de Vries Bart M, Golla Sandeep S V, Zwezerijnen Gerben J C, Hoekstra Otto S, Jauw Yvonne W S, Huisman Marc C, van Dongen Guus A M S, Menke-van der Houven van Oordt Willemien C, Zijlstra-Baalbergen Josée J M, Mesotten Liesbet, Boellaard Ronald, Yaqub Maqsood
Cancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The Netherlands.
Cancer Center Amsterdam, Department of Hematology, Vrije Universiteit Amsterdam, Amsterdam UMC, De Boelelaan 1117, 1081 HV Amsterdam, The Netherlands.
Diagnostics (Basel). 2022 Feb 25;12(3):596. doi: 10.3390/diagnostics12030596.
Acquisition time and injected activity of F-fluorodeoxyglucose (F-FDG) PET should ideally be reduced. However, this decreases the signal-to-noise ratio (SNR), which impairs the diagnostic value of these PET scans. In addition, Zr-antibody PET is known to have a low SNR. To improve the diagnostic value of these scans, a Convolutional Neural Network (CNN) denoising method is proposed. The aim of this study was therefore to develop CNNs to increase SNR for low-count F-FDG and Zr-antibody PET. Super-low-count, low-count and full-count F-FDG PET scans from 60 primary lung cancer patients and full-count Zr-rituximab PET scans from five patients with non-Hodgkin lymphoma were acquired. CNNs were built to capture the features and to denoise the PET scans. Additionally, Gaussian smoothing (GS) and Bilateral filtering (BF) were evaluated. The performance of the denoising approaches was assessed based on the tumour recovery coefficient (TRC), coefficient of variance (COV; level of noise), and a qualitative assessment by two nuclear medicine physicians. The CNNs had a higher TRC and comparable or lower COV to GS and BF and was also the preferred method of the two observers for both F-FDG and Zr-rituximab PET. The CNNs improved the SNR of low-count F-FDG and Zr-rituximab PET, with almost similar or better clinical performance than the full-count PET, respectively. Additionally, the CNNs showed better performance than GS and BF.
理想情况下，应缩短F-氟脱氧葡萄糖（F-FDG）PET的采集时间并减少注入的活度。然而，这会降低信噪比（SNR），从而损害这些PET扫描的诊断价值。此外，已知锆抗体PET的信噪比很低。为了提高这些扫描的诊断价值，提出了一种卷积神经网络（CNN）去噪方法。因此，本研究的目的是开发卷积神经网络，以提高低计数F-FDG和锆抗体PET的信噪比。采集了60例原发性肺癌患者的超低计数、低计数和全计数F-FDG PET扫描以及5例非霍奇金淋巴瘤患者的全计数锆-利妥昔单抗PET扫描。构建卷积神经网络以捕捉特征并对PET扫描进行去噪。此外，还评估了高斯平滑（GS）和双边滤波（BF）。基于肿瘤恢复系数（TRC）、方差系数（COV；噪声水平）以及两名核医学医师的定性评估来评估去噪方法的性能。卷积神经网络的TRC更高，与GS和BF相比，COV相当或更低，并且也是两名观察者对F-FDG和锆-利妥昔单抗PET的首选方法。卷积神经网络提高了低计数F-FDG和锆-利妥昔单抗PET的信噪比，其临床性能分别与全计数PET几乎相似或更好。此外，卷积神经网络的性能优于GS和BF。
