X-ray “ghost” imaging has drawn great attention for its potential to obtain images with a high resolution and lower radiation dose in medical diagnosis, even with only a single-pixel detector. However, it is hard to realize with a portable x-ray source due to its low flux. Here, we demonstrate a computational x-ray ghost imaging scheme where a real bucket detector and specially designed high-efficiency modulation masks are used, together with a robust deep learning algorithm in which a compressed set of Hadamard matrices is incorporated into a multi-level wavelet convolutional neural network. With a portable incoherent x-ray source of ∼37 µm diameter, we have obtained an image of a real object from only 18.75% of the Nyquist sampling rate. A high imaging resolution of ∼10 µm has been achieved, which is required for cancer detection and so represents a concrete step toward the realization of a practical low cost x-ray ghost imaging camera for applications in biomedicine, archeology, material science, and so forth.
