Effects of ROI Placement on PET-Based Assessment of Tumor Response to Therapy

Purpose. Quantitative PET response assessment during therapy requires regions of interest (ROI). Commonly, a fixed-size ROI is placed at the maximum uptake point in the pretreatment study. For intratreatment, the ROI is placed either at the maximum uptake point (ROIpeak) or at the same location as the pretreatment ROI (ROIsame). We have evaluated the effects of the ROI placement on response assessment. Methods. PET scans of 15 head and neck cancer patients were used to evaluate the effects of the two ROI methods on response assessment. Results. The average intratreatment ROIpeak uptake was 13.4% higher than the ROIsame uptake (range ？14% to 38%). The average relative change in ROIpeak uptake was 7.9% lower than ROIsame uptake (range ？5% to 36%), resulting in ambiguous tumour classification in 19% of the tumours. Conclusion. Quantitative PET response assessment using a fixed-size ROI is sensitive the ROI placement. The difference between ROIpeak and ROIsame could be substantial resulting in ambiguous response assessment. Although the fixed-size ROI is simple to implement, it is also prone to the limitations and should be used with caution. Clinical trial data are necessary to establish reliable thresholds for fixed-size ROI techniques and to evaluate their efficacy for response assessment. 1. Introduction As a powerful molecular imaging tool, positron emission tomography (PET) is increasingly being used for early assessment of tumour response to therapy [13–15]. Typically two sequential PET studies are performed and the tumor standardized uptake value (SUV) in the pre-treatment (Pre-Tx) study is compared to that of the intra-treatment (Intra-Tx) study. Response assessment using SUVs requires the selection of either a representative tumor voxel or a region of interest (ROI) for quantification. One of the simplest and most common methods of quantifying tumor uptake is to use the single voxel containing the maximum SUV ( ) [16, 17]. Unfortunately, values are highly sensitive to image noise and voxel size [18, 19], which leads to uncertainties in quantitative response assessment. Moreover, Krak, et al. [19] reported that has poor reproducibility compared to estimates of SUV made using ROI methods. As a more robust alternative, an average SUV within a small fixed size ROI has been recommended to provide adequate statistical quality in SUV measurements and to reduce uncertainties in quantitative response assessment [16]. Table 1 lists representative studies [1–12] that have used the fixed-size ROI method for early tumour response assessment. The Pre-Tx ROI is