Iteration 1 MVP

Iteration goals:

In this iteration we will make most of our research for both sides of the project. The first is numbering the rib's bones, given segmented images. Here we will try to use Matlab. For this feature, there will be article reading, finding ways how to do it. Second, examining networks to learn how rib's bones look like, and finding it in new scans. It might be CNN, using TensorFlow, or other learning algorithms. In this iteration we will decide what algorithm will it be, and how it will be generally implemented. Please review the important links found downstairs and see all the work we've done.

Lists of goals:

• Prototype report - an improvement of the proposal from ZFR.
• Deciding learning algorithm.
• Reviewing different networks.

use cases:

• Networks
• Inputs

Retro prespective - goals and results:

• We wanted to move forward on networks and we got to know 3 kinds of them.
• We tried to make manipulations on images, and we got some toolkits for it.
• We wanted to know more about image processing. Hopefully we will learn more later.
• Project proposal document was upgraded to Prototype Report. To be submitted on project's site and on repo.

Roles:

Team: Alon and stav
Generally:

• Alon - Network work, DeepMedic and article reading
• Stav - Numbering images matrix, Casacded-FCN, and article reading

Issue Management:

Issue Board

Iteration 1 Issues

Tests:

So far we checked ourselves in 2 ways:

• Compiling the code with the tests given by DeepMedic.
• Checking wanted results for our images.
  Everything went well so far.

Points Summary:

• DeepMedic - 8 points
• Cascaded-FCN - 8 points
• Faster-RNNLM - 5 points
• Prototype code - 8 points
• Prototype refactoring - 8 points
• Learning algorithm - 5 points
• Code tests - 5 points
• Prototype video - 2 points
• Article reading - Prototype stage - 2 points
• Prototype update meeting - 1 points
• Prototype report - 1 points
  Total: 53 points

MVP - Minimum Viable Product

This stage gives the main framework of out research. It deals with the main framework of the project: Learning algorithm and Networks. It deals with the main way of implementing what we want, and initial code to be shown. This means showing DeepMedic with our training sets and our data to be checked.

Important Links

• Working Document - All information gathered through this iteration.
• Prototype presentation - working document
• Research Blog

Iteration Summary:

• As a summary, We have worked a lot on this iteration. Since it's the first one, and the project still exists as an idea, and we had to read a lot and to find good information about what we want to do. In the end we were guided by our academic advisor what to read and what to use. Plus, we made one appointment with our academic advisor through Skype, and it was ok. We are making tests for our deep medic code, If image is being loaded correctly, if images are saved correctly. In the future we will make more tests as needed.
• Presentation video

To improve:

• We have worked a lot together, even when there were things we could do without meeting each other. So next time we will make appointments and work on relevant things only.
• Try to focus on what we need to do, and to point to academic advisor problems on time without waiting.
• When we get a mission, we have to know all the background information.

To sum it up:

• The project is on it's first steps, and we are pretty happy with what we see.
• The subject is very interesting.
• Academic advisor is helpful.
• Going forward and higher!

Next Iteration:

Iteration 2 DMC