Dual Tree Complex Wavelet Transform Based Sperm Abnormality Classification

In the proposed study, Dual Tree Complex Wavelet Transform (DTCWT) based statistical features that are derived from normal sperm, abnormal sperm and non-sperm patches are fed to Support Vector Machine classifier with the aim of three class discrimination. The obtained results are compared with the classical dyadic discrete wavelet transform and the superiority of the proposed method has been shown in terms of accuracy and F-measure metrics. The results show that higher accuracy and F-measure scores have been obtained with the proposed approach due to the shift invariance and better direction selectivity property of the DTCWT.