Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||Semi-automatic estimation of image-derived input function in mouse FDG-PET studies using one blood sample|
|Publisher:||Society of Nuclear Medicine|
|Source:||The Journal of nuclear medicine, 2010, v. 51, no. suppl. 2, 514 How to cite?|
|Journal:||The Journal of nuclear medicine|
|Abstract:||Objectives: This study aims to develop a hybrid clustering method (HCM) to semi-automatically define the left ventricle (LV) and myocardium ROIs in the estimation of image-derived input function (IDIF) for mouse FDG-PET studies with 1 blood sample.|
Methods: The HCM consists of probabilistic curve-aligned clustering (PCC) for the early time-activity curves (eTACs)(t<1min) and K-means clustering (KC) for the last frame (t=~60min), which is used to get the LV and myocardium ROIs in one predefined heart ROI. In PCC, the eTACs for pixels in the heart ROI were modeled as spline regression mixtures and divided into 5 clusters by EM algorithm, while 3 clusters were obtained in KC. The myocardium ROI was defined from KC with highest update cluster. Combining the results of PCC and KC, the pixels with the highest peak of eTAC and median activities in the last frame were classified into the LV ROI. Then the mean TACs extracted from these two ROIs were applied to get spillover-corrected IDIF by Cb=(CLV-FbmCmyo)/(1-FbmFmb) with 1 blood sample (~45min) with the corrections of delay, dispersion and uptake of red blood cells. The estimated IDIF was evaluated by the gold standard derived from a serial blood samples for 12 normal mice studies.
Results: Across 12 studies, the correlation coefficients of the mean TACs from manual and automatic ROIs were found to average 0.94±0.06 and 0.98±0.008 for LV and myocardium respectively. The area under curve (AUC) and the Patlak Ki calculated by the estimated IDIF had high linear relation with those obtained by gold standard. The table lists the results.
Conclusions: The semi-automatic estimation of IDIF based on HCM can provide accurate estimation for input function and tissue uptake rates in mouse FDG-PET studies.
|Appears in Collections:||Journal/Magazine Article|
Show full item record
Checked on May 28, 2017
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.