To provide a nuclear magnetic resonance inspecting device for measuring a brain function which is reliable in extraction of an activated site.

Time series pictures are photographed at previously fixed time intervals TR and in advance of photographing each time series picture, MT pulses 102, 105,..., 108, 111,..., 114, 117,... are applied. Photographing 101, 104,... of MTC pictures in a first resting period are executed from a time t1, applying of stimulation for activating a specific brain function is started from a time t2 and photographing 107, 110,... of the MTC pictures in a stimulation applying period are executed. Applying of stimulation is finished at a time t3 and photographing 113, 116,... in a second resting period are executed from the time t3. In the MTC pictures in the resting periods and the stimulation applying period, a pixel whose density value is varied is extracted as the activated site. Thus, a risk of erroneously extracting an area except a nervous area directly related with stimulation applied to an inspecting object is avoided to improve reliability of the extraction of the activated site.