FLASHμ: Fast Localizing And Sizing of Holographic Microparticles
This enables low-cost, real-time holographic imaging for applications requiring fast microparticle analysis, though it represents an incremental improvement over existing machine learning methods.
The authors tackled the problem of reconstructing 3D location and size of small microparticles from holograms, which is computationally expensive with existing methods. Their two-stage neural network FLASHμ detects particles as small as 9μm in real holograms with comparable accuracy to standard approaches while providing a 600-fold speedup.
Reconstructing the 3D location and size of microparticles from diffraction images - holograms - is a computationally expensive inverse problem that has traditionally been solved using physics-based reconstruction methods. More recently, researchers have used machine learning methods to speed up the process. However, for small particles in large sample volumes the performance of these methods falls short of standard physics-based reconstruction methods. Here we designed a two-stage neural network architecture, FLASH$μ$, to detect small particles (6-100$μ$m) from holograms with large sample depths up to 20cm. Trained only on synthetic data with added physical noise, our method reliably detects particles of at least 9$μ$m diameter in real holograms, comparable to the standard reconstruction-based approaches while operating on smaller crops, at quarter of the original resolution and providing roughly a 600-fold speedup. In addition to introducing a novel approach to a non-local object detection or signal demixing problem, our work could enable low-cost, real-time holographic imaging setups.