Study of Heart Disease and Analysis of PPD Algorithm

m5151117Yumiko Kimezawa

October 28, 2011 RPS 1

Outline

• Previous Work• Modification to Methods• Current PPD Algorithm• My Opinion• Future Work

October 28, 2011 RPS 2

Previous Work

• Studying how to actually capture an ECG signal from people

• An A/D converter included in our daughterboard is unfit for capturing the ECG signal

• An input amplifier will be required because the ECG signal is very weak

October 28, 2011 RPS 3

Modification to Methods

• Capturing an ECG signal from people actually

• We use sample data from database as input data

• Capturing an ECG signal from people actually

October 28, 2011 RPS 4

•Hard•Taking long time

Modification

•Studying only

Current PPD Algorithm

• This algorithm seems to process only ECG signals include typical peaks (P, Q, R, S, T and U)

• This algorithm can not process irregular heart rhythm such as ventricular fibrillation (Vf)

October 28, 2011 RPS 5

Ventricular Fibrillation (VF)

October 28, 2011 RPS 6

ventricle

•Extremely dangerous situation•Deadly arrhythmia•Irregular rhythm•P wave is not found•QRS waves are wide-ranging•Severe decrease in blood pressure and no feeling a pulse

Processing Results of VF

October 28, 2011 RPS 7

PPD algorithm must have

detected peaks incorrectly

On the ventricular fibrillation wave, P waves don’t appear and QRS waves are wide-ranging

My Opinion

• Someone’s heart beats hundreds of thousands of times per day (100,000 times/day)

• In addition to extracting ECG wave form, computers have to support analysis of the wave whether heart is normal or not

• Doctors only have to receive information of patients considered at risk for heart disease

October 28, 2011 RPS 8

It is difficult for doctors to diagnose so many patients

My Opinion

Future Work

• Analysis of PPD algorithm• Optimization of PPD algorithm

October 28, 2011 RPS 9

October 28, 2011 RPS 10

October 28, 2011 RPS 11

MIT:16272

RR Interval

PPD Algorithm MIT Database

0.578s (0.000s - 0.578s) 0.570s (0.000s – 0.570s)

0.984s (0.578s – 1.562s) 0.984s (0.570s – 1.555s)

0.953s (1.562s – 2.516s) 0.953s (1.555s – 2.508s)

0.961s (2.516s – 3.477s) 0.969s (2.508s – 3.477s)

0.953s (3.477s – 4.430s) 0.953s (3.477s – 4.430s)

0.969s (4.430s – 5.398s) 0.961s (4.430s – 5.391s)

0.969s (5.398s – 6.367s) 0.977s (5.391s – 6.367s)

0.992s (6.367s – 7.359s) 0.984s (6.367s – 7.352s)

0.961s (7.359s – 8.320s) 0.969s (7.352s – 8.320s)

0.984s (8.320s – 9.305s)

October 28, 2011 RPS 12

MIT:16273

RR Interval

PPD Algorithm MIT Database

0.086s (0.000s - 0.086s) 0.070s (0.000s – 0.070s)

0.609s (0.086s – 0.695s) 0.609s (0.070s – 0.680s)

0.602s (0.695s – 1.297s) 0.602s (0.680s – 1.281s)

0.602s (1.297s – 1.898s) 0.602s (1.281s – 1.883s)

0.609s (1.898s – 2.508s) 0.609s (1.883s – 2.492s)

0.609s (2.508s – 3.117s) 0.609s (2.492s – 3.102s)

0.617s (3.117s – 3.734s) 0.617s (3.102s – 3.719s)

0.633s (3.734s – 4.367s) 0.633s (3.719s – 4.352s)

0.656s (4.367s – 5.023s) 0.656s (4.352s – 5.008s)

0.641s (5.023s – 5.664s) 0.641s (5.008s – 5.648s)

0.641s (5.664s – 6.305s) 0.633s (5.648s – 6.281s)

0.617s (6.305s – 6.922s) 0.625s (6.281s – 6.906s)

0.633s (6.922s – 7.555s) 0.633s (6.906s – 7.539s)

0.617s (7.539s – 8.156s)

0.625s (8.156s – 8.781s)

0.617s (8.781s – 9.398s)

October 28, 2011 RPS 13

MIT:16273

RR Interval

PPD Algorithm MIT Database

0.086s (0.000s - 0.086s) 0.070s (0.000s – 0.070s)

0.609s (0.086s – 0.695s) 0.609s (0.070s – 0.680s)

0.602s (0.695s – 1.297s) 0.602s (0.680s – 1.281s)

0.602s (1.297s – 1.898s) 0.602s (1.281s – 1.883s)

0.609s (1.898s – 2.508s) 0.609s (1.883s – 2.492s)

0.609s (2.508s – 3.117s) 0.609s (2.492s – 3.102s)

0.617s (3.117s – 3.734s) 0.617s (3.102s – 3.719s)

0.633s (3.734s – 4.367s) 0.633s (3.719s – 4.352s)

0.656s (4.367s – 5.023s) 0.656s (4.352s – 5.008s)

0.641s (5.023s – 5.664s) 0.641s (5.008s – 5.648s)

0.641s (5.664s – 6.305s) 0.633s (5.648s – 6.281s)

0.617s (6.305s – 6.922s) 0.625s (6.281s – 6.906s)

0.633s (6.922s – 7.555s) 0.633s (6.906s – 7.539s)

0.617s (7.539s – 8.156s)

0.625s (8.156s – 8.781s)

0.617s (8.781s – 9.398s)

Our Architecture

October 28, 2011 RPS 14

A/Dconverter Filter RAM Processing Results

From patient

Sensor

Processing results are sent through the Ethernet

Important Things in Measurement of ECG Signal

• Clinically-important frequency component of ECG signal- 0 ~ 250Hz

October 28, 2011 RPS 15

• Change in voltage in chest wall- Less than 3mV (very weak)

• Measurement in environments with much noise

Change in voltage inside this range must be faithfully reproduced

Voltage gain of high signal-to-noise ratio (250 to 1000 times) must be gained• Signal-to-noise ratio: The ratio of signal power to the noise power

Book Knowledge

October 28, 2011 RPS 16

Impedance converter

Lead network

Head amplifier

Micro computerADCMain

amplifier

Isolation circuit

Output equipment

Lead Code

Power Circuit

•An example of block diagram illustrating cardiography equipment

Book Knowledge

October 28, 2011 RPS 17

Impedance converter

Lead network

Head amplifier

Micro computerADCMain

amplifier

Isolation circuit

Output equipment

Lead Code

Power Circuit

•An example of block diagram illustrating cardiography equipment

•This circuit prevents body’s impedance from being lower than input circuit’s impedance

Book Knowledge

October 28, 2011 RPS 18

Impedance converter

Lead network

Head amplifier

Micro computerADCMain

amplifier

Isolation circuit

Output equipment

Lead Code

Power Circuit

•An example of block diagram illustrating cardiography equipment

•Constitution of 12-lead from the combination of sensors•A~J : Sensors•Example

- Ⅰ: Output between A and E- aVR: Output between A and F

Book Knowledge

October 28, 2011 RPS 19

Impedance converter

Lead network

Head amplifier

Micro computerADCMain

amplifier

Isolation circuit

Output equipment

Lead Code

Power Circuit

•An example of block diagram illustrating cardiography equipment•ECG signal is very weak

•The signal must not be amplified suddenly

•Differential amplifier is used as head amplifier

An amplifier is needed

The signal is amplified in stages

• Noises are reduced• signal-to-noise ratio is

improved

Book Knowledge

October 28, 2011 RPS 20

Impedance converter

Lead network

Head amplifier

Micro computerADCMain

amplifier

Isolation circuit

Output equipment

Lead Code

Power Circuit

•An example of block diagram illustrating cardiography equipment

•For securing of security, impedance converter, lead network and head amplifier have to be isolated from main amplifier, power circuit and A/D converter

Question•A/D converter (AD9254) is included in HSMC- Pipelined switched capacitor ADC- Sampling frequency: 150MHz- Resolution: 14-bit- Power: 430mW

October 28, 2011 RPS 21

HSMC

Performance is overmuch•Clinically-important frequency component of ECG signal is from 0 to 250Hz

500Hz is enough for sampling frequency

•Using a lot of electricity•Conversion process is complex•Using A/D converter in HSMC is of questionable value

心電計の回路を表すブロック図の例

October 28, 2011 RPS 22

インピーダンス変換器

誘導ネットワーク

前置増幅器

マイクロコンピュータ

AD変換器主増幅器

アイソレーション回路

出力装置

誘導コー

ド電源回路

標準１２誘導を作るための誘導ネットワークの原理

October 28, 2011 RPS 23

A

B C D

E

F

G

HI

J

A~Jの各点の出力を組み合わせる。例えば、第Ⅰ誘導は AとE、 aVRは AとFを組み合わせたものとなる。