A model of semiautomatic type-specific diagnosis of brain tumours by computed tomography: mathematical fundaments and practical application.

Applying Escudero's formula to Bayes' theorem of conditional probabilities, we have developed a mathematical model to diagnose type-specific brain tumours with great accuracy. We have transformed a multidimensional system of n characteristics on computed tomography into a linear system, making this model very easy to use and, thus, accessible to the clinician and radiologist. We tested this model in 177 brain tumours. In 86% of the cases we obtained a type-specific correct diagnosis, achieving better results than those previously reported. Suggestions are given as to how to improve this method further.