基于GPU的低剂量颞骨CT快速优化重建研究

Study of Low-Dose CT with Fast Reconstruction upon the GPU Applying in the Temporal Bone

【目的】探讨在图像处理器（GPU）提高SAFIRE重建速度的基础上，探寻满足诊断需求的颞骨低剂量扫描方案。【方法】选择本院行颞骨CT扫描的成年患者80例，随机分为对照组（行常规扫描，管电压120kV，管电流120mAs）和观察组（管电压80kV，管电流120mAs），每组40例；对照组图像进行常规FBP重建，观察组行常规的FBP、SAFIRE重建和经过GPU处理的SAFIRE重建。分别记录两组的剂量长度乘积（DLP）计算出有效辐射剂量ED，记录两组患者重建方法需要的时间，测量图像的噪声（SD）和CT值并计算出信噪比（SNR），用统计学方法对所测的数据进行分析。【结果】观察组的有效辐射剂量较对照组低；经FBP重建后，观察组SD值较对照组明显升高，SNR值明显降低，差异均具有统计学意义（均P〈0．05）；观察组经SAFIRE重建后图像SD较FBP重建明显降低，sNR明显升高，且差异具有统计学意义（均P〈0．05）；观察组的图像质量差但经过SAFIRE重建后图像质量仅稍低于A组，但是可以满足诊断要求，SD值相对稍高，SNR值相对稍低，但是差异不具有统计学意义（均P〉0．05）。【结论】颞骨以管电压80kV，管电流120mAs的CT扫描后的图像经SAFIRE重建后完全能满足诊断需求，且明显降低了辐射剂量，GPU可明显降低sAFIRE重建时间，提高工作效率。

[Objective] To search the available scanning program of temporal bone to reduce radiation dose and meet the diagnostic requirements upon the Graphic processing unit （GPU） improvement of the speed of SAFIRE reconstructions. [Methods] 80 cases of adult patients with temporal bone CT scan in our hospital were selected and randomly assigned to the control group （routine scanning, 120 kV tube voltage, tube current 120 MAS） and the observation group （80 kV tube voltage, tube current 120 MAS）, each group of 40 cases. The control group images were reconstructed by conventional FBP, and the observation group was treated with conventional FBP, SAFIRE and SAFIRE reconstructed by GPU. The dose length product （DLP） of the two groups were recorded to calculate the effective radiation dose of ED, and the time of the reconstruction method was recorded, the noise （SD） and CT values of the images were recorded, and the signal to noise ratio （SNR） was calculated. All data were analyzed by statistical methods.[Results]The effective radiation dose in observation group was lower than that in control group, and the SD value in the observation group was significantly higher than that in control group, SNR significantly decreased, the difference was statistically significant （ P〈0.05）, and the SD of the observation group was significantly lower than that of FBP reconstruction, SNR was significantly increased, and the difference was statistically significant （ P 〈0.05） ; After SAFIRE reconstruction, the images of the observation group were significantly lower than that of SD reconstruction, and the SNR was significantly higher, and the difference was statistically significant （ P 〈 0.05） ; The image quality of the observation group was poor, but the image quality was only slightly lower than the A group after SAFIRE reconstruction, it can meet the diagnostic requirements; the SD value was slightly higher, and the SNR value was slightly lower, but the difference does not have statistical significance （ P 〉0.05）.[Conclusion] The scanning program （80KV, 120mAs） of temporal bone reconstructed by SAFIRE can meet the diagnostic requirements and reduce the radiation dose obviously, and GPU can reduce the time of reconstruction and improve the work efficiency.