al.  A 0.27V 30MHz 17.7nJ/transform 1024-pt complex FFT core with super-pipelining , 2011, 2011 IEEE International Solid-State Circuits Conference.

[67]  Gert Cauwenberghs,et al.  Kerneltron: Support Vector 'Machine' in Silicon , 2002, SVM.

[68]  Naveen Verma,et al.  A Low-Power Processor With Configurable Embedded Machine-Learning Accelerators for High-Order and Adaptive Analysis of Medical-Sensor Signals , 2013, IEEE Journal of Solid-State Circuits.

[69]  Naveen Verma.  Ultra-low-power SRAM design in high variability advanced CMOS , 2009 .

[70]  Eugene Shih,et al.  Reducing the computational demands of medical monitoring classifiers by examining less data , 2010 .

[71]  Kaushik Roy,et al.  Scalable effort hardware design: Exploiting algorithmic resilience for energy efficiency , 2010, Design Automation Conference.

[72]  Naveen Verma,et al.  A low-energy computation platform for data-driven biomedical monitoring algorithms , 2011, 2011 48th ACM/EDAC/IEEE Design Automation Conference (DAC).

[73]  Eric B. Baum,et al.  Neural net algorithms that learn in polynomial time from examples and queries , 1991, IEEE Trans. Neural Networks.

[74]  James L. Flanagan,et al.  Adaptive quantization in differential PCM coding of speech , 1973 .

[75]  Klaus Brinker,et al.  Incorporating Diversity in Active Learning with Support Vector Machines , 2003, ICML.

[76]  J. Volkmann,et al.  Introduction to the programming of deep brain stimulators , 2002, Movement disorders : official journal of the Movement Disorder Society.

[77]  Bernadette Dorizzi,et al.  ECG signal analysis through hidden Markov models , 2006, IEEE Transactions on Biomedical Engineering.

[78]  A.-T. Avestruz,et al.  A 2 $\mu\hbox{W}$ 100 nV/rtHz Chopper-Stabilized Instrumentation Amplifier for Chronic Measurement of Neural Field Potentials , 2007, IEEE Journal of Solid-State Circuits.

[79]  Mircea R. Stan,et al.  The Promise of Nanomagnetics and Spintronics for Future Logic and Universal Memory , 2010, Proceedings of the IEEE.

[80]  Ali H. Shoeb,et al.  Application of machine learning to epileptic seizure onset detection and treatment , 2009 .

[81]  Eric Panken,et al.  A micropower support vector machine based seizure detection architecture for embedded medical devices , 2009, 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[82]  Russell S. Kirby,et al.  The Dartmouth Atlas of Health Care , 1998 .

[83]  Andrew B. Schwartz,et al.  Brain-Controlled Interfaces: Movement Restoration with Neural Prosthetics , 2006, Neuron.

[84]  Peter J.F. Lucas.  Bayesian analysis, pattern analysis, and data mining in health care , 2004, Current opinion in critical care.

[85]  P. K. Dubey,et al.  Recognition, Mining and Synthesis Moves Comp uters to the Era of Tera , 2005 .

[86]  Seok-Jun Lee,et al.  Microwatt Embedded Processor Platform for Medical System-on-Chip Applications , 2011, IEEE Journal of Solid-State Circuits.

[87]  P. Newacheck,et al.  Childhood chronic illness: prevalence, severity, and impact. , 1992, American journal of public health.

[88]  Naveen Verma,et al.  Enabling system-level platform resilience through embedded data-driven inference capabilities in electronic devices , 2012, 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).