Locate in Lungs
 Locate in Lungs Complete Hardware Setup |  The 8-channel MEMS microphone array for acquisition of Lung sound by implementing around the user's chest. |
|---|---|
 The data visualization interface with flexibility of choosing the desired channel (upto 8 nos.), sampling rate (upto 96 kHz), Filter type, etc. The plot represents to speech being recorded during the testing. |  The AD7768 time-synchronized Analog-Digital-Converter (ADC) with maximum 8-channel input support. |
 The AD7768 Evaluation Board for connecting the ADC to the PC/Laptop. |
Project Multichannel Sensor-based Adventitious Lung Sound Localization Algorithms and Assessment using 3-D Printed Thoracic Phantom, which I love to call Locate in Lungs, is an initiative to develop a a system to non-invasively localize abnormal lung sounds so that doctors can quickly identify the infected or affected portion of the lung, which is the same as the origin of the adventitious sound. In this project, I worked in the development of the multichannel acoustic array hardware system which will be implemented across the user's chest and used for lung data acquisition.
Applications: A non-invasive device for inside the lung disease source identification.
Timeline: May, 2019 - July, 2019
Collaborator(s): Prof. Prasanta K. Ghosh (Indian Institute of Science - IISc, Bangalore)
Theory behind Working: As we know medical practitioners use stethoscope to listen to lung sounds. Also, they place the stethoscope the different position of the chest (front and back) to better identify the probable source of the abnormal lung. As this project tries to automate this process of lung sound origin detection, the hardware system will also require multiple microphones to acquire the lung sound from different directions simultaneously. In the hardware design, two parameters are necessary: 1) Simultaneity 2) Time Synchrony.
For the localization of sound sources, the Time Difference of Arrival (TDOA) at the different acoustic sensors is an important feature from which the location of the origin can be calculated geometrically. However, if the data acquisition in different channels are delayed due to hardware, calculation of the exact TDOA will not be possible; consequently the algorithm will produce in appropriate result. Thus, I developed the time synchronized hardware setup which incorporates the following components/features:
-
Up to 8 microphones integration in single system
-
An amplifier circuit with variable gain for each microphone channel.
-
Time synchronized data acquisition with sampling rate up to 96 kHz.
Here I developed a digital filtering algorithm to remove the DC and Analog-to-Digital-Conversion noise and hardware non-linearities from the acquired audio data.