P.S. Tonight(July 2nd) Gagino finishes his project so we can finish testing my code tomorrow :) YAY!
First Day at Gama: July 1st 2009
First Day at Gama: July 1st 2009
Breast Cancer:
Today was my first day at Gama they had the University driver come by and pick me up at my apartment early in the morning to attend Lourdes class. Regardless, I got there significantly late. Lourdes then introduced me to the class and I said a few things to them in Portuguese, regarding the projects I've been working on and where I go to school at. After introducing me I sat in the class and listened to the lecture, which was nostalgic since it was a topic I knew about. Afterwards another class came in and I was introduced again before leaving with Lourdes to her office to begin the more official research/administration work.
The first thing on the list was to give me a pdf of a book she wrote regarding "virtual reality force feedback surgery training systems" They are creating a way to simulate how to remove cancerous tumors from breast cancer patients. This is a very important because seeing how aside from skin cancers breast cancer is the most prolific cancers for women and the second largest cancer killer for women(lung cancer being number 1). Furthermore, even mammography isn't good enough for prevention because usually when cancer is discovered it had already been growing aggressively for the past 8-10 years. Because of this removal is for most part the only option since prevention has its limitations when it comes to breast cancer.
The technology they are currently using would allow them to train a surgeons how to do the tumor removal using microsurgery(lumpectomy). Hopefully by increasing surgeon skill they want to remove the need of using the more aggressive mastectomies and solely having lumpectomy. For example:
Currently, to remove high order breast tumors the entire breast must be cut open and either the breast itself will be lost or most of it. This is done to prevent the cancer from spreading as any that is left behind will continue to grow. This technique is both emotionally and physically detrimental to the patient. Add chemotherapy and its side effect such as hair loss, fatigue and the many other problems that come with radiation exposure; it's a very demanding and destructive procedure to a patient.
The process this research is focusing on is lumpectomy. Which simply means removing the lump(i.e cancerous tumor) This can be done using microsurgery:
(The pics come from Lourdes poster presentation)
(red dot is the incision area)
As seen in this picture the breast is digitized and a small incision is made into it in which the surgeon can practice the removal of the cancerous tumor using microsurgery tools that provide force feedback that they would normally find while doing the surgery in real life. The challenges to accomplish this include, creating mathematics that simulate tissue reaction as the surgery is being done(i.e. how does it act when it's cut, sutured, pierced ect.). Making the simulation realtime because currently it is very demanding CPU-wise to simulate this and lastly, and most impressively they want this to work via the internet so at anytime and at any place in the world when a surgeon wants to practice his surgery he can log into it using a computer and microsurgery tools to practice the surgery.
After removing the tumor in practice or for training the surgeon now has a deeper idea of how the surgery will be performed. As mentioned above this can also be done for training. With a surgery as important and vital as cancer removal the last thing you want is surgeon who has no experience in the surgery only book knowledge and cadaver practice. This tool assures that the surgeon has a much more of real view as to how this surgery will go and much more practice than surgeons who have to wait for cadavers with breast cancer.
(surgeon performing lumpectomy surgery)
The precision and detail they are working with is incredible. Lourdes group is even implementing the feedback done in closing the incision and really the whole process from start to finish.
Fuzzy Logic:
The additional reading was in regards to fuzzy logic and fuzzy controllers(this is not what happens when you mix a cat and a motor). I'll spare the details but I'll give you a quick Idea of what its about. From the books I read last night(read 2 have 17 left) fuzzy logic is an idea that breaks away from the 1 and 0 mentality of current electronics.
For Example: Let us say we wanted to have a program that separated young people from older people. Currently we would choose a number. Lets say we choose 35 years old to be our cut off. Computers need this discrete cut-off because they live in a world were its 1 or 0. You are young or not young. No compromise.
(Pictures from "Basic Fussy Mathematics for Fuzzy Control and modeling")
This hard cut off can be shown in the above picture. But what makes it that a person that is 34.9 is still young and a person who is 35.1 is old. They are only marginally above the cut-off. This can be seen in society all the time. One example is how when your 21(in the USA) you can drink but day before that you cannot. If a computer was checking ID's you would have to wait until the precise nanosecond you turn 21. This is odd because there is no magical change that happens when you turn 21, your still as irresponsible(and some are responsible) as you were right before the instance in time.
To resolve this issue a Proff. Named L. A. Zadeh from Berkley proposed something called a fuzzy set. For example lets say we have the young criteria again. In a fuzzy set it would not be discrete instead it would flow like a curve that describes how much of "young" you are.
As you can see from the pic above, a fuzzy set does not have a concrete value of what young is. There is a general consensus that 10 or younger is young. Therefore, it has a perfect value of one. However, as we get near 30 the curve dips and then aggressively approaches zero after 35.
However, Fuzzy logic gets even more complicated. This is because the idea of what young is changes depending on the person. For example if you are 70 your idea of young differs from if you are 20. This is were the logical operators of a fuzzy set exist. But instead of cutting everything that does not overlap like the 1 and 0 for an "and" statement. The approach fuzzy logic uses is taking two fuzzy sets and comparing them(I'll spare the mathematics as to how).
As you can see we have fuzzy sets from two different peoples point of view. This is not possible if we used the traditional 1 and 0 approach.
This allows us to break away from asking questions like:
If 35 then (do this)
and lets us ask questions like
If "young" then (do this)
The book offers this example:
IF Speed is High AND Acceleration is Small THEN Braking should be Modest.
This form of talking/thinking is not what most scientist consider concrete and is a very strange way to think. It allows for vagueness and only a fuzzy picture of what's going on(thus called fuzzy logic). Cool HUH! :) now you're an engineer!
Applying this idea to the world of discrete numbers were computers live in, has its challenges but is very important when you're dealing with problems for humans. This is because in the reality humans live in, most things are fuzzy sets that continually change, such as age and our perception of young.
Now as to how this will help with the prosthetic research I will be doing. Well, if we use past ideas as to how to move prosthetics we have a system that is either moving or not moving. This creates jolts of movement and makes things look very robotic. This means the finer movements of human movement cannot be done because the prosthetic is either moving at full speed or no speed, 1 or 0. For example to move a hand to pick up a cup we would have to ask the computer to process things as follows:
IF Hand touches cup THEN Stop
To have some form of fluidity we would need to break the problem into many small discrete questions such as IF 1m from cup then 50% speed, IF 10 cm from cup 5% speed ect.. Even with this approach it's still jumpy since everything has to be so precise and discrete.
Using Fuzzy we can ask the computer question like:
IF Hand is "near" cup THEN "slow" down
This vagueness creates fluidity and precision much more closely to that of a natural hand.
But now lets get to the main project I'm working at UnB Gama:
Gama is a very new school, currently it is only in its second semester and most of the students are just entering college. My primary job here is to work with a group of first year students and teach them the research process along with developing a prosthetic foot. The goal for this foot is to have near perfect balance and weight of an average foot while still having fluidity and precision. We separated the project out into 3 parts the mechanics, electronics and, signal processing. I asked when the project started and the proff at Gama said "right now". I let them know I wanted to do the foot that night before and because of that the project was decided on and as of this moment it has begun. I will be helping in which parts we will buy, what motors to use, the math needed to process the signals, really every part I'll have a hand in. This is a great opportunity and one I'm very excited about it. Most importantly, I hope I can be a good role model to these student and have them excited about doing research. I have no pictures of Gama since it was my first day and I didn't want to seem like such a tourist. But next time I go I will take pictures.
Little Bearded Guys Project: July 2nd 2009
As I'm waiting for Dr. Geovany to email me the source code of the prosthetic leg . I was given the pleasure to check out the work that LBG was working on down stairs. Groupo de Processamento Digital de Sinais(GPDS or Digital Signals Processing Group in english).
He works on facial recognition and the work is really interesting. At the end of the video there is demo on it as well.
No comments:
Post a Comment