透過您的圖書館登入
IP:3.138.33.178
  • 期刊

Comparison of Post-Therapy Y-90 PET Images Using Different Reconstruction Algorithms

使用不同影像重組演算法於治療後釔90正子造影的比較

摘要


Background: Selective internal radiation therapy (SIRT) with yttrium-90 (Y-90) loaded microspheres has been widely used as a locoregional therapy for liver tumors. After SIRT, Y-90 positron emission tomography (PET) is performed to verify microspheres delivery and detect any significant extrahepatic activity. However, the image quality of Y-90 PET is usually poor because of low count statistics and high noise. Q.Clear (GE Healthcare, Milwaukee, WI, USA) is a new Bayesian penalized likelihood reconstruction algorithm that suppresses noise by a penalization factor (termed β). The aim of this study is to compare Q.Clear with conventional reconstruction methods and assess the optimal β value of Q.Clear for visual and quantitative evaluation of Y-90 PET in patients after SIRT. Methods: From July 2019 to August 2020, 13 patients with 30 liver tumors were enrolled. A total of 12 reconstruction methods were used, including ordered subset expectation maximization (OSEM), OSEM with point spread function (PSF), OSEM with time of flight (TOF), OSEM with PSF and TOF, Q.Clear with β values of 500, 1,000, 2,000, 3,000, 4,000, 5,000, 6,000, and 8,000. Quantitative image analyses of the contrast, noise, and signal to noise ratio (SNR) were calculated for comparison. Visual scoring of image quality was also performed. Results: Applying PSF resulted in significantly higher contrast and SNR (both p < 0.001). There was no significant difference in contrast, noise, or SNR between images with and without TOF. In Q.Clear, increasing β value resulted in reduced noise, thereby improving SNR. Compared to OSEM with PSF and TOF, Q.Clear with β values above 2,000 had significantly lower noise (all p < 0.001) and higher SNR (all p < 0.001). All Y-90 PET images had acceptable image quality using Q.Clear with β values of 4,000, 5,000, and 6,000. Conclusions: Our results indicate that Q.Clear with β values between 4,000 and 6,000 had lower noise, higher SNR, and image quality more acceptable than conventional OSEM-based reconstruction methods in post-SIRT Y-90 PET.

關鍵字

image quality Q.Clear Y-90 PET

並列摘要


背景:釔90(Yttrium-90, Y-90)微球體選擇性體內放射治療(selective internal radiation therapy, SIRT)已被廣泛使用於肝臟腫瘤之局部治療。SIRT後,進行正子造影以驗證釔90微球體在肝臟腫瘤的分布,並偵測肝外分布。然而,由於釔90低計數與高雜訊,影像通常很差。Q.Clear是奇異醫療(GE Healthcare)一種新的貝氏懲罰概似(Bayesian penalized likelihood, BPL)重組演算法,透過懲罰因數(β)抑制雜訊。本研究的目的是將Q.Clear與傳統重組方法進行比較,並評估Q.Clear在釔90正子造影影像於目視判讀和定量分析的最佳β 值。方法:從2019年7月至2020年8月,納入13位患者,共30個肝臟腫瘤病灶。使用12種正子造影影像重組方法,包括:序列子集之期望值最大化(ordered subset expectation maximization, OSEM)、OSEM併用點擴散函式(point spread function, PSF)、OSEM併用飛時測距(time of flight, TOF)、OSEM併用PSF與TOF、Q.Clear之β值分別為500、1,000、2,000、3,000、4,000、5,000、6,000和8,000。定量分析比較釔90正子造影影像之對比度、雜訊和訊雜比(signal to noise ratio, SNR),並且進行影像品質目視評分。結果:釔90正子造影影像重組運用OSEM併用PSF,明顯產生較高的對比度和SNR(p值均小於0.001)。OSEM有無併用TOF,在對比度、雜訊或SNR方面沒有顯著差異。在Q.Clear中,增加β值可降低雜訊,從而改善SNR。與OSEM併用PSF與TOF相比,當Q.Clear的β值在2,000以上,具有顯著較低的雜訊(p值均小於0.001)和較高的SNR(p值均小於0.001)。Q.Clear的β值為4,000、5,000和6,000時,所有釔90正子造影影像都有可接受的影像品質。結論:研究結果顯示,與傳統基於OSEM的影像重組方式比較,當Q.Clear的β值介於4,000至6,000之間,可在SIRT後病人得到較低雜訊、較高SNR和更能接受的釔90正子造影影像品質。

並列關鍵字

影像品質 Q.Clear 釔90正子造影

延伸閱讀