Hybrid-Loss Guided 3D CNN For Dynamic Dual-Tracer PET Reconstruction

作者：Jinmin Xu, Huafeng Liu · 年份：2019 · DOI：10.1109/isbi.2019.8759287 · 被引用次数：3 · 研究领域：Medical Imaging Techniques and Applications、Radiomics and Machine Learning in Medical Imaging、Advanced Radiotherapy Techniques

We present a deep-learning based framework for dual-tracer PET imaging, which can achieve the reconstruction by naturally unifies two parts: reconstruction of the mixed images and separation for each tracer. We adopt 3D CNN to learn the full features including spatial information and temporal information at the same time. Further, hybrid-loss layers are used to guide the two parts for the final separation. Experimental results on Monte Carlo simulations provide insights into the superior ability and performance of this strategy. The evaluation results demonstrate that our method can successfully recover the respective distribution from the original mixed sinogram data.