Automatic Detection and Epidemiologic Trends in Head and Neck Cancer Using Fast-R-CNN and SVM Methods

H ead and neck squamous cell carcinoma (HNSCC) is the sixth most frequent cancer in the world. Despite improvements in diagnosis and therapy, the five-year survival rate for patients with advanced HNSCC has risen only slowly. This is mostly due to the high number of patients who present with an advanced disease stage at diagnosis, as well as the high frequency of relapse and secondary tumors. Protein biomarkers that aid in the early diagnosis of primary HNSCC or relapse hold promise for improving clinical outcomes. Finding early changes in the mucosal linings previous to the development of invasive tumors and properly assessing the likelihood of malignant transformation, both currently tough activities, may be important. The proposed procedure makes use of preprocessing, feature extraction, segmentation, and model training. Anisotropic Diffusion Filter and Gaussian Filter are used in preprocessing. Extraction of the techniques follows preprocessing. The Grey Level Run Length matrix is used for feature extraction. Segmenting the nucleus is a highly recommended step in the segmentation process. Training is performed using an RCNN- SVM model. When compared to two other models, RCNN and SVM, the proposed method excels.