Abstract
This thesis is about classification of patients that had been treated for ovarian cancer at Det Norske Radiumhospital.
The classification is based on microscopy images taken right after the treatment.
After a period some patients had a relapse of their illness.
The general question is wether it was possible to foresee the prognosis right after the treatment by analysing the images.
The thesis is limited to applying series of image preprocessing
before the classification to see how that affects the classification.
The lowest classification error achieved with a single texture feature was 0.150.
The thesis is supplied with a CD-rom with lot of tables and diagrams made during the process.