Background-based delineation of internal tumor volumes on static positron emission tomography in a phantom study

Objective[s]: Considering the fact that the standardized uptake value [SUV] of a normal lung tissue is expressed as x +/- SD, x+3xSD could be considered as the threshold value to outline the internal tumor volume [ITV] of a lung neoplasm

Methods: Three hollow models were filled with 55.0 kBq/mL fluorine18- fluorodeoxyglucose [[18]F-FDG] to represent tumors. The models were fixed to a barrel filled with 5.9 kBq/mL [18]F-FDG to characterize normal lung tissues as a phantom. The PET/CT images of the phantom were acquired at rest. Then, the barrel was moved periodically to simulate breathing while acquiring PET/CT data. Volume recovery coefficient [VRC] was applied to evaluate the accuracy of ITVs. For statistical analysis, paired t-test and analysis of variance were applied

Results: The VRCs ranged from 0.74 to 0.98 and significantly varied among gross tumor volumes for delineating ITV [P<0.01]. In two-dimensional PET scans, the motion distance did not affect VRC [P>0.05], whereas VRC decreased with increasing distance in three-dimensional PET scans [P<0.05]

Conclusion: The threshold value [x+3xSD] had the potential to delineate the ITV of cancerous tissues, surrounded by lung tissues, particularly in two-dimensional PET images

Estimation of Internal Tumor Volume：A Phantom Study Based on Semiautomatic Standardized Uptake Value of the Background / 中国医学影像学杂志

Purpose To estimate and discuss the feasibility of the threshold value of the internal tumor volume (ITV) of lung cancer based on the standardized uptake value (SUV) of the background.Materials and Methods The phantom was designed with 3 vacuous models fixed at the bottom of a barrel. The 3 vacuous models were filled with18F-FDG (55.0 kBq/ml) which represented a tumor and the barrel was filled with18F-FDG (5.9 kBq/ml), which represented the background of the lung. The barrel moved sinusoidally with distances of 10.9 mm, 21.8 mm, and 43.7 mm to simulate breathing, and then the phantom 2D PET/CT data were acquired. The SUV in the background was recorded as ±SD, and+3SD was regarded as threshold value to obtain the measured ITVs. The Dice similarity coefficient (DSC) and volume recovery coefficient (VRC) were used to evaluate the accuracy of the ITVs measured by seven methods, including that by SUV 2.5, 35% SUVmax, 41% SUVmax, a×mSUV70+b×, 0.42×(SUVmax+) and Riegel et al. computed threshold value followed the gross tumor volume, tumor-to-nontumor ratio and motion extent.Results Nine ITVs of three lung cancer models were 134.3 ml, 166.1 ml, 223.5 ml, 86.6 ml, 108.5 ml, 150.7 ml, 32.3 ml, 43.8 ml and 63.6 ml, respectively. The ITV measured by Riegel et al. showed no difference with the real ITV (t=-0.48,P>0.05). The ITVs measured by the other six methods were significantly different from the real ITVs (t=-5.11-2.76,P0.05). Moreover, they surpassed the results of the other five methods.Conclusion The threshold value (+3SD) of the SUV determined by the background has potential value in the description of the ITV of lung cancer, so as to provide reference for radiotherapy.