O'Mahoney Eoin, Murray Iain
Nuclear Medicine, Barts Health NHS Trust, West Smithfield, London, UK.
Nucl Med Commun. 2013 Mar;34(3):240-8. doi: 10.1097/MNM.0b013e32835ce5b5.
The aim of our study was to assess improvements in spatial resolution and noise control from the application of the Astonish resolution recovery algorithm for single photon emission computed tomography imaging. Secondary aims were to compare acquisitions made with low-energy general purpose collimators with those obtained using low-energy high-resolution collimators in this context and evaluate the potential of a finer matrix to improve image quality further.
A Tc-filled Jaszczak phantom with hot spheres was used to assess contrast and noise. A National Electrical Manufacturers Association triple line source single photon emission computed tomography resolution phantom was used to measure spatial resolution. Acquisitions were made using both low-energy high-resolution and low-energy general purpose collimators.
Compared with standard ordered subsets expectation maximization reconstructions, the resolution recovery algorithm resulted in a higher spatial resolution (8 vs. 14 mm full-width at half-maximum) leading to reduced partial volume effects in the smaller Jaszczak spheres. Higher image contrast was achieved alongside lower levels of noise. An edge enhancement artefact was observed in the resolution recovery corrected images. An overestimate of the target-to-background activity was also observed for the larger spheres.
The use of such an algorithm results in images characterized by increased spatial resolution and reduced noise. However, small sources of the order of 2-3 cm can be significantly overenhanced.
本研究旨在评估将阿斯顿分辨率恢复算法应用于单光子发射计算机断层扫描成像时,空间分辨率和噪声控制方面的改善情况。次要目的是在此背景下比较使用低能通用准直器与低能高分辨率准直器进行的采集,并评估更精细矩阵进一步改善图像质量的潜力。
使用装有热球体的锝填充贾斯扎克模体来评估对比度和噪声。使用美国国家电气制造商协会三线源单光子发射计算机断层扫描分辨率模体来测量空间分辨率。使用低能高分辨率和低能通用准直器进行采集。
与标准的有序子集期望最大化重建相比,分辨率恢复算法产生了更高的空间分辨率(半高宽从8毫米降至14毫米),从而减少了较小贾斯扎克球体中的部分容积效应。在实现更高图像对比度的同时,噪声水平更低。在分辨率恢复校正后的图像中观察到边缘增强伪影。对于较大的球体,还观察到目标与背景活性的高估。
使用这种算法可得到具有更高空间分辨率和更低噪声的图像。然而,2 - 3厘米量级的小源可能会被显著过度增强。