Denoising and block gridding of microarray image using mathematical morphology

Microarray has been used as an important tool for monitoring gene expression levels. Through microarray analysis, gene expression levels can be obtained and the biology of a disease can be identified based on the obtained result. Hence, the therapeutic drugs to treat the disease can be decided. However, analysing microarray images are complicated because the images are noisy (contain unwanted pixels) and have low contrast between spot of interest and its background. Thus, it is crucially essential to optimally remove noise from microarrays images through pre-processing techniques before gridding them into blocks. Therefore, this paper proposes a technique that comprises two steps; pre-processing and processing (gridding). The pre-processing step is conducted by using mathematical morphology to remove noise while for processing, block gridding relies on mean profile or also known as projection profile, followed by morphology operation closing. The obtained results from pre-processing reveal that Tophat morphological operation is the best operation for microarray denoising among other existing operations like Bottomhat operation, Opening operation and Closing operation with SNR value as low as 1.3248dB, while still able to preserve information; thus producing better block grids of a microarray.