DEVELOPMENT OF FD FILTER BANKS FOR THE SCREENING OF INTERSTITIAL LUNG DISEASES
Fractional derivative filters provide rich texture features for the medical images. The concepts of fractional calculus were 300 years old but their applications to the image-processing field are novel. More over fractional derivative filters preserve low-frequency features near smooth areas and enhances high-frequency features like edges, which are prominent in an image. So, in this paper we designed fractional derivative filters extracted from the Srivastava-Owa operator. The fractional derivative filters are rotation invariant that solves the rotational variance problem caused by the pose variation of the patient during CT scanning. For this texture enhanced image, the features are extracted using Local binary pattern (LBP), gradient features by Histogram of oriented gradients (HOG), and labelling is done with support vector machine (SVM). The current work is implemented on ILD database which gives 90.67% accuracy.