Quantitative brain imaging using the new, fast iterative histogram-mode reconstruction for the HRRT PET scanner

Specialized methods to overcome limitations in low count statistics HRRT brain imaging have been developed and implemented recently. Combination of ordinary Poisson (OP) OSEM-3D reconstruction with advanced randoms calculation has been shown to decrease quantification bias in low count statistics, but a residual bias is however seen. In the present study the extent and origin of the quantification bias in the histogram-mode iterative reconstruction (OP-OSEM-3D) scheme was further investigated. Aim of the study was to assess the minimum emission measurement signal strength, to express the minimum signal strength in noise- equivalent-counts (NEC) and to identify the origin of the residual bias. To assess the minimum count statistics anthropomorphic brain phantoms were scanned in the scanners dynamic range. Dynamic range was simulated by histogramming the coincidence data into overlapping variable length frames. True distribution was estimated from a high statistics frame and ROI analysis was applied to observe biases in the quantification. ROI analysis showed significant quantification errors in low count statistics. Moreover, regional sensitivity to bias was shown to correlate with image features. Our conclusion is that the residual quantification bias is originated in the iterative reconstruction, and that it needs to be taken into consideration in dynamic HRRT imaging when OP-OSEM-3D is used.