Quantitative analysis of bronchiectasis using local binary pattern and fuzzy based spatial proximity

Quantitative analysis of bronchiectasis in computed tomography(CT) images of the lungs is needed in diagnosis of the disease, so that proper and correct medication can be provided for the patients. In this work new methods to analyze bronchiectasis quantitatively using Local binary pattern and fuzzy based spatial proximity are proposed and the results are compared with K-Means. Local binary pattern serves as a powerful tool in diagnosing diseases in terms of accuracy and reduction of computational complexity. This method of texture analysis which takes into account the local area around a particular pixel rather than single intensity eliminates the noise in the image. The disease severity is then analyzed by region growing which quantifies the amount of disease spread. Fuzzy based spatial proximity system provides efficient identification of bronchial walls by which the cystic lesions which is the primary characteristic of bronchiectasis, can be identified and separated. The diameter of each lesion can be found separately and quantification can be done by finding the probability of the amount of disease spread over the entire lung area. The k-means algorithm can act as the supportive tool in the quantification of the disease.