Master Thesis Conclusion & Acknowledgements

		Back to Main Back to Master Thesis Features :: Results :: Conclusion :: Downloads

Conclusion

The work showed that it is possible to perform fast and accurate semi-automatic left ventricle segmentation with the LiveWire algorithm. A correlation of 97% was found when comparing volumes from LiveWire contours and from manually drawn contours. The comparison with the parametric model lies in the range of 84% correlation for both methods. Details about these evaluations can be found in the thesis.

Publication

The most interesting outcomes of the diploma thesis were published at the 26th Workshop of the Austrian Association for Pattern Recognition in Graz (September 2002). The pdf paper and the poster can be found in the Download section. The paper was entitled "The LiveWire Approach for the Segmentation of Left Ventricle Electron-Beam CT data" and mainly dealt with the comparison of LiveWire segmentation with the "Gold standard" i.e. manually drawn contours and with the parametric model.

Acknowledgements

• The work was kindly supported by Univ.-Prof. Dr. Rainer Rienmueller from the Department of Radiology, University Hospital Graz. He helped me in providing medical knowledge and anonymized CT data-sets. Furthermore he manually evaluated and segmented about 30 data sets, which normally can't be taken for granted.
• I'd like to thank my colleague Armin Kanitsar from Vienna Univ. of Technology in this way for providing me with the idea of trying the LiveWire algorithm for image segmentation.