Brain tumor extraction method using KNN algorithm by the K parameter variation

Early tumor detection, especially of brain tumors is a paramount recommendation of the world health organization, The earlier tumors are detected, the higher the chances of a cure. Detecting tumors at the first stage significantly increases these chances. The number of MRI images to be processed every day is immense and very huge, requiring the automation this step of MRI images processing and analysis. In this article we propose a new brain tumor detection and extraction method based on the KNN algorithm known for its efficiency and precision. Before tumor detection and extraction, we proceed to the MRI image filtration and denoising by the Nl-means filter to improve the accuracy of the results as a first pretreatment step. The second step of preprocessing is the brain extraction by the EMBE method to reduce the overlapping of the results thresholds.The tumor detection and extraction are obtained by the KNN algorithm. We proceed to the "K" parameter variation, and we fix on the other hand the method of the distance calculation "d" to the Hamming distance. at each combination (K, d) we check the error rate to choose the right combination of the "K" parameter and the "d" distance.