A multimodal matching module that aligns material imagery across different microscope modalities by learning shared representations.