A Method for Enhancing Lung Nodules in Chest Radiographs by Use of LoG Filter

Abstract

To make the visual of any region with a spherical structure (where a potential nodule may happen to occur) in lung fields in a chest radiograph more clearly and to better determine the extent of that region and the presence of any other region of the same nature, this paper proposes a method for enhancing lung nodules in chest radiograph images by use of Laplace of Gaussian (LoG) filter. The key for the implementation of the method is the selection of the standard deviation of the LoG kernel. After LoG filter convolution on a chest radiograph image, regions of high circularity in the image are enhanced, while other regions are suppressed, and thus the contrast between nodules and normal anatomy is improved. With respect to other methods, the main advantage of LoG filter based method is that no prior explicit knowledge about the actual shape of the nodules and structure of image background is needed. The methods have been tested on a publicly available database of 52 chest radiographs, in which the absence and presence of nodules in the chest radiographs were confirmed by use of CT examinations. Experimental results demonstrate that the proposed method in enhancing lung nodules in chest radiographs is efficient and effective.