1. A nonuniform projection distribution CT method for solitary lung nodule follow-up: personal previous lung image-guided, patchwise, low-rank constrained imaging
- Author
-
Sun Jianqi, Bai Huiping, Song Ying, Zhao Jun, and Zhang Weikang
- Subjects
Rank (linear algebra) ,Computer science ,Signal-To-Noise Ratio ,030218 nuclear medicine & medical imaging ,Image (mathematics) ,03 medical and health sciences ,0302 clinical medicine ,Region of interest ,Image Processing, Computer-Assisted ,medicine ,Humans ,Radiology, Nuclear Medicine and imaging ,Computer vision ,Projection (set theory) ,Lung ,Projection View ,Radiological and Ultrasound Technology ,business.industry ,Solitary Pulmonary Nodule ,Nodule (medicine) ,Distribution (mathematics) ,030220 oncology & carcinogenesis ,Artificial intelligence ,medicine.symptom ,Tomography, X-Ray Computed ,business ,Algorithms ,Follow-Up Studies - Abstract
In this paper, a nonuniform projection distribution (NUPD) CT method is proposed for the region-of-interest-specific examination in the solitary lung nodule follow-up application in order to reduce redundant x-ray projections exposed on normal tissues. The method exploits personal previous lung CT scan information to design a nonuniform x-ray projection modulation scheme where x-rays are sparsely modulated over the areas outside of the nodule in each projection view. The nonuniform projection modulation scheme could obstruct 71.84% of the x-ray projections within each scanning view in the case of a 40-mm-diameter region of interest of a solitary lung nodule, and eventually, 96.80% of the x-ray projections are eliminated in a complete circular scan in cooperation with the double sparse sampling protocol. We also devise a prior image-guided patchwise low-rank reconstruction in NUPD CT to improve the imaging quality. The proposed reconstructions have the highest Peak Signal to Noise Ratio values compared with other methods for both the full field of view (FOV) and region of interest (ROI). The structural similarities with the references are 0.9922 for the complete FOV and above 0.9999 for the ROI.
- Published
- 2020