This paper presents the potential of derived shape features in the classification of the brain hemorrhage. Derived shape features are secondary features which are calculated from commonly used popular primary shape features. These features contain more relevant information having higher dependency on the shape of the target hemorrhage. Selection of high potential features is done to reduce the dimension of the input feature set to optimize classifier accuracy. The potential of these derived features is demonstrated and discussed with respect to the primary features. © 2020, Springer Nature Singapore Pte Ltd.