Rotational invariant fractional derivative filters for lung tissue classification

Abstract

A new and powerful rotation invariant fractional derivative convolution neural network model is proposed for the classification of five categories of interstitial lung diseases. Fractional derivative convolution neural network model employs fractional derivative filters for the texture enhancement of the lung tissue patches instead of the raw image, is given as input directly to the convolution neural network. These FD filters are rotation invariant which solves the problem of rotation invariance of lung tissue patterns caused by pose variations of the patient during CT scanning. Also, the problem of the poor performance of most classifiers such as, support vector machine and K-nearest neighbours caused by an imbalanced dataset is solved, by oversampling the minority categories emphysema and ground glass patches, and under-sampling the majority category, micronodules patches. The experimental results are executed on the publicly available interstitial lung disease database which shows the fractional derivative convolution neural network model performs better than the state-of-art with average F-score and accuracy noted as 93.32% and 93.33% respectively. INTRODUCTIONInterstitial lung diseases (ILD), is a diverse group of disorders found in the tissue between the air sacs of the lungs that leads to fibrosis, or scarring of an interstitium. The scarring of an interstitium makes it difficult for the lungs to receive the oxygen. 15% of all the cases seen by pulmonologists that account for ILD is caused by autoimmune disease, genetic abnormalities, infections, drugs, or long-term exposure to the hazardous materials [1]. Most of the cases, the cause remains unknown and the disease is left as idiopathic. High-Resolution Computerized Tomography (HRCT) imaging is being widely used for visualizing the texture variations of different ILD patterns [2]. However, distinguishing between lung tissue patterns such as Normal, Micronodules, Ground Glass, Fibrosis and Emphysema is a challenging problem because these tissues exhibit similar appearance between different tissue categories and also exhibits great variations within the same categories as shown in (Figure 1). When the radiologists are under heavy workload to manually identify the type of tissue pattern, whichThis is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.