TY - JOUR T1 - Effect of time-of-flight on lesion detection in PET/CT: Is it simply an effective sensitivity gain over non-time-of-flight? JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1875 LP - 1875 VL - 57 IS - supplement 2 AU - Chuanyong Bai AU - Andriy Andreyev AU - Zhiqiang Hu AU - Jun Zhang Y1 - 2016/05/01 UR - http://jnm.snmjournals.org/content/57/supplement_2/1875.abstract N2 - 1875Objectives Time-of-flight (TOF) demonstrates an effective sensitivity gain over non-TOF. It is argued that if TOF only has a sensitivity gain then acquiring more counts for non-TOF can achieve the same performance as TOF. This work is to study if TOF provides more than just a sensitivity gain over non-TOF for lesion detection. Methods We simulated TOF and non-TOF data of three lesions (10 mm in diameter) in torso: two in lungs (LL1, activity 2.0; LL2, activity 1.5) and one in mediastinum (ML, activity 2.5). The background activities were 0.5 for lung and 1.0 for soft tissue. We compared TOF and non-TOF at different situations: (a) with no data degradation and (b)-(d) with different levels of degradation. In (a), both emission data and the attenuation map were noise-free. In (b), the attenuation map was noise-free, the TOF emission data had 0.4M events per 2 mm slice and non-TOF had 0.4M, 0.6M, 0.8M and 1.2M events. In (c), the emission data was noise-free, but the attenuation map had noise level from that of low-dose CT (LDCT) to ultra LDCT (uLDCT) and to the level similar to transmission-based attenuation maps prior to PET/CT. In (d), the emission data had 0.4M events for TOF and 1.2M events for non-TOF while the attenuation maps had LDCT and uLDCT noise levels. For (b) and (d), 32 different noise realizations of each of the emission data were generated and reconstructed. Between (b) and (d), the same noise realizations were used for 0.4M TOF and 1.2M non-TOF. Lesion visibility was scored from 0 to 3, with 0 not visible, 1 marginally visible, 2 visible, and 3 definitely visible. Two-tailed t-tests were performed between TOF and non-TOF visibility scores for statistical significance assessment. Results (a) With no noise in both emission and attenuation maps, non-TOF and TOF images had the same lesion visibility and quantitation and both were the same as the truth. (b) With no noise in attenuation maps but noise in emission data, 320 ps TOF showed an effective sensitivity gain with a factor between 2 and 3. Lesion visibility was comparable in 0.4M TOF and 0.8M non-TOF with no statistical difference (p > 0.05). Lesion LL2 had significantly lower visibility in 0.4M TOF than in 1.2M non-TOF (0.81 vs 1.59, p=0.00) while LL1 and ML had similar visibility with no statistical difference. (c) With no noise in emission data but noise in attenuation maps, at the noise level similar to transmission-based attenuation maps, LL2 and ML could not be identified in non-TOF images but could in TOF images; at the LDCT noise level, all the three lesions were clearly visible. This suggested that when data degradation level was high, some lesions that could be detected using TOF might not be detected using non-TOF, no matter how many non-TOF counts were acquired. (d) The effective sensitivity gain of TOF over non-TOF increased with increased data degradation level. For attenuation map at LDCT noise level, 0.4M TOF images and 1.2M non-TOF images showed comparable visibility for all the lesions; at uLDCT noise level, 0.4M TOF images had significantly higher visibility for the two lung lesions than 1.2M non-TOF images (LL1: 2.38 vs 2.06, p=0.02; LL2: 0.75 vs 0.25, p=0.00) while the mediastinum lesion visibility showed no significant difference (Figure). The effective sensitivity gain of TOF was a factor of 3 or higher, which was greater than that in (b). Lung lesion visibility in non-TOF images reduced significantly using attenuation maps with uLDCT noise level as compared to noise-free ones (LL1: 2.06 vs 2.59, p=0.02; LL2: 0.25 vs 1.59, p=0.00), yet the differences were insignificant in TOF images (p>0.05), indicating non-TOF was more susceptible to attenuation map degradation than TOF. Conclusions TOF showed effective sensitivity gain over non-TOF. However, the gain was greater when the level of data degradation was higher. At certain degradation level, some lesions that were visible in TOF images could no longer be detected in non-TOF images regardless of the acquired number of events. $$graphic_5B7322DD-39DC-41D6-94A2-EE8CB4B97860$$ ER -