Journal of my ten weeks at BU

Originally, I had a journal of all my ten weeks at BU. However, while I was making my final report I accidently overwrote my journal. So now I don't have any record of my journal, which means I will have to give you a overview from what I remember and then focus on a report of my last week. You can read an overview of my summer in my final report. I can't believe I did that!

Week 1

This week Katelyn and I mainly tried to get stuff done to work at BU. We had to get ID cards, set up what seemed like half a billion accounts and also get caught up on the current status of the project. Prof. Betke wasn't in town, so her graduate students showed us around.

Week 2

This week I continued to read up on past papers that were written. I also looked at some of the code that was done. Prof Betke came back from her trip so Harry, Bill, katelyn, she and I met to discuss what had to be done. Harry and Bill had papers that had to be written for a conference so we had to run tests to generate results for them.

Week 3

This week involved more testing and writing of short programs to do that. Harry and I worked on optimizing his nearest neighbor search code. He was using a 1D array to represent a 3D array and was converting it when he needed to use it. So we have to fix it so it would use a 3D array. This really made things go faster.

Week 4

We had to put the finishing touches to Harry's paper. We had to read and reread it to check for errors and also made suggestions to Prof Betke on how to improve it. The paper was finally handed in on Friday. It was a long day but it paid off in the end.

Week 5

Now that the paper is done, Harry and Prof Betke need to start working on his thesis which was also a submission to a journal. They got it back with many comments so it was now our task to include the recommended changes. I started out by implementing a search method that finds the location of the sternum and spine based on a template matching code that Prof. Betke had written. I first had to crop the image out of a scan and then use that as a template to find the locations of the two images in a given slice. I came in to work on Saturday as well because of the 4th of July. This was already implemented in the first version of the paper but they were using gre y scale units rather than Hounsfield Units. Hence, the accuracy isn't always guaranteed.

Week 6

I had to test my code and also run it on many different cases. We achieved a detection rate of 100% which was encouraging because the location of the sternum and spine weren't very accurate in the first paper and so this was a major concern for the reviewers. I also ran other tests for many different versions of the data. I implemented a bounded box method that takes in a set of points and finds 8 coordinates that if connected would completely cover the box.

Week 7

We still have a lot of work to do before the paper will be ready for submission. We worked on more methods to use a comparison. I tried to help Harry implement a weighted intensity method to find the intensity of a given point. It takes in a point as input and then assingns it an intensity value based on its neighbors and the distance of those neighbors from the point concerned.

Week 8

Prof Betke wanted me to implement the 3D chamfer matching method. She came up with an algorithm to use. The methods we found are written for 2D spaces so we had to extend it to include 3D. She explained it to me so it was much easier to implement. I then had to run it on the results that Harry had found using his method and see how they compared. The chamfer matching method is very fast and so it wins but isn't as accurate as ICP.

Week 9

Prof. Betke was working on writing the paper. So I had to read over it and make suggestions. It is a long paper so it takes quite a long time to read. So I had to take breaks in between. katelyn and I took our lunch outside and enjoyed the outdoors while we read. I am also thinking of working on the chamfer matching code some more and see if we could use it to find the optimal rotation and transformation without starting with the result obtained from ICP. This project is very interesting, but since we don't have much time, I might have to work on it once I get to school.

Week 10

This week I mainly was trying to tie up the loose ends from the summer. I had to make sure that all the tests I ran, code I wrote and any research I did was saved in a place where others could get to it. I had to print out papers that would be useful for school. Prof Betke and I are talking about doing a remote project based on the Chamfer method. Since its making such good progress both of us would like to continue working on it. I might even be able to make it count for an internship and get credit for it at Bethel. So thats an interesting prospect to look forward to. Even though I was officially done on Tuesday I am still coming in to finish up stuff, make copies of papers I may need and also kill times since I don't have anything to do any home anyways.