A new diffusion magnetic resonance imaging method was developed and optimized to probe the smallest structures to date (0.54 ± 0.06 µm), micron-sized tissue structures, such as small axon diameters (AxD). This method relies on probing the shortest possible diffusion time scales. Current temporal diffusion spectroscopy methods cannot distinguish small structures because the pulse sequences limit the ability to probe the shortest diffusion times. This method circumvents those limitations using oscillating gradients in lieu of pulsed gradients and will have important neuroscience applications such as probing AxD distributions. The required imaging time will need to be shortened in order for the method to be clinically useful. The thesis presents three sets of experiments that guided the selection of pulse sequence and imaging parameters needed for measuring micron-sized restrictions. Measurements were made on water diffusing between 3 μm polystyrene spheres. The gradient frequencies and amplitudes used were appropriate to infer the pore sizes between the 3 μm beads while the fit results indicated a better signal-to-noise ratio would help make the results more accurate. Measurements made on water diffusing inside various tubes indicated the gradient frequency range was not appropriate for inferences of tube diameters that were 250 μm and larger. The second set of experiments used information from the first set to choose imaging and gradient parameters for measurements of water diffusing between 3, 6, and 10 μm beads and 150 μm tubes. Pore sizes, surface-to-volume ratio and diameters could be inferred from these samples with this method. The results indicated that the method had a resolution limit which is greater than 0.5 μm given that the pore sizes between the 3 μm and 6 μm beads were indistinguishable. The more accurate inferences of diameters of the bead rather than diameters of the tubes suggest the method is better optimized for micron-sized samples rather than 150 μm-sized samples. The final experiment tested the method on human corpus callosum. The results suggest that the resolution limit of the method needs to be decreased and careful attention to the axon direction would be needed. Suggestions for improvements to the model are made.