You are on page 1of 17

Texas A&M University at Qatar

ECEN 403: Electrical Design Lab 1

Customer Needs Analysis Report


Personalized Quantification of Facial Normality using AI

Team 7
Khalid Al-Emadi
Salma Aboelmagd
Layan Al-Huneidi
Sara Mohamed

Mentor
Dr. Erchin Serpedin

Due 24th of October, 2021

“An Aggie does not lie, cheat, or steal, or tolerate those who do.”
1

ECEN 403: Electrical Design Lab 1 0

Customer Needs Analysis Report 0

Introduction 2

Methods 2

Survey Analysis 3

Expert Interview 10

Evaluation 14

Conclusion 15

Appendix 16
2

Introduction

A facial generation application was created to assist reconstructive surgeons in generating

an image of a realistic image of a patient post a facial reconstructive surgery. Embedded within

this program is a scale for the personalised quantification of facial normality. It uses a scale that

measures the degree of abnormality, making it useful to quantize the improvement pre-surgery

and post surgery. This particular feature can aid a spectrum of people: reconstructive surgeons,

medical communities, patients, patients’ families, as well as researchers. Seeing as this spectrum

deals with a specific targeted group of people, it is of the utmost importance to adjust this scaling

feature to our target audiences’ needs. Therefore, a customer needs analysis was to be conducted

and this analysis was done through surveying people with a variety of questions to better guide

the team towards the market’s needs better. The aim of this report is to analyse these responses

and explain our methods used for conducting the survey.

Methods

For our feedback, a survey was conducted and sent out, as well as an interview with a

professional. As our project was aimed at the medical industry primarily, we conducted our

interview with a reconstructive surgeon then sent our survey out to the general public to get a

better understanding of what a general audience would expect to see. The specific survey

questions are included in the appendix.


3

Survey Analysis

This question was asked to see what category the people who do the survey are in. The

overwhelming majority of the people who did the survey are students. That is because the

method used to share the survey link around was through “WhatsApp” groups that consisted

mostly of students. This method of spreading the survey proved to be not effective, as the goal

was to get responses from people in as many backgrounds as possible. A solution to this problem

is to setup stands in public spaces, offering people something in return as an incentive for them

to do the survey. Due to time constraints and complications for getting permits to setup a stand,

this was not possible.


4

An overwhelming majority of the participants agree that there is a need for the project.

The lack of diversity in the professions may have shifted the results towards “yes” more than

“no”. Those who have selected “no” may not have any children, so they do not see a need in the

project.
5

A great majority of the participants believe the program would be very effective in rating

the abnormality of the facial images. To reach the expectations of the participants, a large

number of facial images will be acquired. This is important because the larger the dataset, the

more accurate the program is.

To avoid any copyright and privacy issues, we will be using a website that generates

images of people that do not exists from a dataset of real people. More than three quarters of the

participants believe that having those images as a source for our code would not negatively affect

the accuracy of our program. Since the website we are getting our images from has a dataset of

thousands of images, we believe that the website produces very accurate and convincing images

of faces. It is one of the best sources for us as the issue of copyright and privacy would slow us

down and only get us a few photos to use.


6

Most of the participants believe that it is difficult to recognize the degree of abnormality

in a face. While it is true that some “abnormal” faces have very minor abnormalities that even

they don’t notice, there are degrees of abnormality that are obvious. This response shows that

there may be small deviations between the rating of the program and the human ratings of

abnormality.
7

Please elaborate:

· It does not raise any ethical concerns

· its a consensual decision

· At this point in time, I’m not sure what to expect

· If it’s used for it’s specified purpose, I don’t think it poses any ethical concerns

· Wouldn’t people then be able to choose the ethnicity they want to look like?

· There is an ethical dilemma present when we are creating a standard of beauty based

on numbers. If said standard is put in place who’s to tell what the social ramifications are.

· It can address any faisal abnormality

· the only thing i’d be concerned about is when we say abnormal, do we mean a

dysfunctional feature or just not within the beauty standard?

· On a large scale, the merits would definitely outweigh as people are more inclined

towards a better face

· Can use that facial screening to get someone’s identity. Make videos as if they are

talking for example

· I think that it might result with identity theft

· humans have unrealistic expectations of beauty that most of the time don't exist in real

life or naturally. this app seems it will worsen this issue instead of helping people

overcome unrealistic beauty standards


8

· It might be offensive to the people with facial abnormalities not wanting to do facial

reconstruction in terms of them catching on the idea that they need to look a certain way

or have specific things on their faces in order for people to understand them or

communicate with them or be accepted in society.

· It doesn't create any concerns.

· Many people might use it even if they don’t need it to change there look completely.

Also is it ethical to use tea people to construct it is another concern

· I don’t think it’s ethically concerning

· I don’t think it would cause any ethical concerns, but it could lead to people having

more and more things they see as flaws about themselves.

· Yeas it would raise ethical as it’s helpful for the future

· It does not provide any ethical concerns

· I cannot see why there will be any ethical issues.

· If there is no consent given from the person whose face is being scanned, it can cause

many legal issues

· its kind of like shaping a baby, and that could lead to doctors playing God on how the

baby should look like

· No it will help in the implementation of the project

· I would wonder if what the program identifies as abnormal and normal would

influence beauty standards

· It’s fine so as long as the data is not sold or given to someone else, only the hospital

has the right to this


9

· Software are prone to making errors to some degree

· Depends on how it's implemented but maybe in the worst case scenario the program

might flag some uncommon but normal features as abnormal and avoid these features

entirely. Maybe it will converge to one 'perfect' face

· It will be beneficial for people to feel comfortable in their own skin again

· The program is excellent for recommendations during surgeries. However, it does

require human interference to confirm facial predictions.

· Yes, because we might not see the objects clearly and the quality can be judged

negatively.

· There is a very broad spectrum of faces around the world it would be extremely hard

to distinguish the “ normal “ features, cause in reality there is no normal face structure,

normal is very broad and hard to achieve.

· It all depends on the training of whatever Machine learning method you use. There are

differences between the way people look and ethnicity/race plays a part in that, however

minor.

Here, the participants are evenly split between believing that the program is ethical and that it is

not ethical. Some of the people that chose “no” stated religious reasons on why they think it is

unethical. Others stated reasons that would reflect on social issues. For example, there are some

responses that were concerned about the beauty standards that society withholds. They were not

comfortable with the word “abnormal” being used, so rephrasing might be helpful for the future.

It seems that the participants were not aware that our focus is on medical anomalies and not
10

“beauty flaws”. Also, there were privacy concerns. This is not an issue because privacy is on the

top of our priority, so we will make it clear the next time a survey is sent out.

The majority of the participants would recommend the program to people looking for facial

reconstructive surgery. The difference between this pie chart and the last is huge. Even though

50% of participants believed that the program would be unethical, most of them would still

recommend the program. This shows that the information about our goals and the program must

be shown in a clearer way.

Expert Interview:

The list of questions below were asked during an interview with an expert. The answers

were roughly transcribed from the audio.


11

How often do you get patients that are in need of facial reconstruction surgery?

…The plastic and craniofacial surgery department treats kids with craniofacial deformities and

that's what we do every day. So in a typical clinic day, we'll see anywhere between 15 and 25 or

more patients. Over those, more than half, maybe 2/3 are facial reconstructive problems…

Which is the more common cause for the need for official reconstruction surgery? Birth

defects or accidents?

At this hospital it's definitely birth defects. We don't see tremendous amount of trauma and in the

pediatric population the trauma is mostly lacerations; occasionally fractures, nasal fractures but

by large I'd say 90% of what we're doing is congenital.

What is the procedure done before the surgery? (How do you plan out what the patient

needs and how the reconstruction happens)

… In terms of measuring the face, there is no great tool to measure the face. So, how do we

prepare? We prepare by basically asking the child and the parents: what bothers you and why did

they come in. They usually know when they came in they came in because there's something

about the face or the skull that they know isn't right, and either the child if they're old enough is

aware and maybe is being picked on at school let's say or if it's a baby his evidence is if a child is

born with a cleft lip you know a lot of these deformities are quite obvious... Basically the short
12

answer of what do we do in preparation: it's communication, it's understanding what the problem

is and explaining what we can do and what we can't do.

Do you use photo references before and/or while doing the surgery?

In our clinic here we have a photography suite. I have a dedicated photographer. We photograph

almost every patient certainly for facial reconstruction… We do it for a few reasons… to help us

plan so I may not operate on a patient who I see today for two or three months or four months or

longer. The day before the surgery or the week before the surgery or several weeks before you

know depending upon the complexity of the case, I'll go back, and I'll look at those images and

oftentimes I'll zoom magnify into those faces and really analyze what I'm seeing to prepare

myself mentally for the next day or for the upcoming case so it's helpful in planning...

Do you expect the application to make the pre-surgery planning process easier?

I'm not sure that it will make it easier necessarily. I think it'll make it maybe more valid. What I

mean by that is if you can have data that shows sort of a normal curve, a range of normality, and

you can place a patient on that normal curve somewhere, I think it helps to give them some

context as to where they sit. In some places also there's a real struggle to get payment for

procedures... insurance payers want some justification as to why we're doing the procedure. It's

very clear if a child or a patient has something like a cancer or something that can be easily

demonstrated. Insurance is not going to argue about payment but when you're talking about a
13

facial deformity, again, there's no objective measures. So, the application that we're working on

can help to provide some context…

Do the patients feel anxious about what they would look like after the surgery?

I'm sure they do… the children don't say what they may be thinking about, it is certainly the

older children, I'm sure. The younger children probably are not thinking about it. Again, a large

part of our practice are babies three months, four months, five months with cleft lip… we do a lot

of skull reconstructions of babies who are anywhere between three and 12 months of age so

they're clearly not thinking about what they're going to look like after. I'm sure their parents are.

Parents being parents tend to focus much more on safety issues and concern about risk. They just

want their child to be safe and to come to the surgery comfortably. I don't think they focus quite

as much on the static outcome. Of course, they do care about that but I think for certainly, the

younger kids focus more on safety and prevention of risk. The older kids: we treat kids who are

in middle school and high school. Teenagers, of course, they're thinking about that. It is their

motivation: the aesthetic outcome. I'm sure they worry about it. I try to tease that out but we also

don't want to overly focus on the negative. We try to focus a little bit more on the positive in our

conversations with the patients, and we always do stress realistic expectations. This is something

that is drummed into your head when you're in training as a plastic surgeon: is to always make

sure that you emphasize that patients and their parents have realistic expectations that they

understand what we can provide…


14

Do you believe that the application would help the patients feel more confident about the

reconstruction surgery?

I guess it depends on what the delta is, right? I mean, so, if we start at you know on a scale of 1

to 10 if we started at a four and we end up at a 4.1, I presume they're not gonna be very happy. If

we started with a 4 and we finished it at an eight, I think that would be something that they

would be happy about…

Evaluation

Through the conduction of these surveys and the interview with Dr. Mitchell, we were

able to reach out to some key considerations in developing and implementing our project. Most

notably, the design and the implementation of this model is found to be crucial, due to the

increasing demand for Facial construction surgeries. For that, we are eagerly anticipating

working on this project and being able to provide a computer model that will intend to measure

the level of facial normality of any facial image in order to deliver an abnormality score pre and

post-reconstruction. Thus, helping the surgeons evaluate their progress by comparing the

outcome deformity degree to the initial one. It will stand out as a scale on which improvements

will be measured upon. Dr.Mitchell acquainted us with the fact that the computer model needs

around 70,000 images database for it to handle faces of all different ethnicities extremely well.

For that, we will be focusing during the upcoming period on training our algorithm to recognize

abnormalities. Moreover, we will be constructing a large and labeled database of images to be

able to detect the degree of abnormality precisely. To do so, we will be using a website that

generates images of people that do not exist to generate a final ethical model of high diversity.
15

Lastly, we will implement new approaches to reach out to a wider range of audiences. Since the

methods we attempted to spread the survey were not very effective.

Conclusion

The interview along with the surveys conducted enhanced our understanding of the needs

and goals and provided us with the needed data that will be used to carry on this project. The

intention of collecting these data was to attain a comprehensive understanding of the target

market and their demands. Our goal is to definitely have the best model possible that will serve

the doctors, patients ,and their families as they will get the opportunity to track the facial

normalization process.
16

Appendix

Survey Questions:

1. Please state your profession:


2. Do you believe there is a need for such a project?
3. On a scale from 1 to 10, how effective do you believe this program would be?
4. The source of our images will come from a machine learning algorithm that uses real
photos of people to create an image of a person that does not exist such as the one shown
below. Do you think this would negatively affect the accuracy of our program?
5. On a scale from 1 to 10, how challenging do you believe it is to recognize the degree of
abnormality?
6. Does this application raise any ethical concerns for you?
7. Please elaborate:
8. Would you consider using/recommending such a program for someone looking for facial
reconstructive surgery?

Expert Interview:

1. How often do you get patients that are in need of facial reconstruction surgery?
2. Which is the more common cause for the need for official reconstruction surgery? Birth
defects or accidents?
3. What is the procedure done before the surgery? (How do you plan out what the patient
needs and how the reconstruction happens)
4. Do you use photo references before and/or while doing the surgery?
5. Do you expect the application to make the pre-surgery planning process easier?
6. Do the patients feel anxious about what they would look like after the surgery?
7. Do you believe that the application would help the patients feel more confident about the
reconstruction surgery?

You might also like