Lung Cancer Classification Using Deep Learning Coupled with Medical Imaging Techniques to Assist Early Diagnostics

Cancer is oncogenic transformation of cells stimulated by carcinogens which stops contact inhibition that leads to tumors (benign or malign). As per World Health Organization (WHO), 10 million people died in 2020 due to cancer, most prominently by lung cancer, colon cancer, liver cancer, stomach cancer and breast cancer. Nearly, half of the cancer patients die due to late diagnosis. Current diagnostic techniques include biopsy and histopathology of tissues, radiography, computed tomography (CT), and magnetic resonance imaging (MRI). Some treatment techniques include chemotherapy, hormone therapy, and surgery. Although medical sector have these techniques, cancer at higher stages is still incurable, therefore late detection of cancer is fatal. With the onset of Industry 4.0, the era of Artificial Intelligence (AI) has been established. AI can be used to speed up the process of medical diagnosis, for example, convolutional neural network (CNN) to detect cancer based on medical report of the patient (X-ray, CT scan or MRI scan). In this research, we have developed two CNN models on MRI and histopathological images for lungs cancer diagnostics, with achieving validation accuracies as 96.45% and 99.57%, and validation losses as 0.12 and 0.01, respectively. A full stack website was developed by using flask as backend and React.js as frontend where CNN models were hosted on as backend API to serve image classification requests from user. The website was deployed using open-source deployment platforms.