Cancer Detection: A Rule-Based Method for Categorizing Brain Tumors

The most important thing in the process of brain tumor diagnosis is to be correct and at the earliest level that the treatment can be administered in a timelier manner which will prove to be a difference in the survivability of brain tumor patients. The indications of biopsies and MRI scan are very good, but it requires very expensive equipment and/or trained personnel and therefore is a very costly and time-consuming procedure. This paper provides a rule-based system to facilitate this process and requires the provision of clear medical rules that will be followed to diagnose benign and malignant brain tumours. They include size of the tumor, shape of the tumor, borders of the tumor, type of cell which is found to be involved in a tumor and the existence of certain hormone receptors in the system. Testing was done to a publicly available number of MRI brain images that had been specifically tested after being subjected to proper and comprehensive preparation and cleaning of the information before testing. The system code was developed in python and checked based on accuracy, confidence matrices and classification reports.

Перевод пока недоступен