Wavelet based segmentation of Liver Tumors from CT Images

Computer aided liver image analysis is of profound interest in recent years due to their faster, accurate and reliable characteristics. Qualitative methods include contrast stretching, smoothness, brightness, sharpness etc. Quantitative analysis, on the other hand make use of geometrical measurements such as area, perimeter, diagonal length, horizontal and vertical width etc. These methods of analysis have their own shortcomings due to the characteristics of tumors. Usually they are not in geometric shape which makes the analysis much more difficult and complex. In this work wavelet based image segmentation of liver tumor was performed. The computation time for segmentation is significantly reduced by using Wavelet decomposition method. The segmented results were further subjected to quantitative and qualitative analysis to compare the segmentation results obtained from three different wavelets such as Daubechies, Haar and Coiflet wavelets.