LG223: Exploring visual representations on histopathological images

Date: 
Friday, March 27, 2015 - 12:00
Speaker: 
David Romo-Bucheli, MS
Abstract: 
Automatic analysis of histopathological images is a potential source of knowledge and an important support tool towards quantitative diagnoses in pathology. This field has opened up the possibility of analyzing not only the relationships between cells, but also complex spatial arrangements of cell groups that may be related to a particular disease pattern. Although many image processing techniques determine quantitative relationships among tissue parts, most of them lack flexibility and interpretability. An ideal representation of histopathological information would preserve features that are relevant for medical diagnosis across multiple scales and facilitate the discovery of unrecognized cues or patterns useful in the differential diagnosis workflow.