Comparison of two detection algorithms for spot tracking in fluorescence microscopy images

Abstract

Research in biological image analysis plays an important role in understanding the underlying mechanisms of cellular processes, which may lead to better knowledge of certain aspects of the cell function. The primary analysis of biological images requires the detection and tracking of hundreds of spots. In this paper we presents an approach for the tracking of spots in microscopy images based on the modification of the algorithm presented by Feng et al. The improved algorithm consists of replacing the original detection algorithm, Feature Point Detection(FPD) with Isotropic Undecimated Wavelet Transform (IUWT). The tracking algorithm based on Interacting Multiple Model (IMM) remains fixed. The performance of the presented method, IMM-IUWT along with two others, MHT, based on Multiple Hypothesis Tracking (MHT) and IMM-FPD, were validated on numerous challenging realistic synthetic image sequences, and their performance was evaluated using root mean square error (RMSE) metrics. The results indicate that the presented method outperforms the original method. At high level of signal to noise ratio (SNR), it is noted that the performance of the modified method, IMM-IUWT, is similar to that of MHT method. The quantitative comparative results demonstrated the importance of spot detection in tracking contexts.