The CT-perfusion image shows information of brain abnormalities such as its size and location. Generally, neurologist diagnoses stroke disease using CT-perfusion images such as Cerebral blood flow (CBF), cerebral blood volume (CBV). In our previous report, we applied threshold technique to divide amount of CBV and CBF into low and high level. Then, their levels are applied to identify normal tissue areas, dead tissue areas (infract core) and blood-cot tissue areas (infract penumbra). However, it's not totally correct, if the same threshold is applied to the whole area (it must depend on size of blood vessel in that area. In this report, we propose double thresholds to divided CBV and CBF into 3 levels: very low, medium and very high levels. Very low and very high levels are definitely implied to bad areas and good areas, respectively. The proposed double thresholds makes stroke prognosis more accurate. The simulation results confirm that our proposed results closed to results defined from neurologist comparing to the conventional results. © 2013 SPIE.