Researchers have developed a new end-to-end neural network that can speed up the reconstruction of holographic images. Unlike other deep learning techniques, the approach can be used on samples not encountered during training, making it particularly useful for label-free holographic biomedical imaging.
