A comparison of three image segmentation techniques for PET target volume delineation

Abstract

Incorporation of positron emission tomography (PET) data into radiotherapy planning is currently under investigation in numerous sites including lung, brain, head and neck, breast and prostate. Accurate tumor volume quantification is essential to the proper utilization of the unique information provided by PET. Unfortunately, the subject of target delineation within PET currently remains a largely unaddressed problem. Identification of the precise geometric boundary separating tumor from its surrounding background can, in many cases, prove vexing. The current difficulties associated with accurate geometric interpretation of PET data pose fundamental limitations to the full utilization of this powerful imaging modality for planning more accurate and effective radiotherapy treatments. The need to improve this situation is immediate if the full potential of this powerful imaging modality is to be properly assessed and utilized. To this end three segmentation methods are examined with regard to their ability to yield accurate delineation of PET target cross sections. These are thresholding, Sobel edge detection, and the watershed approach. A phantom study employing well defined volumes and activity distributions provides an opportunity to assess the efficacy of these three segmentation approaches with regard to their relative abilities to yield accurate target delineation in PET. Results reveal threshold based segmentation as the superior method of the three delineation techniques investigated.