Nonlinear holospectral imaging: scatter removal from curvilinear data in multidimensional energy space [nuclear medicine]

After a brief review of the scatter removal technique in holospectral imaging, the authors describe its implementation, taking into account the nonlinear nature of the data. Qualitative results show that structures are better defined in the processed images when the scatter removal technique is applied locally after a segmentation of the multidimensional raw data. The proportion of the variance contained in the principal axis due to this curvilinear model is highly object-dependent and ranges from almost nothing (adequacy to linear model) to 5 to 8% in the worst case