An Automatic Image Enhancement and Analysis Technique for Head and Neck Cancer Detection

Abstract

Every year, thousands of people are diagnosed with head and neck cancer. In hospitals, head and neck cancer detection is done by radiation therapy but it has some side effects and due to this therapy life quality of patient becomes less. It is detected manually by taking Computed Tomography images but it is very time consuming method. Therefore aim of this research is to detect cancer using some machine learning algorithms as this cancer rapidly increases nowadays. Head and neck cancer detection is performed by collecting 26019 CT scan images from Cancer Imaging Archive (TCIA). This research mainly focuses on classifier deep learning framework in h2o and decision tree followed by ensembling of both which gives better accuracy. Firstly, CT scan image of head and neck cancer is given as input to the system and processed through the image processing technique called weiner filter. Then the images are processed through the segmentation technique called fuzzy c-means algorithm. After that, feature extraction technique named Gray Level Co-Occurrence Matrix (GLCM) is used to extract the features. These features are given to classifier to train the model and finally it obtains the satisfactory results with 99.41% accuracy.