Skip to ContentSkip to Navigation
About usNews and EventsEvents and open daysPhD ceremonies

PhD ceremony Mr. F.N. Kiwanuka: Exploring morphological attribute filters in medical image enhancement

When:Fr 20-09-2013 at 11:00

PhD ceremony: Mr. F.N. Kiwanuka, 11.00 uur, Academiegebouw, Broerstraat 5, Groningen

Dissertation: Exploring morphological attribute filters in medical image enhancement

Promotor(s): prof. N. Petkov

Faculty: Mathematics and Natural Sciences

The thesis presents a number of extensions to attribute filtering, in particular the two shape descriptors, techniques to automatically select threshold parameters, and attribute cluster filtering for 3D morphological attribute filtering.

Shape description plays an important role in generating feature vectors from a given shape upon which features of interest can be selected irrespective of their size. The goal of description is to uniquely characterize the shape using its shape descriptor vector. The required properties of a shape description scheme are invariance to translation, scale, and rotation.

The work in the thesis describes both the theoretical mathematical formulation and practical considerations where for each shape descriptor, an algorithm and implementation using the Max Tree is given. The thesis also describes how threshold parameters can be set automatically. Computing the threshold automatically not only increases reliability, is time saving and relatively more accurate in selecting features of interest from the image volume but also guarantees reproducibility. The thesis also describes attribute cluster filters for 3D enhancement that replaces the single attribute with an attribute vector, which is a feature vector describing each connected component. This allows a better discrimination of different classes of objects.

Experiments that demonstrate the properties, capabilities and limitations of the shape descriptors, automatic threshold selection and attribute cluster filters are provided in the thesis. Of particular consideration is performance evaluation of the methods in relation to existing methods.

printView this page in: Nederlands