A pattern detection apparatus based on a scanning transmission electron microscope having an electron gun for generating and accelerating an electron beam, a plurality of convergent lenses for converging the electron beam, a deflection circuit for deflecting the electron beam so that it scans an object to be inspected, such as an X-ray mask, a detection circuit which receives electrons that have been dispersed and transmitted in the object and converts the detected electrons into an electrical signal, and an image forming circuit which forms a detected image of the object under test in response to the detected signal from the detection circuit and in synchronism with the deflection signal applied to the deflection circuit.