The computer evaluation and diagnosis of polypoid lesion of the stomach, using radiologic findings

Summary  Intending to improve the radiologic diagnosis of the stomach, we summarized various kinds of polypoid growths of the stomach as “Polypoid Lesion” and, dividing them into four groups by their profile view, we investigated size, roughness of the surface and indentation of the base. (Yamada T.et al. ’65) Using the computer technique and facilitating the multiple calculations, we purposed the standardization and the objectivity of the radiologic diagnosis in this study. Applying the Bayes theorem, we calculated the probability of malignancy, basing on the macroscopical type, size, roughness of the surface and indentation of the base of 264 operated polypoid lesions. Investigating the diagnostic validity on 77 polypoid lesions of the stomach, we were revealed the diagnostic accuracy of 91% and 87% by the computer and radiologie diagnosis respectively. This is due to the characteristic data processing of the computer. On the other hand, the computer could diagnose correctly 82% of the minimal lesions, like the early gastric carcinomas. This is due to the poor data acquisition and/or deficient plan processing for the rarely experienced lesions. The computer diagnosis could become a great help to the standardization, objectivity and the pre-operative final diagnosis, investigating the plan processing and the data processing.