Article Structure and remarks

[massich]

This issue is intended to track and discuss the article's structure and remarks. Anything that is here should be present in the final article, and everytime some one modifies something should make sure that either the relevant information remains on the article or should be reported back here to place it where needed.

Things that should be on the article

• [x] OCT
  • [x] Why is important
  • [x] Why classification
  • [ ] image modality problems
  • [x] liu, OCT images have a drift (misalignment) while acquisition therefore either is treated as single slices or the volume needs to be aligned.
  • [x] liu, coexistence of multiple pathologies
  • [x] liu, variability in shape size and magnitude within the same pathology
  • [x] liu, retina reflectivity (schuman et al., 2004)
  • [x] liu, inconsistent image quality (Barnum et al., 2008)
  • [x] Venhuizen, evaluation of the volumentric scan is time consuming, expensive and the AMD signs are easy to miss
• [x] Clarify that we want to do a fair comparison
  • [x] We bring a library (or framework) protoclass
  • [x] We bring our implementation:
  • [x] Liu
  • [x] Venhimzen
  • [x] GMM
  • [x] ours
  • [x] Srinivasan
  • [x] How we validate the implementation of other people's methods
• [x] Dataset
  • [x] our dataset
  • [x] why Duke (45) and Duke (384) cannot be used

    Outline

• Intro
• Materials and Methods
  • list the articles
  • our implementations
  • our framework
  • data
• Results
• Discussion

[massich]

I don't find my notes on the reasons why the public datasets were unsuited for us, do you remember the arguments we discussed? @mrastgoo

[mrastgoo]

Duke(45) - AMD - DME - the data are cropped, denoised and resized Duke(384) - AMD and normal (you said it, I dont remember )

[massich]

That's exactly it. The think is that the large duke has no DME therefore we cannot use it. While the short Duke would still suit us, it has the problem that the data has been manipulated. This is the argument that needs to be pitch in issue #8

[massich]

I'll start referring to the lack of data. Do we have a better reference than Maryellen L Giger et al. “Anniversary paper: History and status of CAD and quantitative image analysis: the role of Medical Physics and AAPM”. . In: Medical physics 35.12 (2008), p. 5799 @I2Cvb/not-that-bad