Tomographic Reconstruction

Tomographic reconstruction aims to create three-dimensional images from multiple two-dimensional projections, addressing ill-posed inverse problems common in medical imaging, materials science, and other fields. Current research emphasizes developing advanced deep learning models, including convolutional neural networks, diffusion models, and implicit neural representations, often integrated with physics-informed approaches or traditional iterative methods to improve reconstruction accuracy and efficiency, particularly in sparse-view or low-dose scenarios. These advancements are significant for reducing radiation exposure in medical imaging, enhancing the resolution and speed of various tomographic techniques, and enabling new applications in areas like 3D microscopy and industrial inspection.