Creating Biomedical Technologies to Improve Health

2017 BESIP Project

Occupational Therapy
Walter Reed
Mentor Name: 
Mark Lindholm
Mentor Telephone: 
Computational Bioscience and Engineering Laboratory, Office of Intramural Research
Mentor Name: 
Thomas Pohida Project #4
Randall Pursley
Mentor Telephone: 
(301) 435-2904

Laboratory and Project Description

Walter Reed National Military Medical Center (WRNMMC) is one of the nation’s largest and most renowned military medical centers. They serve the needs of handicapped and disabled veterans, including those severely injured during military service. For those soldiers who have sustained an amputation or traumatic brain injury, WRNMMC provides prosthetics and rehabilitation to help the veterans regain their mobility and independence.
We are currently developing an audible-key-identifier keyboard assistive technology for patients with visual and/or physical limitations. In short, without a user physically touching the keyboard, the system provides audio feedback indicating finger proximity hovering over any key. Given the audio feedback, the user will know when it is appropriate to move their finger or other prosthetic downward to press a key. This video-based add-on system enables specific patients to use a standard computer keyboard for accessing a desktop computer and running any program. The audible-key-identifier system design and implementation is based on real-time video processing, and is comprised of custom hardware and software. The current system design uses a camera to stream video of the user’s keyboard and finger. The video processing software determines the real-time fingertip location relative to pre-defined keyboard image regions corresponding to each key. As the fingertip is moved around the keyboard, the software continuously updates the computer voice audio to inform the user of fingertip position over keys. This first generation system utilizes a separate computer, a basic web camera, ambient room lighting, manual keyboard key-region calibration, a fiducial mark on the fingertip, and rather routine video processing methods to segment the marker.
A second generation audible-key-identifier system must be developed to improve performance and reliability. Two example design enhancements include tolerance to changes in ambient room lighting, and automated calibration of key regions in the video image. At least two new approaches will be evaluated to achieve these improvements. First, we will consider incorporating wavelength specific lighting (e.g., near-infrared), and correspondingly limiting camera wavelength sensitivity. Second, we will investigate the use of a Microsoft Kinect as a fingertip detection device, which would eliminate the existing use of a camera and need for illumination. Other system improvements will require the development and testing of more advanced video processing algorithms and methods.
The BESIP student working on this project would be involved in the design, prototyping, and testing of biomedical instrumentation and methods. Working closely with the interdisciplinary team the intern will gain valuable hands-on biomedical engineering experience with multiple procedures and technologies including sensors, optics, scientific programming (e.g., LabVIEW or MATLAB), data acquisition, real-time signal and video processing, motion analysis, laser-cutting and 3D-printing for rapid prototyping, and clinical considerations and method development. The work for this project will take place on the neighboring campuses of NIH and WRNMMC.
Lindholm lab: The mission of the Assistive Technology Program is to promote independence through training and use of tools and technology, and to support health care providers and patients as they explore solutions to existing and new functional challenges. The Assistive Technology Program at Walter Reed is based on the Human Activity Assistive Technology (HAAT) model, which has been localized for Walter Reed National Military Medical Center - Bethesda. The HAAT model provides a framework for understanding Assistive Technology services. The four components of the model are the patient, the activity, the assistive technology, and the context in which these exist. Enabling a patient to carry out an activity is the goal. Areas of patient activity include activities of daily living, work / productive activities, and leisure activities. The context includes physical, social, cultural and institutional aspects of the environment in which the patient carries out the activity. The patient's abilities and the context in which an activity will be carried out are considered when selecting an assistive technology. The ultimate goal is to provide our patients every opportunity to succeed and return to an independent and quality life.
Pohida lab: Provides electrical, electronic, electro-optical, mechanical, computer, and software engineering expertise to NIH projects that require in-house technology development. Collaborations involve advanced signal transduction and data acquisition; real-time signal and image processing; control and monitoring systems (e.g., robotics and process automation); and rapid prototype development. Collaborations result in the design of first-of-a-kind biomedical/clinical research systems, instrumentation, and methodologies.