Real-Time Measurement of Parkinson’s Tremors Sponsor: National Institute of Aging (Under Grant #R25 AG046114) Mentor: Aaron Crandall Author: Biswaranjan Das CASAS Center for Advanced Studies in Adaptive Systems Brief Background Parkinson’s Disease: • Progressive nervous system disorder. • Major Symptom is 3-7Hz tremors. • Affecting over 7 million people worldwide. • Proper pharmaceutical treatment requires frequent tracking of tremor severity • Objective tremor measurement via visual inspection which is subjective and often unreliable Intrusive and their weight can mask tremor symptoms. Previous Works ○ Senior Design Project at EECS, Washington State University. ○ Used Microsoft Kinect V2. ○ Implemented a pipe-and-filter architecture. ○ Using OpenCV and a FFT module. Not real-time due to the complexity of the code. Lot of constraints and some bugs. + ○ Mahew J. Johnson, W. University in St. Louis “Detection of Parkinson Disease Rest Tremor” Glossary OpenCV: Open Computer Vision FFT: Fast Fourier Transform F200: Intel Camera Frontal Series 200 Project Goals 1. Find an aﬀordable and mobile vision system for analysis of hand tremors. 2. Implement a cross-platform web-based interface for viewing live and historical hand movement data. 3. Validate the data obtained from vision processing with traditional accelerometer with clinical metrics. INTEL™ REALSENSE F200 FFT Module ↓↓ Weighted Freq. Fourier Linear Combiner (WFLC) *Allows for continuous tracking of frequency and amplitude modulations. 22 Hand Points 3 cameras : 1 IR + 1 RGB + 1 IR Projector. > In-built Image Processor >Provides <1mm resolution between 0.2-0.85 meters. >Shipped with major OEMs starting 2015. Already in market. > Cheap, small and cross-platform*. Lightweight Webserver Running on Subject’s Laptop or Home. All Major Browsers Client can be used be anyone like caregivers or family. Uses Python 3: Cross-Platform -> 1. Python3 Powered: WFLC Module => RabbitMQ 2. Flask Web Framework: Werkzeug Server (WSGI: PEP 3333) Jinja 2 Templates 3. Socket.IO For Websockets and fallback bi-directional comm. Validation Tools/Devs Shimmer Sensing V3 : Accelerometer > Low-Noise 2G Output > Bluetooth Low-Energy > Light weight. Results 1 Realsense F200 Driver 2 RabbitMQ Exchange 3 Postgres DB for storage 4 Shimmer V3 Data to RMQ 5 Webserver setup 6 Data Collection (initial) 7 Data Analysis The measurements are now almost real-time. Works on all major browsers. Graphs are generated on the ﬂy. Live Demo can be viewed at ntg-reu.eecs.wsu.edu now. Special Thanks Presentation/References Diane Cook, Aaron Crandall, Quinn B., Declan E. www.ntg-reu.eecs.wsu.edu/bd Project Design Implementation / Tech Stack

Real-Time Measurement of Parkinson’s Tremors …ntg.ailab.wsu.edu/posters/2015/BishuDas.pdfSenior Design Project at EECS, Washington State University. Used Microsoft Kinect V2. Implemented

Download PDF Report

Upload
others
View
1
Download
0

Embed Size (px)

Citation preview

Page 1: Real-Time Measurement of Parkinson’s Tremors …ntg.ailab.wsu.edu/posters/2015/BishuDas.pdfSenior Design Project at EECS, Washington State University. Used Microsoft Kinect V2. Implemented

Real-Time Measurement of Parkinson’s TremorsSponsor: National Institute of Aging (Under Grant #R25 AG046114)Mentor: Aaron CrandallAuthor: Biswaranjan Das

CASASCenter for Advanced Studies in Adaptive Systems

Brief BackgroundParkinson’s Disease: • Progressive nervous system disorder. • Major Symptom is 3-7Hz tremors. • Affecting over 7 million people worldwide.

• Proper pharmaceutical treatment requires frequent tracking of tremorseverity• Objective tremor measurement viavisual inspection which is subjectiveand often unreliable � Intrusive and their weight canmask tremor symptoms.

Previous Works

○ Senior Design Project at EECS,Washington State University.○ Used Microsoft Kinect V2.○ Implemented a pipe-and-filterarchitecture.○ Using OpenCV and a FFT module. � Not real-time due to the complexity of the code. � Lot of constraints and some bugs.

○ Ma�hew J. Johnson, W. University in St. Louis “Detection of Parkinson Disease Rest Tremor”

GlossaryOpenCV: Open Computer VisionFFT: Fast Fourier TransformF200: Intel Camera Frontal Series 200

Project Goals1. Find an affordable and mobile vision system for analysis of hand tremors.2. Implement a cross-platform web-based interface for viewinglive and historical hand movement data. 3. Validate the data obtained from vision processing with traditionalaccelerometer with clinical metrics.

INTEL™ REALSENSE F200

FFT Module ↓↓Weighted Freq.Fourier LinearCombiner (WFLC)*Allows for continuous tracking of frequency and amplitude modulations.

22 Hand Points3 cameras : 1 IR + 1 RGB +1 IR Projector.> In-built Image Processor>Provides <1mm resolutionbetween 0.2-0.85 meters.>Shipped with major OEMsstarting 2015. Already in market.> Cheap, small and cross-platform*.

Lightweight WebserverRunning on Subject’s Laptop or Home.

All Major BrowsersClient can be used be anyone likecaregivers or family.

Uses Python 3: Cross-Platform ->

1. Python3 Powered: WFLC Module => RabbitMQ2. Flask Web Framework: Werkzeug Server (WSGI: PEP 3333) Jinja 2 Templates3. Socket.IO For Websockets and fallback bi-directional comm.

Validation Tools/Devs

Shimmer Sensing V3 : Accelerometer > Low-Noise 2G Output > Bluetooth Low-Energy > Light weight.

Results

1 Realsense F200 Driver2 RabbitMQ Exchange3 Postgres DB for storage4 Shimmer V3 Data to RMQ5 Webserver setup6 Data Collection (initial)7 Data Analysis

� The measurements are nowalmost real-time.� Works on all major browsers.� Graphs are generated on the fly.� Live Demo can be viewed atntg-reu.eecs.wsu.edu now.

Special Thanks

Presentation/References

Diane Cook, Aaron Crandall,Quinn B., Declan E.

www.ntg-reu.eecs.wsu.edu/bd

Project Design

Implementation / Tech Stack