SpotNet - Learned iterations for cell detection in image-based immunoassays

Accurate cell detection and counting in the image-based ELISpot and FluoroSpot immunoassays is a challenging task. Methodology recently proposed by our group matches human accuracy by leveraging knowledge of the underlying physical process of these assays and using state-of-the-art iterative techniques to solve an inverse problem. Nonetheless, thousands of computationally expensive iterations are often needed to reach a near-optimal solution. In this paper, we exploit the structure of the iterations to design a parameterized computation graph, SpotNet, that learns the characteristic patterns embedded within several training images and their respective cell secretion information. Further, we compare SpotNet to a customized convolutional neural network layout for cell detection based on recent advances. We show empirical evidence that, while both designs obtain a detection performance far beyond that of a human expert, SpotNet is substantially easier to train and obtains better estimates of cell secretion.