Image Segmentation Techniques in Nuclear Medicine Imaging
Abdel‐Ouahab Boudraa, Habib Zaidi · Kluwer Academic Publishers eBooks · 2006
It is gratifying to see in overview the progress that image segmentation has made in the last ten years, from operator-dependent manual delineation of structures, through simple thresholding, the use of classifiers and fuzzy clustering, and more recently atlas-guided approaches incorporating prior information. Recent developments have been enormous particularly in the last ten years, the main opportunities striving towards improving the accuracy, precision, and computational speed through efficient implementation in conjunction with decreasing the amount of operator interaction. The application of medical image segmentation is well established in research environments and is still limited in clinical settings to institutions with advanced physics and extensive computing support. As the above mentioned challenges are met, and experience is gained, implementation of validated techniques in commercial software packages will be useful to attract the interest of the clinical community and increase the popularity of these tools. It is expected that with the availability of computing power in the near future, more complex and ambitious computer intensive segmentation algorithms will become clinically feasible.