双探头符合探测图像的噪声分析

Noise analysis on dual-head coincidence detection FDG imaging

目的 分析双探头符合探测图像的噪声成分。方法 采用Hawkeye双探头SPECT仪及东芝模型 ,对 7种活度进行采集重建图像 ,测量1 8F 脱氧葡萄糖 (FDG)均匀分布区域像素均值及标准差 ,对数据进行曲线拟合。结果 噪声分为 3个成分 :成分 1与信号强度成正比 ,由随机符合造成 ;成分 2与信号强度的平方根成正比 ,由放射性统计涨落造成 ;成分 3与信号强度无关。图像的信噪比平方与计数成二次 线性关系 ,随计数增加信噪比增加 ,但其速度逐渐缓慢 ,达最大值后开始降低。结论 符合探测图像噪声的平方与计数成二次 线性关系。与单光子显像相比 ,其噪声中增加了与信号强度成正比的随机符合噪声部分。

Objective To investigate components of noise in dual-head coincidence fluorodeoxyglucose (FDG) imaging in order to improve the image quality. Methods Using GE Hawkeye dual-head coincidence detection equipment, emission scan was performed for 7 times on Toshiba AZ-618 phantom. The average pixel counts and standard deviation were measured within the region of interest (ROI) in the uniform FDG area of the phantom. The data were processed with curve fitting. Results The noise on the image could be categorized into three components. The first one varied directly with signal intensity and depended on random coincidence; the second one varied directly with the square root of signal intensity and depended on Poisson fluctuation; and the third one was independent of signal intensity. The relationship between signal-to-noise ratio and the signal intensity could be expressed with a quadratic polynomial. With the increment of the signal intensity, signal-to-noise ratio increased at first, but decreased later. Conclusions The relationship between the square of the noise on the image and the signal intensity could be expressed with a quadratic polynomial. Comparing with single photo imaging, a random coincidence noise component is added to the coincidence imaging.