Abstract
We developed a computer system to assist the differential diagnosis of clinically important oral mucosal diseases of oral squamous cell carcinoma, leukoplakia, and oral lichen planus using intraoral photographs. In 2013, we reported the use of a fuzzy inference-based soft computing diagnostic system for oral mucosal diseases. In the current study, we replaced the fuzzy inference system with a support vector machine and used five features for identifying disease characteristics after upgrading the soft computing system. We used digital intraoral photographs of oral squamous cell carcinoma, leukoplakia, oral lichen planus, and normal mucosa to validate the soft computing system. The resultant sensitivity values for oral squamous cell carcinoma, leukoplakia, oral lichen planus, and normal oral mucosa were 90％, 90％, 87％, and 93％, respectively, and the specificity values were 97％, 92％, 98％, and 100％, respectively. Therefore, the soft computing system developed in this study can be used as a diagnosis support system for clinically important oral mucosal diseases.
