Diagnostic Supporting System for CT Scan Diagnosis of Cervical Lymph Node Metastasis in head and neck cancer using Fuzzy Inference : Tuning of Membership Function and Reconstruction of the System

Abstract

Diagnosis with medical imaging such as CT scan essentially contains fuzziness because of subjective assessment of the findings by examiner. We have developed a computer diagnostic supporting system using fuzzy inference, which allows objective assessment of cervical lymph nodes metastasis in head and neck cancer. The input and output membership functions were prepared for each item and the final assessment was made by the center of gravity of the max-sets of the output membership function. Evaluation of the adequacy for this system was performed in 67 CT scans of cervical lymph nodes in head and neck cancer patients by three examiners with different clinical experience. The diagnostic accuracy of this system was approximately 90% for every examiner. This diagnostic supporting system appears to be beneficial for diagnosis of cervical lymph node metastasis by CT scan, however there was 4 falsenegative cases in 67 cases. Therefore this system was tuned up a membership function, and the possibility for a clinical use was evaluated.