Medical image segmentation is an actual task in biomedical engineering. It can help physicians analyse the medical image to achieve a more accurate diagnosis. The aim of this work was to develop an effective segmentation algorithm. A classic FOREL clustering algorithm was used as basis for development of the modified clustering algorithm. This paper describes the principles of the suggested method and the achieved results. The modified algorithm was tested on a tomographic image of a brain, acquired by computer tomography. The experimental results shows that the modified algorithm allows solve the problem of image segmentation more effectively than the classic FOREL algorithm.