Discriminant analysis is still foremost statistical method for classifying individual subjects into desired groups. In the beginning, individual subjects are classified based on simple criteria. In the next step, they are reclassified using better discriminator variable.

Stroke is still the major health problem because of its morbidity and mortality. It results in various neurological signs and sequelae. Classifying stroke patients into appropriate groups is important to manage them properly. The National Institute of Health Stroke Scale (NIHSS) is common to be used for classifying stroke patients into groups. This scale is considered to be simple to implement. However, it has limitation since it is influenced by various factors. Magnetic resonance imaging (MRI) is considered to be gold standard for classifying stroke patients into groups as well as for diagnosis. Then MRI is considered to be better for classification than NIHSS. The purpose of this study is to classify ischemic stroke based on MRI results using two groups discriminant analysis.

A non-reactive or unobtrusive research is implemented by using secondary data obtained from Machfoed et al. [11]. Twenty eight patients with ischemic stroke are selected. They attended Brain Clinic Surabaya who underwent MRI examination and NIHSS within 24 hours of stroke onset. Four discriminator variables are selected. These were volume of lesion (cm³) in first localization (Vol), the number of lesion localizations (Noll), age in year (Age), and the number of hemispheres where lesion is located (Noh). Two groups discriminant analysis of data is performed to classify ischemic stroke patients.

The results show that age of patients varies around 40-89 years old. Score of NIHSS varies around 1-9. Male patients are higher than female ones (60.7% compared to 39.3%). A quarter of patients (25%) does not show ischemic stroke lesion while half of patients (50%) shows ischemic stroke lesions in one localization. Four discriminator variables classify the patients into minor and moderate stroke with degree of correct classification 72.2%. It is recommended to perform further study with larger sample size and various categories of stroke patients.

discriminator, ischemic stroke, NIHSS, MRI.