Design and Analysis of an Isotropic Wavelet Features-Based Classification Algorithm for Adenocarcinoma and Squamous Cell Carcinoma of Lung Histological Images

Abstract

One of the most prevailing types of lung cancer is non-small cell lung cancer (NSCLC). Differential diagnosis of NSCLC into adenocarcinoma (ADC) and squamous cell carcinoma (SCC) is important because of prognosis. Histological images are taken from a database consisting of 72 lung tissue samples collected indigenously with a core needle biopsy. In this work, a novel method has been developed where the features of ADC and SCC for a histological image are taken from various statistical and mathematical models implemented on the coefficients of the wavelet transform of an image. The method provides a precision of 95.1% and 96.2% in classifying malignant and non-malignant tissue type respectively. This methodology of classifying ADC and SCC without coding clinical diagnostic features into the system is a necessary step forward towards an autonomous decision system.