Disease discrimination based on disease subspace of organ shape using orthogonal complement of normal subspace

Diagnostic modeling based on computational anatomy is an important topic. In previous work, discrimination method using support vector machine based on principal component analysis of the hippocampus shapes have been proposed. However, disease-specific component was not considered explicitly. In this paper, we propose a method for constructing the disease subspace using orthogonal complement of the normal subspace. The proposed method was tested using the hepatic cirrhosis and hip osteoarthritis datasets and was compared to a previous method. In our experiments, the proposed method was effective for disease discrimination based on organ shapes.