Major Project Outline
Project Name: Entropy-based metrics for joint image alignment or Enantiomorph for short
Supervisor: Neil MacParthalain
Kick-off date: 25th January 2016
Research Question: Can entropy-based image alignment metrics offer improved image aggregation of tissue density for mammographic risk assessment?
Project description: "The joint-alignment of images is an important area of computer vision with application to many different areas such as medical imaging, handwritten text recognition and classification etc." Congealing will be used to align a set of images (initially handwriting samples, hopefully eventually mammography scans). Users will be able to chose an alignment metric from a simple GUI to allow for easy comparisons between the outputs given by the different metric types.
Outcome of the project: To be able to output an average image after using different alignment metrics.
Stretch goal: Predict upon new mammography images the likelihood of Breast Cancer - given the 4 BI-RADS composition categories.
Extra reading: