智能4D-CT；模体

Intelligent 4D-CT; phantom

呼吸伪影减小

Motion artifacts were reduced

呼吸自适应断层扫描；模体

Respiratory adaptive computed tomography; phantom

靶区和肺≥4 mm伪影发生频率降低70%和76%；呼吸信号不规律减少27.3%

The frequency of respiratory-induced image artifacts ≥ 4 mm decreased by 70% and 76% for the tumor and lung; Irregular respiratory signals decreased by 27.3%

呼吸运动引导4D-CT；模体

Respiratory motion guide 4D-CT; phantom

成像剂量减少17.8%；呼吸信号不规律减少12.6%

Imaging dose decreased by 17.8%; Irregular respiratory signals decreased by 12.6%

基于B样条配准；主成分分析

B-splines; principal component analysis

模型误差为(2.35±0.26) mm\(2.48±0.18) mm

Model error was (2.35±0.26) mm\(2.48±0.18) mm

基于B样条配准；主成分分析；贝叶斯原理

B-splines; principal component analysis; bayesian

单周期\双周期构建模型误差为（0.57±0.06) mm\（1.52±0.41) mm

Error of model based on Single-cycle\Double cycles was（0.57±0.06) mm\（1.52±0.41) mm

基于SUM、MIP和EXS勾画靶区

Target contouring based SUM; MIP; EXS

MIP、EXS勾画靶区与SUM勾画靶区的体积比分别为0.888±0.061、0.883±0.064

The ratios of for MIP and EXS to SUM were 0.888±0.061, 0.883±0.064

基于MIP和AVG勾画靶区

Target contouring based MIP and AVG

MIP勾画不规则运动范围较大的靶区 ((20.8±2.6) mm)剂量一致性差

MIP delineation of moving irregularly target with a larger range ((20.8±2.6) mm) has poor dose consistency

基于模型4D-CT

Model-based 4D-CT

ITV区域内模型平均误差为(1.71±0.81) mm；呼吸伪影减少

Mean model error within the ITV regions was (1.71±0.81) mm; motion artifacts were reduced

鲁棒主成分分析；

霍恩&舒克的光流法

Robust principal component analysis;

Horn and Schunck optical flow method

PSNR与SSIM分别提高25.4%与7.6%（MC-FDK）；PSNR与SSIM分别提高37.9%与17.6%（FDK）

The improvements of PSNR by 25.4% and SSIM by 7.6% (MC-FDK) and of PSNR by 37.9% and SSIM by 17.6% (FDK)

60个角度下PSNR与SSIM分别为49.02与0.9855

PSNR and SSIM (AP-AirNet) were 49.02 and 0.9855 respectively under 60-view

自适应4D-CBCT；MC-FDK

Adaptive 4D-CBCT; MC-FDK

成像剂量减少85%；扫描时间减少70%

Imaging dose was reduced by 85%; scan times were reduced by 70%

18个呼吸周期下PSNR与SSIM最大为32.73和0.935；扫描时间减少约90%

PSNR and SSIM were up to 32.73 and 0.935 under 18 breathing cycles;

Scanning time was reduced approximately 90%

PSNR和SSIM分别提高了约18%和51%

The improvements of PSNR by 25.4% and SSIM by 7.6% approximately.

光流约束

Optical low (OF) constraint

SSIM提高了2.8% (ART-TV)和23.4% (FDK)

SSIM increases 2.8% (ART-TV) and 23.4% (FDK)

U-net；迁移学习

U-net; transfer learning

PSNR和SSIM分别为38.42和0.958

PSNR and SSIM were 38.42 and 0.958, respectively

下叶肺肿瘤

Lower lobe lung tumor

运动范围最大减少6.4 mm

The range of motion is reduced by up to 6.4 mm

上/中叶肺肿瘤;下叶肺肿瘤

Upper/mid-lobe lung tumor;

lower lobe lung tumor

运动范围分别减少0.8 mm；3.5 mm

The range of motion is reduced by 0.8 mm and 3.5 mm respectively

肺肿瘤

Lung tumor

运动范围减少7.5 mm

The range of motion is reduced by 7.5 mm

肺肿瘤

Lung tumor

PTV从148 mL减少至110 mL

Planning target volumes were reduced from 148 mL to 110 mL

肺肿瘤

Lung tumor

PTV减少6%

PTV reduced by 6%

肺肿瘤

Lung tumor

平均肺剂量减少29%

There was a 29% reduction in the mean lung dose

肺肿瘤

Lung tumor

肺肿瘤

Lung tumor

放射性肺炎风险降低11%

The risk of Radiation pneumonia is reduced by 11%