|Topic:||Multimodal digital pathology|
|Supervisor:||Prof. Ing. Jiří Matas Dr.|
|Description:||(The work will be supervised with Dr. Vlad Popovici, email@example.com)
The current methods in digital pathology try to model the human expert approach
for understanding the histopathology images and to propose a number of quantitative descriptors of the images. However, this approach provides a limited view on the underlying biology and will likely lag behind the human expertise in interpreting the histopathology image data for the foreseeable future.
On the other hand, nowadays a biological sample is characterized by a richer
set of features raging from clinical to molecular information. It is not uncommon to have whole-genome expression data, mutational and clinical data available for analysis.
In this context, the present project aims at combining gene expression, clinical
and imaging features to provide a more comprehensive description of the pathology slides, description that will rather complement than replace the usual pathologist assessment. This represents a paradigm shift in digital pathology and is expected to advance the current state of the art. Using gene expression data to guide the development, the project will create and implement novel software tools for histopathology image analysis. The utility of the resulting methods will be evaluated in collaboration with an expert pathologist.
|Instruction:||Requirements: familiarity with image processing; Java; pattern recognition