Microarray Image Segmentation Using Fuzzy Transformation

摘要

Microarray image segmentation is a crucial issue in the microarray image analysis. Most existing segmentation methods do not consider the artifacts presented in the data, and therefore are prone to yield inaccurate measurement of the signal. This paper proposes a microarray image segmentation algorithm based on the fuzzy transformation (FZT) theory which is able to perform artifact correction. The algorithm utilizes the clustering property of FZT to identify the clusters of the input data, then classify each cluster into background, spot or artifacts according to the cluster's size and mean value. Compared with the well-known Mann-Whitney segmentation method and the recently developed greedy expectation maximization (EM) based method, the proposed algorithm is simple and robust, yielding better spot extraction as shown by the experimental results.