Jurisica, I. & Glasgow, J. (2000). Extending case-based reasoning by discovering and using image features in in-vitro fertilization. In B. Bryant, J. Carroll, E. Damiani, H. Haddad, & D. Oppenheim (Eds.) Proceedings of the 2000 ACM symposium on Applied computing - Volume 1 March 19-21, Villa Olmo, Italy. pp 52--59

@InProceedings{JurisicaGlasgow2000,
  author = 	 {Jurisica, Igor and
                  Glasgow, Janice},
  title = 	 {Extending case-based reasoning by discovering and
                  using image features in in-vitro fertilization},
  booktitle = 	 {Proceedings of the 2000 ACM symposium on Applied
  computing - Volume 1},
  crossref =     {},
  key = 	 {},
  pages = 	 {52--59},
  year = 	 {2000},
  editor = 	 {Bryant, Barrett and 
                  Carroll, Janice and 
                  Damiani, Ernesto and 
                  Haddad, Hisham and 
                  Oppenheim, Dave},
  volume = 	 {1},
  number = 	 {},
  series = 	 {},
  month = 	 {March},
  organization = {ACM},
  publisher = {ACM Press},
  address = {New York, NY, USA},
  note = 	 {},
  annote = 	 {}
}

Author of the summary: Jim Davies, 2005, jim@jimdavies.org

Cite this paper for:

• changes from one visual representation to another for case-based reasoning (CBR)
• hypothesis: one has to deploy image-based knowledge discovery in order to identify many possible features that describe the quality of oocytes and embryos.
• why CBR is good for medical informatics [52]
• morphometry: techniques for the measurement of the size and shape of biological structures.
• SYSTEM: TA3 (a component)

hypothesis: one has to deploy image-based knowledge discovery in order to identify many possible features that describe the quality of oocytes and embryos.[53]

In this work they create visual descriptions from images. In this way it's changing from one visual representation to another.

morphometry: techniques for the measurement of the size and shape of biological structures.

Visual aspects of eggs (oocytes) and embryos can help with the in-vitro fertilization (IVF) process. This work automates morphometry to identify features like cytoplasmic inclusions and granularity, size of the perivitelline space, cell number, cell membrane regularity, fragmentation, compaction, zona pellucide thickness. This extracted information is linked to more traditional symbolic information to better predict pregnancy outcome.

Two main ways to do segmentation (telling what objects are in an image) are region-oriented (clustering areas of same-color) and edge-oriented (looking for abrupt changes in color to demarcate object boundaries). In this work, the KB is used to know which kinds of objects might have fuzzy boundaries.[55]

The limitations of computational vision techniques means that retrieval is either based on expert annotation or simple image characteristics like color, shape, and shade. This work uses deformable models.

deformable models: Available models are recognized and the task is to align and deform the new model to fit it. Somehow, deformable models can also be used or feature extraction. 11 features are found for the eggs and 4 for the embryos in this work. It uses "snakes." TA3 is used for retrieval.[56]

SYSTEM: TA3 (a component)

Summary author's notes:

• The hypothesis is that they don't think purely symbolic knowledge will work, but then make a system that extracts symbolic visual knowledge from images.
• The retrieval seems to happen from these attributes, so what is the point of choosing "to use explicit image representation, preserving all relevant visual information"?