Classification of multiplex fluorescence in situ hybridization images using fuzzy clustering and wavelets-based preprocessing

Multiplex or multicolor fluorescence in situ hybridization (M-FISH) imaging is a recently developed technology in medical diagnosis such as cancers and other genetic diseases. M-FISH images are used to evaluate chromosomal aberrations, mainly responsible for medical disorder. The reliability of this technique greatly depends on the correct karyotyping or classification of the M-FISH images. Currently available methods are inefficient to meet requirements of cytogeneticians in disease identification.; Our aim is to improve the classification accuracy of M-FISH images. We introduced fuzzy clustering approaches which outperforms the conventional Bayesian approaches. We also introduced two different approaches for registering the images using principal component analysis and shift invariant wavelet analysis. To improve the speed and accuracy, multi-resolution search is employed. Our algorithm is tested on real world M-FISH database, demonstrating the superior performance. The increased classification accuracy improved the reliability of the M-FISH imaging technique for disease diagnosis and clinical research.