Structured correlation detection with application to colocalization analysis in dual-channel fluorescence microscopic imaging

Current work flows for colocalization analysis in fluorescence microscopic imaging introduce significant bias in terms of the user's choice of region of interest (ROI). In this work, we introduce an automatic, unbiased structured detection method for correlated region detection between two random processes observed on a common domain. We argue that although intuitive, using the maximum loglikelihood statistic directly suffers from potential bias and substantially reduced power. Therefore, we introduce a simple size-based normalization to overcome this problem. We show that scanning using the proposed statistic leads to optimal correlated region detection over a large collection of structured correlation detection problems.