Objectives: In this study, we developed a neuro-fuzzy based system for classification of cancerous and non-cancerous lung cells. Methods: Images were pre-processed using median filter algorithm, segmented using marker-controlled watershed algorithm, and were extracted using gray-level co-occurrence matrix. A hybridized diagnosis system that made use of neural network and fuzzy logic for classification of lung cells into cancerous and non-cancerous cells is modelled. Computed tomography (CT) scan image dataset of the lung was downloaded from The Cancer Imaging Archive dataset. Neural network performed the training and classification of the lung cells with back-propagation algorithm, while the cancerous cells were passed into fuzzy inference system to determine the lung cancer stage. Results: Our system was able to successfully classify the imported CT scan images into normal or abnormal with considerably high accuracy of 70% and precision of 89%. This system can support physicians in decision making when diagnosing cancer.
Select your language of interest to view the total content in your interested language