Comparison of Respiratory Motion Correction Methods in PET Lung Tumor Quantification

摘要

During PET acquisition, tumor motion due to respiration poses a major challenge for accurate localization and quantification of PET images. Respiratory gating in PET was proposed as a solution to the motion artifacts. However, PET images collected at discrete bins can be significantly affected by noise as there are lower activity counts in each gated bin. In this project, we developed a computer-assisted method that can automatically identify tumors in lung PET images of discrete bins within the breathing cycle, followed by the algorithms that register all the information of a complete respiratory cycle into a single reference bin. Four correction/registration algorithms were tested: Centroid-based registration, Intensity-based registration, Rigid Body registration and Optical Flow registration; as well as two registration schemes: Direct registration and Duccessive registration. Validation and comparison with these methods were performed by conducting experiments with a computerized phantom and a dynamic lung-chest phantom. Iterations were conducted on different sizes simulated tumors and different noise levels. Static tumors without respiratory motion were used as gold standard, quantitative results were compared with respect to tumor activity concentration, tumor correlation and signal-to-noise ratio. Optical Flow registration with Successive method demonstrates the best correlation result but is more sensitive to noise, Centroid based registration with Direct method requires the least processing time but is less accurate. After motion correction, the best compromise between short PET scan time and reduced image noise can be achieved, while quantification and clinical analysis become faster and more precise.