Identification of Nano-particle Absorption by Circulating Macrophage

Detection of cells within a photoacoustic image of a slab of tissue is an inherently difficult problem. Not only are the cells themselves shape and intensity variant, but the background is bright as well. Taking a 3-D slab and reflecting it into a 2-D plane causes different orientations of the same cellular structure to emerge. This task is currently accomplished by humans, though the process is laborious and error-prone. In this paper, we describe our approach and results in trying to solve this problem using various machine learning techniques. We chose features that reflect that cells are ovular, bright, and will have some contrast with the background. We compare results obtained with different algorithms and different combinations of features, concluding that more work is necessary to obtain acceptable results.