Content based medical image classification using discrete wavelet and cosine transforms

In this study, we aimed to determine whether the medical image belongs to that class or not, using the textural features of medical images. The study was performed on the images in IRMA (Image Retrieval in Medical Applications), the international database. After performing pre process on the our current medical images, discrete wavelet transform (DWT) was applied and then discrete cosine transform (DCT) was applied to each band components. After feature extraction, using of 1%, 3%, 5% and 7% of the obtained data were classified. K-Nearest neighbor algorithm (KNN) was used in the classification phase. The classification performance was around 87%.