Edward Jezisek Brandon Autrey Edward Nowlin Renato Ortega Group 2

1. Edward Jezisek Brandon Autrey Edward Nowlin Renato Ortega Group 2

2. Motivation • To improve the quality of ECG signals acquired from patients in an MRI environment • ECG quality in MRI machines is lacking • Current filter selection is confusing • Current filters are imperfect • New MRI scans create problems

3. Objectives • To further the use of adaptive algorithms in medical practice • Make patient monitors easier to use • Improve the quality of ECG Signals

4. Specifications and Requirements

5. Noise Contributions • Electrical interference from power lines add 50 or 60 Hz power-line frequency. • Muscle contraction and muscle activity can generate high frequency electromyography (EMG) noise. • Motion artifacts such as movement of the electrode over the skin surface.

6. Overall System Diagram

7. Hardware

8. PCB Layout • 417 components • 6 layers • Digital and Analog ground plane • ESD/Defibrillation protection • Front end amplification and filtering • 4 ECG inputs and outputs • 3 gradient antenna inputs • ARM Cortex M4 Processor

9. Digital to Analog Converter (AD5640) • 14 Bit conversion resolution • Single channel • 3 wire serial interface • Clock rate up to 30 MHz

10. ARM Cortex M4F • RISC based microprocessor • 80 MHz clock speed • Two 12-bit ADC modules with maximum sample rate of 1 MSPS • 256 KB Flash memory • Four SSI Modules • 105 GPIOs including 24 shared analog input channels • Single-precision Floating Point Unit (FPU)

11. PCB Schematic

12. Input from each ECG lead • Removed because of NDA

13. Input from each gradient antenna • Removed because of NDA

14. ESD/Defibrillation Protection • In a situation when a patient goes into heart failure, a defibrillator may need to be utilized • A SG75 gas discharge tube is used to short the current going into the filter when voltage exceeds 75V • Clamping diodes are also used after the discharge tube to short the circuit then the voltage exceeds +/-3.3V

15. ESD/Defibrillation Protection • Removed because of NDA

16. Sallen Key Filter • Front end filtering is used to eliminate as much interference as possible before the analog signal is converted into a digital signal • The Sallen key filter has a cut off of about 5500 rad/s or 9kHz

17. 3rd order Sallen key filter • Removed because of NDA

18. Op amp biasing and gain • Op amp is biased to 1.2V to allow for proper operation of the analog to digital converter. • Biasing is achieved through the use of applying a voltage to the circuit and using a voltage divider • The Op amp circuit is configured to have a gain of 1.5

19. Op amp biasing and gain Removed because of NDA Op amp biasing Gain

20. PCB Power • Analog and Digital ground planes are separated by a CLC filter • Input voltage is 5V • Voltage reference is used to step voltage down to 3V for applicable components • Linear regulators are used to maintain a steady voltage • Digital and Analog grounds are separated by a CLC filter

21. PCB Power • Removed because of NDA

22. Printed Circuit Board

23. Software

24. Linear Filters • Pros: • Easier to make, and is already implemented in many machines • Is possible to implement with an analog system(few or no extra parts) • Quick and usable for diagnostic quality devices • Cheaper • Cons: • Requires a professional to select the correct linear filter • Easy to make mistakes; if a mistake is made that can be costly in terms of receiving useless data. And in the medical profession; time is of the essence. • Training is confusing • Quality is lacking

25. Adaptive Filters • Pros: • Much better quality • Less input data from user • Cons: • Difficulties filtering the incoming amount of data • More research is required for a full implementation of it • Costs more (R&D)

26. Least Mean Square Filter Example

27. Real-Time Matlab Plot • Video removed due to file size limitations

28. Important Data Points

29. Analog to Digital Converter • Two ADC modules that use Successive Approximation Register (SAR) architecture to deliver low-power 12-bit conversion resolution • Supports 24 input channels • Maximum sampling speed of 1 million samples per second • Four programmable sample conversion sequencers • ADC logic runs at 16 MHz • 3.3 V reference voltage

30. Programmable Sequencers • We used sequencer SS0 is used to sample 6 signals at once • 3 ECG signals • 3 Gradient signals • Sample sequencer is triggered by processor interrupt

31. Synchronous Serial Interface (SSI) • Programmable data frame size from 4 and 16 bits • Transmit FIFO 16 bits wide and 8 locations deep • Master or slave operation • Three types of frame formats • MICROWIRE and Freescale SPI • FSS pin is active low for duration of frame transfer • Texas Instruments Synchronous Serial Interface • FSS pin is pulsed for one serial clock period prior to frame transmission • Three SSI modules used for the three DACs.

32. Continuous TI Synchronous Serial Frame Format

33. SPI Digital Data

34. Digital to Analog Converter • Single channel 3-wire serial interface compatible with SPI • SYNC, SCLK, and DIN • Clock rate up to 30 MHz • Reference voltage of 2.7 V

35. Write Sequence for DAC

36. DAC output of 1 KHz sine wave

37. Class Diagram

38. Testing Procedures

39. Functional Testing • MATLAB • Used to confirm the device sends the correct signal • Oscilloscope • Used to read the data from the PCB, ensures correct signals sent and received • Function Generator • Supplies various signals that were filtered with the LMS

40. ECG Generator

41. Connections

42. ECG Signal with No Filtering

43. ECG Signal with Adaptive Filtering • Video removed due to file size limitations

44. Issues • Noise of about +-30 mV on the board • UART is not working on PCB • Difficulty selecting step size and filter order

45. Budget

46. Progress

47. Work Distribution

48. Questions?