Recently, computer-aided diagnosis (CAD) has become one of the major research subjects in medical imaging and diagnostic radiology. CAD provide a computer output as a "second opinion" to assist radiologists' image readings. The early evolution and quantitative analysis of the human spontaneous intracerebral brain hemorrhage (ICH) are significant for the treatment and the follow-up of patient's recovery. This paper presents a new method for semiautomatic segmentation and tracking of ICH from computerized tomography (CT) head images, which is based on a suitable combination of digital image processing and pattern recognition techniques. The segmentation and tracking processes involved two main steps: first selection of a query object (ICH) on one study slice (image) and second the automatic object retrieval on the rest of the slices of the patient study. The proposed method had been tested successfully by medical researchers in a representative dataset of 36 ICH CT head images patient studies.