A method for imaging retinal structures involves offsetting an imaging aperture from the center of the reflected/scattered light distribution. Images are obtained with the imaging aperture at multiple offsets with respect to the center. The various obtained images are co-registered to a reference confocal image in a reference wavelength and averaged to generate high signal to noise ratio images. The images may be combined in various ways to enhance the contrast of retinal structures that would not be visible in confocal images.