Lung Cancer Detection Using combination of Gabor Filter, Histogram Equalization and Multi-Layer Perceptron

Globally, lung cancer ranks second in prevalence and highest in mortality among all cancers. Previous research that utilised computer vision and machine learning faced limitations in computation cost, processing time, and the increasing number of images required for accurate assessment. The purpose of this research is to construct a machine learning multi-layer perceptron (MLP) that is simple yet effective in detecting lung cancer, and to design a more effective non-invasive detection approach. To achieve good classification, an investigation of feature extraction is important to achieve high accuracy. An analysis of Gabor Filter (GF), Histogram Equalisation (HE), and MLP to detect lung cancer has been conducted. This research comprises 800 CT lung image datasets categorised into cancerous and non-cancerous classes. The result shows the MLP itself achieved the highest accuracy with 96%, GF with MLP with 50%, and GF HE with MLP with 85%. MLP itself without feature extraction is suitable for early lung cancer detection, although it might slow down the computer because no feature extraction is used. To meet the needs of early detection where quick and accurate results are significant, the proposed model GF HE and MLP show potential.