To classify fat within a tomographic image with high accuracy.

A real tissue image forming section 310 forms a real tissue image by utilizing a real tissue threshold corresponding to the distinction between the muscle and fat of real tissue and distinguishing between muscle pixels and fat pixels within the tomographic image obtained from an X-ray CT apparatus 10. A reference image forming section 320 forms a reference image by utilizing a reference threshold shifted to more on the fat side than the real tissue threshold and distinguishing between the muscle pixels and the fat pixels within the tomographic image obtained from the X-ray CT apparatus 10. An image composition section 330 forms composite images by combining the fat pixels surrounded by the muscle pixels within the reference image with the muscle pixels within the real tissue image. An expansion processing section 340 forms visceral fat pixels corresponding to the visceral fat of the real tissue by expanding the fat pixels within the composite image.