Threshold calculation for segmented attenuation correction in PET with histogram fitting

Various techniques for segmented attenuation correction (SAC) have been shown to be capable of reducing transmission scan time significantly and performing accurate image quantification. The majority of well established methods are based on analyzing attenuation histograms to classify the main tissue components, which are lung and soft tissue. Methods using statistical approaches, i.e. class variances, to separate two clusters of a measured attenuation map have been shown to perform accurate attenuation correction at a scan time within a range of 2-3 min, but may fail due to peak deformations, which occur when the transmission scan time is further reduced. We implemented a new method for segmented attenuation correction with the aim of minimizing the transmission scan time and increasing the robustness for extremely short scan times using a coincidence transmission device. The implemented histogram fitting segmentation (HFS) allows accurate threshold calculation without assuming normally distributed peaks in the histogram, by adapting a suitable function to the soft tissue peak. The algorithm uses an estimated lung position (ELP) for patient contour finding and lung segmentation. Iterative reconstruction is used to generate the transmission images.