Self-Test Hearing Diagnostic for the PDA

Document Sample
Self-Test Hearing Diagnostic for the PDA Powered By Docstoc
					Self-Test Hearing Diagnostic for the PDA
                                   By:
Nizar Ghneim, Sara Siddiqui, George Mwangi, Catrina Nelson, Ebrima Ndure

                        Under the Supervision of:
                          Murat Torlak, Ph.D.

                    The University of Texas at Dallas
                   Eric Jonsson School of Engineering
                  Department of Electrical Engineering

                             May 10, 2010




                               Page 1 of 14
Table of Contents
1     Introduction ...............................................................................................................................4
2     Achievements.............................................................................................................................4
    2.1     The Developed Software ............................................................................................................... 4
      2.1.1        Semester One ........................................................................................................................ 4
      2.1.2        Semester Two ........................................................................................................................ 5
    2.2     Clinician Visit ................................................................................................................................. 5
    2.3     Strategy ......................................................................................................................................... 6
      2.3.1        Flowchart .............................................................................................................................. 6
      2.3.2        Gantt Charts .......................................................................................................................... 6
    2.4     Results from Exams ....................................................................................................................... 9
3     Lessons Learned and Future Work ...............................................................................................9
    3.1     Problems Encountered ................................................................................................................. 9
      3.1.1        Semester 1 ............................................................................................................................. 9
      3.1.2        Semester 2 ........................................................................................................................... 11
    3.2     Design Day................................................................................................................................... 11
      3.2.1        Suggestions Received on Design Day I ................................................................................ 11
      3.2.2        Conclusion after Evaluating the Suggestions ...................................................................... 12
    3.3     Future Project Ideas for Senior Design ....................................................................................... 12
      3.3.1        Suggestions for PDA program ............................................................................................. 12
      3.3.2        To make more difficult, say for the next level of training: .................................................. 12
      3.3.3        Other Ideas: ......................................................................................................................... 12
4     Conclusion................................................................................................................................ 13
    4.1     Webpage, Source Code, and Documentation............................................................................. 13
    4.2     Discussion.................................................................................................................................... 13
5     Acknowledgements .................................................................................................................. 13
6     References ............................................................................................................................... 14




                                                                    Page 2 of 14
Table of Figures and Tables
Figure 1:    Screenshot from Exam .................................................................................................................. 4
Figure 2:    Flowchart of Software ................................................................................................................... 6
Figure 3:    Gantt Chart for Semester 1's Work ............................................................................................... 7
Figure 4:    Gantt Chart for Semester 2's Work ............................................................................................... 8
Figure 5:    Sample Results File........................................................................................................................ 9


Table 1: Summary of Results ...................................................................................................................... 10




                                                                   Page 3 of 14
1 Introduction
This project was initiated as a response to a request for proposal from Philip Loizou, Ph.D. He was
hoping to design a graphical user interface (GUI) that would allow patients who have hearing disabilities
to carry out self-tests on their ability to identify the 11 vowels in the American English at their leisure.
The software would be a good tool not only for the patients carrying out the tests, but also for the
doctors who rely on test results to evaluate their patients’ conditions. The software would save time
spent at the doctor’s office conducting tests and therefore allow the doctor to spend more time
addressing the patient’s problem and plan of treatment. By the end of this two-semester endeavor, we
coordinated with an audiologist from the UT-Dallas Callier Center, Linda Thibodeau, Ph.D., and designed
a product as per her specifications.


2 Achievements
2.1 The Developed Software
2.1.1 Semester One
During the first semester under the supervision of Dr. Philip Loizou, the team was able to design a fully
functional part of the GUI: a test using a set of vowels on a Portable Data Assistant (PDA). The GUI
allows the patients to hear eight words that all begin
with the letter H and end with the letter D with their
only difference being the vowels in between the two
letters (i.e. "hood" or "head" or "heed"). With the repeat
button, patients can opt to hear the randomly selected
words as many times as they wish before choosing the
word they think they heard from a list. Once they make a
selection, the program will inform the patient whether
they correctly identified the word or not and then
proceed to read out another randomly selected word
and the patients repeats the steps mentioned above. Th
is loop continues until the patients choose to view their
progress so far by pressing the 'See Results' button.

The results were displayed as an 8X8 confusion matrix
which should have a value of one displayed diagonally
from the top left corner to the bottom left corner if the
patient successfully chooses the correct words read out
by the program. If the patient selected the wrong word,
the program will display a value of one at the
intersection of the correct word row and the column of
the selected word. These results are then saved onto the
PDA's hard drive for doctor’s review.
                                                                         Figure 1: Screenshot from Exam

                                               Page 4 of 14
2.1.2 Semester Two
During the second semester under the supervision of Dr. Murat Torlak, the team coordinated heavily
with the UT-Dallas Callier Center's audiologist Dr. Linda Thibodeau. Using her expertise as a guide, we
then began to design a fully function prototype as per her specifications. The key elements wanted
available were to have multiple types of tests (such as tests for high and low frequency sound words),
the ability to take and record multiple tests, a questionnaire at the beginning that requested certain
information about the current patient who will take the test, and the ability to measure the ambient
noise of the room the patient is in.

We are proud to announce that we succeeded in accomplishing each one of these requirements. The
software begins by reading a "user.txt" file to automatically fill in the information in the questionnaire.
Once the user completes the questionnaire, the "user.txt" is updated so that it reflects the last patient
who has taken the test. This way, if a single patient is using the software alone, he or she would not
have to fill out the same basic information every time. All this information is also saved as a part of the
results of the test.

After the questionnaire, the patient is prompted to maintain silence for 5 seconds while the program
records the ambient noise in the room. Using C#.net, we created a P/Invoke library as outlined on the
MSDN website with a few minor adjustments. As the software accesses the microphone of the device,
the data, instead of being saved into *.wav file, was stored into an integer array. Taking the root mean
square (RMS) of this signal allowed us to output the ambient noise a single value, the RMS of the signal
recorded during these 5 seconds. The code can easily be adjusted so that the measurements are taken
in the background and various noise samples are taken throughout the test.

After the software records the ambient noise, the patient has the option of choosing one of three tests,
the "Vowel Sounds" test that was available during the first semester, the "High Frequency" test (i.e.
"pick" or "click" or "thick"), and the "Low Frequency" test (i.e. "blob" or "job" or "mob"). The software
requires the patient to go through a "practice" session that requires them to listen to each word at least
three times before being able to take the actual test. This ensures that the patient is familiar with the
words he or she is going to be tested on.

The test is conducted the same way it was during the first semester; the software plays a word and the
user has the option to repeat the word as many times as they want before choosing the word that they
heard. Once the test is complete, the data is saved in the "Results" folder as a text file, each with a
different name. The results file displays the the information the patient submitted during the
questionnaire, the value of both the RMS and variance of the noise, as well as the results of the test in a
similar confusion matrix.

2.2 Clinician Visit
The team was able to visit a physician who was performing a test on a patient. The doctor had a
computer that was loaded with a program which read out the sentences the patient was supposed to
identify by listening through a cochlear implant. The patient repeated the sentences that they heard and
the doctor evaluated their response.


                                                Page 5 of 14
Observing the test being conducted gave the team a sense of deeper understanding of the patients’
needs as well as an understating of the need to being able to provide a solution to a real world problem
that affects millions of people. The advantages of having self-test software on a portable device were
immediately highlighted by this visit.

2.3 Strategy
2.3.1 Flowchart
We provide a simplified overview of how our program is design below in the form of a flowchart. This
flowchart provides only a simple overview of the flow of the software:




                                       Figure 2: Flowchart of Software

2.3.2 Gantt Charts
We planned our two semesters work according to our abilities and time. The following two pages
summarize our efforts in Gantt Charts:


                                               Page 6 of 14
                                                              Semester 1 Gantt Chart
                        8/25/2009           9/14/2009                10/4/2009             10/24/2009               11/13/2009          12/3/2009


       Initial Team Meeting   8/31 1 2

      Project Resource and
                                    9/3/   7     10
     requirements research

Proposal Letter preparation                9/14/ 1 4

  Software Skeleton Design                     9/17/    6           12

Software development tools
                                                       9/24/ 1 2
        acquisation

        Word List Research                                  9/28/          14              15

           Software
                                                                                10/12/                  38                                  23
      Development/Coding
      Software Testing and
                                                                                                                         11/20/    6        8
          Modification

   Final Report Preparation                                                                                                      11/26/ 4       4

   Project Presentation and
                                                                                                                                            12/4/ 1 3
        Demonstration


                                               Start Date          Duration (Days)       Hours Spent         Time Left



                                                        Figure 3: Gantt Chart for Semester 1's Work




                                                                         Page 7 of 14
                                                     Semester 2 Gantt Chart
                             1/2/2010          1/22/2010         2/11/2010            3/3/2010       3/23/2010      4/12/2010            5/2/2010   5/22/2010

 Design day comments analysis       1/14/ 1 3
    Graphics incorporation and
                                        1/14       12       9
       Iphone app Research
     Callier Center Presentation                  1/27 11

  Prof. Thibodeau's suggestions
                                                   1/29/ 12
             analysis
        New program Flowchart
                                                           2/2/ 8        4
             Development
Vowels,High and Low Frequency
                                                                      2/15       15         12
        words Research
 Acquired Words modifications                                                         3/4/ 4 2

 Software Development/Coding                                                             3/10/              36                     15
          Software Testing and
                                                                                                                        4/18/ 4      5
              Modification
       Final Report Preparation                                                                                              4/24/       6 22
       Project Presentation and
                                                                                                                                    4/30/ 12
            Demonstration
                                                            Start Date        Duration (Days)    Hours Spent     Time Left(Days)




                                                            Figure 4: Gantt Chart for Semester 2's Work



                                                                             Page 8 of 14
2.4 Results from Exams
 Upon completion of the software, our team provided the
Callier Center clinician Dr. Thibodeau with a PDA with our
software loaded. She asked four of her hearing impaired
patients, as well as four of her graduate students (who did
not have any hearing impairment) to take the test. We
extracted the data from the txt files located in the Results
folder of our software and summarized them in Table 1
(see next page). As can be seen from the table, the test
was almost always completed to perfection amongst her
students who had no hearing impairment; however the
patients with at least some hearing loss were often unable
to make a perfect score on the test.

The confusion matrix (seen right in Figure 5) is very simple.
The rows of the matrix correspond to the words that were
presented to patient, while the columns correspond to the
words selected. This is a very insightful way to display
one’s results. This allows patients to see what words they
missed, as well as what words they most often mistook
them for. A patient who makes a perfect score on the
exam would have number only along the diagonal.                            Figure 5: Sample Results File



3 Lessons Learned and Future Work

3.1 Problems Encountered
3.1.1 Semester 1
The idea was to provide a solution through self-test software on a portable device like PDA, iphone or
any other smart phone. While doing that, we encountered several challenges.

Lack of documentation for the programming the PDA was a major setback. Programming the PDA is
most often done on an industry level scale, as most people do not need to program the PDA for their
own use. Because of this, there is very little documentation available online regarding this. As most of
the program could be copied directly from the Windows 7 to the Windows Mobile platformed VB.net,
we only had trouble with learning how to play *.wav files.

Once the PDA was programmed, the next challenge was to get the program to execute as needed. We
were able to successfully simulate running the program on a virtual PDA using Microsoft Visual Studio
2008's Windows Mobile 6.0 SDK for Pocket PC but upon downloading the program on the actual PDA,
the audio files we were using could not be played. Every time we tried to play the files, the program
would return an error stating that the files were inaccessible. We soon discovered that the audio files


                                               Page 9 of 14
                                                      Table 1: Summary of Results
Examinee    Age   Comments                           Test       % correct      Most missed word    Environment   Noise Level (RMS)

                                                     High       100%           n/a                 Lab           0.53
                  - Mild sloping to severe HF loss
Patient 1   79                                       Low        100%           n/a                 Lab           4.94
                  - Without hearing aide
                                                     Vowel      100%           n/a                 Lab           3.56

                                                     High       87.5%          Flick, Pick, Kick   Lab           0.53
                  - Severe hearing loss
Patient 2   45                                       Low        95.8%          Blob                Lab           0.52
                  - With hearing aide
                                                     Vowel      87.5%          Who’d               Lab           2.22

                                                     High       100%           n/a                 Lab           0.58
                  - Normal to moderate HF loss
Patient 3   56                                       Low        100%           n/a                 Lab           0.67
                  - Without hearing aide
                                                     Vowel      100%           n/a                 Lab           0.61

                                                     High       95.8%          Pick                Outside       19.51
                  - Normal to moderate HF loss
Patient 4   53                                       Low        91.7%          Mob, Bob            Outside       12.35
                  - Without hearing aide
                                                     Vowel      91.7%          Hid                 Outside       1.87

                                                     High       100%           n/a                 Lab           0.64
                  - Normal hearing
Student 1   26                                       Low        100%           n/a                 Lab           0.59
                  - Graduate student
                                                     Vowel      100%           n/a                 Lab           0.76

                                                     High       100%           n/a                 Lab           1.66
                  - Normal hearing
Student 2   34                                       Low        95.8%          Mob                 Lab           0.67
                  - Graduate student
                                                     Vowel      91.7%          Who’d               Lab           0.78

                                                     High       100%           n/a                 Lab           0.62
                  - Normal hearing
Student 3   23                                       Low        100%           n/a                 Lab           0.77
                  - Graduate student
                                                     Vowel      100%           n/a                 Lab           0.72

                                                     High       100%           n/a                 Lab           0.83
                  - Normal hearing
Student 4   33                                       Low        100%           n/a                 Lab           0.84
                  - Graduate student
                                                     Vowel      100%           n/a                 Lab           1.03


                                                             Page 10 of 14
could not easily be incorporated into the *.exe file but rather, had to be downloaded with it. Once we
did this, we managed to find the directory locally, rather than hard coding it.

Initially the program had words divided by grade level but after discussions and inputs, we realized the
need to divide words based on frequency pitch of the word. A sample test program now includes noise
frequencies like high and low.

To increase the versatility of the program, we wanted to be able to modify the program to run on
iPhones or other smart phones. Initially it seemed like we would need a Mac and learn how to program
using Mac applications but the team members did not have any Mac. So the option did not work viable.
Finally an alternative way was researched. The flow chart was redesigned and adapted to place the
program on PDA and other smart phones.

3.1.2 Semester 2
During the second semester we were able to meet with Dr. Thibodeau and students from the Callier
Center. During this meeting suggestions were put forth to expand the scope of the program on the PDA.
One of the suggestions included having the PDA vary the frequency of the word during the test session.
After thorough research of this suggestion the team was not able to find a code to successfully perform
this task.

The team's initial idea for the PDA was to have it connect to the patients hearing aid via bluetooth. This
idea was presented to Dr. Thibodeau during the presentation. However the team learned that bluetooth
hearing aids were not readily available for patients due to lack of development and high cost. Instead
available hearing aids operate on an FM signal so the PDA would have to output the words through an
FM signal.

Some other problems that were encountered during this semester included getting the PDA to measure
ambient noise during the testing session and report this information in decibels after the session was
over. The team did extensive research to see if there was code available that would accomplish our
desired goal. The team was not able to find code to do exactly what was wanted, but it was possible to
design the code so that the background noise could be recorded and then calculate the RMS value.

3.2 Design Day
3.2.1 Suggestions Received on Design Day I
During Senior Design Day, we had a few guests visit our booth and some of them offered suggestions on
what we could do to improve the software. One of the suggestions that we heard from various guests
was to design an application that could run on an iPhone.

The guests pointed out that our program would be even more helpful if we could make it more versatile
by making it compatible with some of the more popular phones of the day such as the iPhone, Droid or
the Blackberry. This would ensure that the patients are not forced to buy a PDA to use the program.




                                              Page 11 of 14
3.2.2 Conclusion after Evaluating the Suggestions
The team received feedback and suggestions from the guests and immediately went into evaluating the
ideas brought forward and their practicability within the set budget. We determined that it would be
impractical to develop the iPhone application due to budget constraints and lack of apple specific
application developmental tools.

Developing applications for the other phones would also stretch our budget and would require a lot of
research and development time. The team felt these tasks could not be accomplished in one semester.

3.3 Future Project Ideas for Senior Design
After completing the tests on several of her patients and students, Dr. Thibodeau gave us the following
recommendations (note: this section and the following sub-sections are her words):

3.3.1   Suggestions for PDA program
       Results of RMS only to be 2 decimal places
       Prompt user to be quiet for measuring so user knows the test is in progress
       Provide msg that says results of ambient test (everything is ok, press ok to proceed)
       Give instructions after the practice segment so user knows what will happen next
       Tell user they will be getting feedback throughout the test and consider giving correct response
        when they guess incorrectly
       Show results at the end in percentage
       Revise Random number generated so that words presented at least once and no more than 2
        same words in a row possible; one person reported getting same word 3 times in a row and
        never getting another one

3.3.2   To make more difficult, say for the next level of training:
       Add background noise
       Different speakers (male and female)
       Increase length of words (phrases to sentences; might reduce set to four because of small
        screen)
       Increase number of words (hick, sick, thick)
       Repeat digits

3.3.3   Other Ideas:
       Data logging function to keep track of how long the program was used and whether it was done
        with or without hearing aid
       Consider expanding routine to larger screen (iPad or PC based) that might also be used to play
        video to practice speech reading




                                             Page 12 of 14
4 Conclusion

4.1 Webpage, Source Code, and Documentation
We have uploaded all of our work onto the following webpage:

                                  http://www.utdallas.edu/~nsg051000

We have decided to publish our code under a Creative Commons Attribution-Noncommercial 3.0 United
States License and would welcome anyone to use or improve the code as the please, provided our
consent and the acknowledgement of our work. We also provide *.cab files that can be used to directly
install the application onto any Windows Mobile device. We provide two separate *.cab files; one that is
optimized for WQVGA resolution (240x480) and one that is optimized for QVGA resolution (320x240).
The VB.net source code for the form layouts is only for the WQVGA version.

The webpage also provides all the documentation we have kept throughout our two semesters of work,
including our original proposal, the Design Day slides and posters for both semesters, as well as the final
reports for each semester. We also provide our contact information as well as anything else we would
chose to publish.

4.2 Discussion
The scope of this project was well defined, and the tasks were completed to a satisfactory finish. With a
prototype product that can successfully test the patient on any one of three tests (vowel sounds, high
frequency and low frequency), measure the ambient noise, and record the results, we feel we have
developed the functionality to the point where a clinician can easily customize the program to fit his or
her interests and needs. There are still lots of tasks that need to be accomplished before this software
can be said to be completed, but these tasks will be left to future senior design students. To emphasize,
the true value of engineering is not its complexity but its practical use. We hope that the final software
will be used by clinicians who work closely with hearing impaired patients on a regular basis. We hope to
have designed the beginning of a product of clinical value to the hearing impaired and will continue to
be developed in the near future.


5 Acknowledgements
We would like to thank UTD professors Dr. Murat Torlak, Dr. Philipos Loizou, and Dr. Linda Thibodeau as
all as each of their respective TA’s and RA’s.

This project was sponsored by the National Science Foundation (NSF).




                                              Page 13 of 14
6 References
[1] P. C. Loizou, Speech Enhancement: Theory and Practice (Signal Processing and Communications), 1st
ed. CRC, June 2007.

[2] L. M. Thibodeau, The Advanced Hearing Research Center, Callier Center for Communication
Disorders, The University of Texas at Dallas.

[3] G. Schwabb, et. al., “.Net Compact Framework: Creating a P/Invoke Library,” MSDN, January 2004,
[Online], http://msdn.microsoft.com/en-us/library/aa446550.aspx

[4] J. Box, et. al., “.Net Compact Framework: An Introduction to P/Invoke and Marshaling on the
Microsoft .NET Compact Framework,” MSDN, March 2003, [Online], http://msdn.microsoft.com/en-
us/library/aa446536.aspx




                                            Page 14 of 14

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:5
posted:9/6/2011
language:English
pages:14