Biophotonic imaging can be used to characterize tumor growth in animal models. Estimation of the numbers and location of target cells is dependent on accurate segmentation of a target region from the background luminescence of the animal body. Existing software systems extract general regions of interest in complex images but can fail to detect details or faint regions of interest in the presence of higher luminescent activity. Limited work has been published in the analysis of these images. We explore the use of popular medical image segmentation techniques in segmentation of bioluminescence images. The ability of each algorithm in detecting regions of interest is tested using established performance measures. Based on the characteristics of bioluminescence images of animal tumor models, and the performance of the algorithms, we formulate a framework for segmentation techniques best suited for this application.