Diagnostic Stethoscope Project Synopsis

6
Smart Stethoscope By: Shabana Banu, Chetan D, Bharath Kumar V, Bhanu Pratap Reddy Project Group USN: PES13FYP15 Guide: Kiran Kumar 3 February 2014 1 PES INSTITUTE OF TECHNOLOGY - BSC

Transcript of Diagnostic Stethoscope Project Synopsis

Page 1: Diagnostic Stethoscope Project Synopsis

!Smart Stethoscope By: Shabana Banu, Chetan D, Bharath Kumar V, Bhanu Pratap Reddy Project Group USN: PES13FYP15 Guide: Kiran Kumar 3 February 2014 !

!1

PES INSTITUTE OF TECHNOLOGY - BSC

Page 2: Diagnostic Stethoscope Project Synopsis

EXECUTIVE SUMMARY !Cardiac murmurs are pathologic sounds that are produced by turbulent blood flow in the heart. Detailed diagnoses of pathologic murmurs often require echocardiogram procedures. Although the procedure is effective, it requires special equipment and trained technicians to capture the necessary images and measurements. On the other hand, heart murmurs can sometimes be detected by a physician using a standard stethoscope during auscultation. This procedure is commonly performed during routine check-ups. However, depending on the grade or severity of the murmur, the quality of the stethoscope, and the training and skill of the physician, it can be difficult for a physician to distinguish a murmur from a normal heartbeat. This design project aims to assist physicians in detecting heart murmurs by analysing cardiac signals in real time during auscultation and reporting any detected abnormalities.

The task of designing a cardiac murmur detection algorithm has been previously explored by several researchers in various academic groups. In general, the task can be described as a pattern recognition problem using 1-dimensional medical data. Pattern recognition typically involves two key steps:

Feature extraction

Classification

In feature extraction, one or more discriminative metrics are calculated using the input data. These metrics are then used in classification to assign a specific class label to the input data. For the problem of detecting cardiac murmurs, the classification is binary – assigning either a normal or murmur label to the analysed data.

!!!!!!!!!

!2

PES INSTITUTE OF TECHNOLOGY - BSC

Page 3: Diagnostic Stethoscope Project Synopsis

OBJECTIVE !The purpose of this project is to design and implement a digital stethoscope to serve as a platform for potential computer aided diagnosis applications for the detection of cardiac murmurs. The system uses a custom-built sensor to capture heart sounds and convert them into electrical signals to be processed and visualised in real time using MATLAB with the help of an Arduino Uno board.

System Design and Overview The system Hardware Mainly Consists of two components namely:

Sensor (Piezoelectric Transducer Circuit) Arduino Uno

The Software Part is heavily based on MATLAB where the obtained electrical signal from the sensor is processed and visualised in real tie. !The overview of the system can be diagrammatically represented as !!

!!

!!!!!

!3

PES INSTITUTE OF TECHNOLOGY - BSC

Electrical Signal Derivation from Acoustic Signal

SENSOR

Signal Capturing Interface

PC Interfacing

ARDUINO UNO

Real Time Audio Processing

Real Time Visualisation

MATLAB

Page 4: Diagnostic Stethoscope Project Synopsis

!HARDWARE DESIGN & IMPLEMENTATION ! ACOUSTIC SENSOR AND AMPLIFIER CIRCUIT !The piezoelectric sensor in the acoustic sensor needs to be biased in order for proper operation. In addition, the output of the piezoelectric sensor is on the order of millivolts, which is relatively small in magnitude. This makes it challenging for the Arduino to detect changes in sensor output. In order to address both these issues, a bias and amplifier circuit was designed and implemented to interface the raw sensor output with the Arduino Uno board.

"

!!Murmur Detection

Dataset As the aim of the project states our main objective to acquire data from a patient in real time and process it and diagnose the heart condition. But in order to do that we need to train the algorithm with many datasets, i.e build up a knowledge base upon which the decisions are made. The knowledge base would contain data. Sound based on all major types or frequently occurring heart conditions it would also contain a repository of normal heart sounds too. An Example for the datasets that would be collected, is given as:

!4

Page 5: Diagnostic Stethoscope Project Synopsis

Fig1: Heart Sound plot of a Normal Heart with no Abnormalities

Fig2:Heart Sound plot of a patient diagnosed with ‘Mitral Regurgitation’ !!!!5

Page 6: Diagnostic Stethoscope Project Synopsis

Feature Extraction In this step we try to single out features unique to the murmur. As the characteristic of every murmur is different with respect to the other in a specified domain. This can be done through methods such as:

Mean and Variance using pdf.

Low Energy Rate Using RMS Equation.

Frequency Component.

and so on…

We aim to find the best suited among the above mentioned for the algorithm we develop and implement it in the prototype.

!CONCLUSION !This may not be the first implementation of such a device but it sure as well won’t be the last, the algorithm that we develop in the end is what’s vital, it is the soul of the project. The algorithm will determine the efficiency of the prototype but will also help improve its knowledge base, which is vital as the better the knowledge base then better the prototype. Thus, the algorithm will itself be the main achievement of our project.

Any future implementation of the same could build upon our prototype to make it better, as it will hold a great deal of potential for improvement and room for customisation based on need.

!

!6