You are on page 1of 4

Final defense script:

1. (trisha) intro

Ulcerative keratitis refers to the inflammation of the cornea, which can affect a dog’s vision. Such eye
problem should be identified early in order to give priority due to its risks, to prevent adverse outcomes.
In this scenario, veterinarian expertise and clinical service is sought after.
In our modern world, it is advantageous to use our capability to access information online. But with the
abundance of such information, it could be difficult for a dog owner to properly acquire the accurate
information they need.
That is why we developed an Android-based application for identifying keratitis eye disease of dogs. We
utilized image processing for the identification process as it is a growing field that has multiple
applications including medical science.
The application helps dog owners to get information regarding the eye disease, with the assurance of
veterinary expertise. The features of the application will be discussed in the progression of our
presentation. Overall, we, the researchers, developed a system that provides convenience, accessibility,
and reliable information for dog owners.

2. (eryca) conceptual framework

We interviewed our locale, Sudah Veterinary Clinic in Magalang, for information regarding image
processing in the area of dog diseases. After the collection, we reviewed the related studies in order to
get a better understanding of the current study. All the necessary resources for developing the
application were prepared. After all the preparation, the system’s design was considered along with its
functions and features.

3. (eryca) statement of the study

We aimed to design and develop an online Android-based application utilizing image processing for dog
ulcerative keratitis eye disease.

Specifically, the researchers aimed to:

1. Develop a system that registers the dog’s owner profile


2. Develop a system that identifies the dog’s ulcerative keratitis eye disease using image processing
3. Develop a system that provides relevant information, tips, and first aid regarding the dog’s eye
disease
4. Develop a system that provides scheduling for video consultation
5. Evaluate the system’s functional suitability and usability
4. (eryca) scope and limitation

A broad scope of all common dog eye diseases is complex, hence, the study is limited to only one eye
disease, which is ulcerative keratitis. The developed system also does not have the capacity to recognize
non-dog eye images, it will classify any image inputted.

There are also determined factors affecting image processing algorithm, such as image resolution,
environment and lighting, contrast, lens distortion, and noise. In the study of Kim et al. (2019), geometric
operations such as rotation and scaling generate artifacts and destroys pixels, lowering the image’s
quality. This was also noted by the veterinary doctor of the study’s locale, Sudah Veterinary Clinic,
Magalang.

5. (trisha) methodology

Now, for our methodology. Our research design is descriptive. We collect data or the images, and then
we analyze and describe the phenomenon, or in this case, the eye disease keratitis, through image
processing.
Again, our locale is the Sudah Veterinary Clinic in Magalang. And we had 28 Sudah Veterinary Clinic cdog
owner clients as our respondents, as well as the main veterinary doctor as the admin, and 3 IT experts.
Which, there are 32 respondents in total that evaluated the system.
The data collection methods for the study were in the form of survey questionnaires, interview, and
Internet research. Three sets of questionnaires were prepared for: the user, admin, and IT expert. An
interview with the Sudah Veterinary Clinic veterinary doctor was also conducted in order to grasp the
existence of the problem. Internet research was consistently done before and during the development of
the system.
Purposive sampling was the main sampling strategy used in selecting the respondents. As the project
involved dogs, it was apparent that dog owners are the respondents. Convenience sampling was done
when gathering the data, only asking dog owners who were coincidently visiting the veterinary clinic and
who were willing to be respondents. Other known nearby dog owners were asked to be respondents
even if they were not currently visiting the said clinic but do sometimes.
(Statistical Treatment of Data)
Using the Slovin’s formula, n=N/(1+〖Ne〗^2), N is the population which is the number of clinic visitors per
week, and e is the margin of error of 5%; n is the number of respondents. There are 30 dog owners who
frequent the clinic for eye-related inquiries on a week on average. So, the number of dog owner
respondents were 27.907 ≈ 28. Statistical analysis was conducted to analyze the data gathered. Means
for the response ratings were acquired by averaging, where the total rating is added together then
divided into the number of responses. A scale for data interpretation was then referenced to assign to
the overall means.
We utilized an Agile approach in developing the proposed system with the following phases:
 For the (Exploration phase): We established our objectives, scope, and then the features of our
proposed system.
 For the (Planning phase): It includes the user interface mockup design and the system
architecture.
 For the (Iteration phase): It is the development phase, which will consist of reviews and revisions
until completion.
 Then lastly, the (Testing phase): It ensures that the developed system will function properly as
indicated. And so, it can also include user feedback.

6. (trisha) results and discussion

Now for the data interpretation and discussion.

Table 3 shows the 28 dog owners’ evaluation of the developed application. Having an overall mean of
3.83, which is interpreted as “strongly agree”, means that the dog owners found the application easy to
use and helpful.

Table 4 shows the veterinary doctor’s evaluation of the developed application, which he serves as its
admin. Having the mean of 3.42, the veterinary doctor agrees and acknowledges the application being a
useful tool in his occupation.

Table 5 shows the IT experts’ evaluation of the developed application. Their questionnaire is divided into
two categories: functional suitability and usability. Having the overall means of 4 and 3.77, all 3 IT
experts strongly agree and recognize that the application is aligned with its objectives, being functional
and usable.

The developed application utilized CNN-based models that were trained at Google colab, exported into
TensowFlow Lite. Having 90% accuracy levels for training and testing, and validation achieved an 89%
accuracy percentage. This is lower than the ideal accuracy percentage which is 90%, yet it is still a
useable image processing algorithm.

Upon research, there were factors identified that could be affecting the accuracy levels, which are image
resolution, environment and lighting, contrast, lens distortion, and noise. These were in alignment with
the findings of the study of Kim et al. (2019), that low image resolution due to geometric operations such
as rotating and scaling, lowered some of their models consequently. The aforementioned factors were
noted by the veterinary doctor the study’s local Sudah Veterinary Doctor.

7. (kenneth) conclusion

The application underwent evaluation by 28 dog owners from Sudah Veterinary Clinic, one veterinary
doctor, and three IT experts. Dog owners gave it a mean score of 3.83, indicating a strong agreement that
the application is easy to use and beneficial. The veterinary doctor scored it 3.42, expressing agreement
on its usefulness for clinic operations. IT experts gave a mean score of 3.77, strongly agreeing that the
system demonstrates functional suitability and usability. Overall, the evaluations suggest positive
feedback on the application's ease of use, benefits for clinic operations, and validation of its design and
functionality.

8. (kenneth) recommendation

The researchers recommend that future researchers improve the features and functionalities of the
developed application. Specifically:

1. Add more eye diseases that can be identified by the system in order to maximize its capabilities

2. Create a solution to the limitation so that the system should be able to recognize non-dog eye images.

3. Figure out a way to resolve the enumerated factors affecting the image processing algorithm.

You might also like