Rugged encapsulation structures and acoustic instrument development for inexpensive medical sensors
Mentors: K. Mani Chandy and Julian Bunn
Abstract - Public health in developing countries is limited by insufficient and inaccessible medical care. An innovative way to address this issue is by means of a health system capable of operating without direct interaction with a physician. The method proposed here consists of portable cellphone based medical sensing devices that can be used by members of the local society. The instruments collect medical data and upload it into the Cloud where they are evaluated and can be assessed by physicians. Two such devices proposed are auscultation and electrocardiograph (ECG) tools, both of which exist as prototypes. The problems with these devices are structural. The ECG and stethoscope lack robust casings and can break easily. The key elements of this study include creating an encapsulation for the current auscultation unit and developing a combined stethoscope and ECG device. We have investigated different acoustic shapes and optimized them for microphone recordings in the stethoscope.
The lack of an established medical infrastructure, physicians, and medical facilities make care inaccessible to many patients in developing countries. Individuals that reside in rural areas and villages often have to travel a full day to see a doctor or visit a hospital. Medical care is also a financial burden on the citizen in need. These obstacles often create severe situations for treatable conditions. Interestingly, in the same regions where medical care is scarce, modern cell phone technology is present. India, for example, offers wireless coverage to the majority of its population and the number of cell phone users is growing at a rapid rate (Figure 1). Still, India is plagued by poverty, overpopulation, and a severe lack of accessible medical treatment.
Figure 1. The figure on the left indicates the amount and location of the Indian population living below the poverty line. On the right, it is clear that there is sufficient cell phone coverage in these same areas.
The goal of this project is to develop a new method for providing healthcare to the medically under-served without the need of professionally trained healthcare professionals. Our research focuses on developing sensor and computer systems that offer low-cost medical evaluations to those in less than favorable economic and living conditions. This network will use medical sensors, cell phones, tablet computers, cloud computing devices, and algorithms to allow a person with minimal training to administer a physical exam. The idea is to upload medical information recorded with sensors (stethoscopes, capillary refill meters, electrocardiograms (ECG), blood oximeters, blood pressure instruments, thermometers, etc.) from a cell phone to the Cloud. Equipped with machine-learning algorithms and easy access to panels of human experts, the Cloud carries out an evaluation of the information and classifies basic results in a triage (“Most likely healthy”,”Unsure”,”Seek further medical help”).  In a broader sense, the collection of data from multiple patients allows physicians to analyze general health problems statistically.
Researchers are using cameras, microphones, audio tools, and many common built-in features of cell phones to generate medical testing devices.With this, university research groups have produced a number of cell phone based instruments. The University of Pittsburgh developed HeartToGo, a cell phone medical technology capable of continually monitoring and recording real-time ECG signals, generating a cardiac health
summary, and detecting certain abnormal cardiac conditions. The Massachusetts Institute of Technology (MIT) produced Catra, a sensor that beams light across the eye and diagnoses patients with cataracts.  In this paper we discuss two cell-phone based devices, an electric stethoscope and otoscope, as well as an ECG tool that requires a laptop. The otoscope attaches to a Nexus Android phone and uses the built in camera to take pictures of lesions on the visible skin, mouth, or ear. The stethoscope records heart auscultations with the phone’s microphone.
This paper discusses the improvement of existing devices, primarily the stethoscope and electrocardiographing (ECG) tools. The key elements of this investigation include: creating an encapsulation for the current auscultation unit and ECG (Figure 2), as well as investigating acoustic shapes better suited to microphones. Part of the investigation focuses on developing a stethoscope suited to the use of a microphone as a detector, rather than the human ear. Traditionally, a stethoscope is a two-sided acoustic instrument that uses a bell and a diaphragm, for high and low frequency vibrations, respectively. We will determine the types of vibrations more suitable for a microphone by analyzing recordings from the instrument. Testing the optimum size for such an instrument is essential since this question will affect manufacturing details, cost, and our ability to integrate the stethoscope into a cell phone or cell phone accessible device.
Figure 2. a) Electronic chestpiece prototype that employs a PUI microphone embedded in an AllHeart chestpiece b) ECG prototype with exposed wires and insufficient electronics casing.
II. DESIGN AND FARICATION
ELECTROCARDIOGRAPH (ECG) CASING
The prototype version of the ECG combines a stethoscope and ECG recordings so both the audio and ECG signals can be viewed at the same time. This will aid the diagnosis of patients. Exposed wires and electronics complicated the first version of the ECG. We mounted the ECG unit on a glove. A custom electronics box was designed in SolidWorks and produced using a rapid prototyping machine (Figure 3). Three electrodes attach to the glove surface using epoxy. Wires leading from the electrodes to the electronics casing were threaded through the fabric of the glove to minimize exposure. A sleeve on the back side of the glove houses the casing.
Figure 3. a) SolidWorks drawing of a custom made electronics casing with inlet and outlet holes for wires and a switch. This also shows the battery holding location. b) The completed lead side of the ECG unit. c) Casing end of the ECG with completed box
The electronic medical devices have to be rugged because of their use in remote areas. It is imperative that they function reliably. Early prototypes for the stethoscope device were not robust and tended to break. A layer of tape covered the electronics involved, which were held together by glue. We have developed two types of aluminum casings that attach to the stethoscope head and protect the electronics. One goal was to generate a modular design so that if the device fails. Remote users can open, analyze, and hopefully fix it, without compromising its structural integrity. SolidWorks was used to design these alternative structures. The first design had a thin aluminum tube that clamped onto the stethoscope head (Figure 5a). The issue with this design is the clamp, which has some sharp edges. The second design is a bit thicker and is composed of two parts. The bottom part is press-fitted permanently onto the stethoscope, while the other is screwed onto the first section (Figure 5b). The second design is slightly bulkier but it minimizes sharp edges while still encasing and protecting the electronics. The final products are shown in Figure 5c.
ACOUSTIC SHAPE RESEARCH
Once we designed the stethoscope encapsulation, we wanted to develop a better acoustic shape. Though the bell-shaped stethoscope is widely used, it may not necessarily be the best for translating vibration into electrical signals. A variety of acoustic instruments, ranging from guitars to conch shells, are used every day. One of these designs, created by either man or nature, may be ideal for amplifying vibrations for a microphone rather than our own ears.
Three new stethoscope head pieces have been designed and modeled on SolidWorks. The inspiration for the shapes include a normal stethoscope, a trumpet, and an old car horn that has a looped airway (Figure 6a and d). These pieces have been printed using a Stratasys prototyping device and were tested and evaluated based on acoustics performance.
A group of medical students under Dr. George Chandy, from the University of California Irvine (UCI), have taken completed devices to India for testing. The instruments will be used in major hospitals and rural communities. The feedback the team collects is important for creating a database of medical recordings and improving existing devices.
a) b) c)
Figure 5. a) First casing design that slides onto the stethoscope chest piece. b) Second two-part stethoscope design. c) The completed aluminum casings with attached electronics.
Figure 6. Three proposed stethoscope headpiece shapes. a) This design has a large surface area. b) This design has a common straight edged cone. c) This design uses an exponentially decaying cone. d) This piece combines the exponentially decaying cone with a looped receiving end airway.
In order to observe acoustic differences between the shapes, we inserted a microphone into the back end of each piece and record the sound entering the airway. A speaker with a sound surface area larger than the diaphragm will feed white and pink noise into each piece and the microphone will record. White noise has a random signal with a flat power spectral density, whereas pink noise exhibits an exponentially decaying spectrum. By finding the power spectrum of each recorded signal and using a Fourier transform, we can compare the results to the expected response. Using Audacity, the white and pink noise audio signals were generated. The software allows us to simultaneously play the noise and record from the microphone embedded in the stethoscope. The experiment was conducted in a quiet room in order to insure accurate test results. First we tested the procedure with a bare microphone. This set a baseline comparison for the recordings and allowed us to assess the microphones limitations.
The speaker was set up 6 inches from a tripod with the stethoscope shape mounted on top. Prior to mounting, the body of each piece was wrapped in polyurethane foam. This ensures that the microphone only records sound from the entry airway of the shape. Each piece was subjected to 20 seconds each of white and pink noise. The process was repeated to ensure that the recordings were a good representation of the recording ability of the device.
Figure 7. a) Speaker setup with a microphone mounted on a tripod. b) Stethoscope piece that is wrapped in polyurethane foam. The wire connects the microphone in the piece to the computer for recording.
After conducting the first recording with the bare microphone, it was apparent that the microphone itself had limitations. Figure 8 is the power spectrum for the bare microphone recordings. Compared to the theoretical spectrum (top blue lines), the recorded signals display minimal similarities. The pink noise recorded spectra show a decline in decibel level with the increase in frequency. The white noise recording shows a large number of oscillations. The graphs are not overwhelmingly similar. Another consistent issue with the recordings is the notable difference between the output decibel level of the generated signal compared to the recording. Both graphs, white and pink noise recordings, also exhibit a decline in decibel level when the frequency is above 15000 Hertz. This can be attributed to the microphone since it appears in subsequent tests. Fortunately, the devices are being generated for heart auscultation recording. Typical heart sounds register audible frequencies much smaller than 15000 Hertz. Taking this into account, the considered frequency range will be from 0 to 2000 Hertz when comparing power spectra for the different designs shown later.
Figure 8. The top blue line shows a theoretical expectation for the power spectrum of the noise recording. The bottom line shows the actual spectra from a recording using a bare microphone.
After testing was completed, the power spectra from the recordings were compared by overlapping them on the same graph. The power spectra related to the white noise recordings allow for a few comparisons, the first between peaks in the graphs (Figure 10). The graphs generated for the white noise recording of all the shapes, including the bare microphone, displayed similar peaks up until 1400 Hertz. It is apparent that comparisons between graph features are not possible. The amount of data points is too small. However the data points appear to be similar in each graph. The decibel levels on both the white and pink noise recordings are similar (Figure 10). Interestingly, the recording decibel level of both white and pink noise is increased when the cone and the exponentially decaying cone were used. Both these shapes cover comparable surface area
Figure 10. a) White noise power spectra for the three designs. b) Pink noise power spectra for the different stethoscope shapes.
There was not a clear distinction between the performances of the shapes. The second and third designs appear to have better recording amplitude than the other recordings, though it is still significantly lower than the generated amplitude. The next step for improving the device is testing the performance of the shapes when they are fabricated out of different materials. The rapid prototyping machine, used to make the tested shapes uses an ABS plastic material, a lightweight material used for plastic molding that might not be as robust as desired. The plastic may have absorbed some of the sound at the frequencies we wish to amplify.
Currently, the devices that demonstrate functionality use an existing stethoscope headpiece and add electronics casings. In the future these pieces will also be made using non-metal material and tested for reliability. In order to reduce the cost of this medical initiative, manufacturing techniques and material options will be explored.
I would like to express gratitude toward my mentors Julian Bunn and Mani Chandy for their guidance throughout the project, as well as my fellow SURF students in the infospheres group. As a team, we also greatly appreciate the students and faculty at UCI for their help in testing our devices. I would also like to thank John van Deussen and the Mechanical Engineering department for use of the equipment in the ME shop. Finally, thanks to Fred and Jean Felberg for their contribution to this project and the SFP office for allowing us to participate in the SURF program.
1 10 cent Medical Checkups for the Developing World, http://www.infospheres.caltech.edu/10cent_checkup/
2 Victor Chu, Connecting Medical Devices to Cellphones, Masters Thesis, Caltech, June 2010.
3 Danny Petrasek, Alan Barr, and Krishna V. Palem. 2010. The virtual hospital: the emergence of telemedicine. In Proceedings of the 2010 international conference on Compilers, architectures and synthesis for embedded systems (CASES '10). ACM, New York, NY, USA, 53-54.
4 Zhanpeng Jin, Joseph Oresko, Shimeng Huang, and Allen C. Cheng. HeartToGo: A Personalized Medicine Technology for Cardiovascular Disease Prevention and Detection. University of Pittsburgh. 2009.
5 Tim Hornyak. MIT smartphone clip-on detects cataracts in minutes. June 2011.