A method comprises a first step of dividing a measured medium into N voxels; a second step of assuming an imaginary medium identical to the measured medium except for absence of absorption; a third step of setting a light injection position and a light detection position for the imaginary medium; a fourth step of calculating an impulse response s(t) at the light detection position; a fifth step of calculating a probability density Ui(t,t) that the impulse response at a time t at the light detection position has existed in voxel i (i=1, . . . , N; 1N) at a point of time t (0tt); a sixth step of calculating a cumulative probability Ui(t) that the impulse response detected at the time t has existed in voxel i, from Ui(t,t); and an eighth step of calculating a time-resolved path length li in voxel i from Ui(t) and s(t).