Lifang Pang1,2,3, Wentao Zhu4, Yun Dong5, Yang Lv5, Hongcheng Shi6,7,8. 1. Department of Nuclear Medicine, Zhongshan Hospital, Fudan University, No. 180, Fenglin Road, Shanghai, 200032, People's Republic of China. 2. Shanghai Institute of Medical Imaging, Shanghai, 200032, China. 3. Institute of Nuclear Medicine, Fudan University, Shanghai, 200032, China. 4. UIH America, Inc, 9230 Kirby Dr, Suite 600, Houston, TX, 77054, USA. 5. Shanghai United Imaging Healthcare Co., Ltd, 2258 Chengbei Rd, Jiading District, Shanghai, 201807, China. 6. Department of Nuclear Medicine, Zhongshan Hospital, Fudan University, No. 180, Fenglin Road, Shanghai, 200032, People's Republic of China. shi.hongcheng@zs-hospital.sh.cn. 7. Shanghai Institute of Medical Imaging, Shanghai, 200032, China. shi.hongcheng@zs-hospital.sh.cn. 8. Institute of Nuclear Medicine, Fudan University, Shanghai, 200032, China. shi.hongcheng@zs-hospital.sh.cn.
Abstract
PURPOSE: Delayed positron emission tomography (PET) imaging may improve sensitivity and specificity in lesion detection. We proposed a PET data-driven method to estimate the attenuation map (AM) for the delayed scan without an additional x-ray computed tomography (CT). PROCEDURES: An emission-attenuation-scatter joint estimation framework was developed. Several practical issues for clinical datasets were addressed. Particularly, the unknown scatter correction was incorporated in the joint estimation algorithm. The scaling problem was solved using prior information from the early CT scan. Fourteen patient datasets were added to evaluate the method. These patients went through two separate PET/CT scans. The delayed CT-based AM served as ground truth for the delayed scan. Standard uptake values (SUVmean and SUVmax) of lesion and normal tissue regions of interests (ROIs) in the early and delayed phase and the respective %DSUV (percentage change of SUVmean at two different time points) were analyzed, all with estimated and the true AM. Three radiologists participated in lesion detection tasks with images reconstructed with both AMs and rated scores for detectability. RESULTS: The mean relative difference of SUVmean in lesion and normal liver tissue were 3.30 and 6.69 %. The average lesion-to-background contrast (detectability) with delayed PET images using CT AM was 60 % higher than that of the earlier PET image, and was 64 % higher when using the data-based AM. %DSUV for lesions and liver backgrounds with CT-based AM were - 0.058 ± 0.25 and - 0.33 ± 0.08 while with data-based AM were - 0.00 ± 0.26 and - 0.28 ± 0.08. Only slight significance difference was found between using CT-based AM and using the data-based AM reconstruction delay phase on %DSUV of lesion. The scores associated with the two AMs matched well consistently. CONCLUSIONS: Our method may be used in delayed PET imaging, which allows no secondary CT radiation in delayed phase. The quantitative analysis for lesion detection purpose could be ensured.
PURPOSE: Delayed positron emission tomography (PET) imaging may improve sensitivity and specificity in lesion detection. We proposed a PET data-driven method to estimate the attenuation map (AM) for the delayed scan without an additional x-ray computed tomography (CT). PROCEDURES: An emission-attenuation-scatter joint estimation framework was developed. Several practical issues for clinical datasets were addressed. Particularly, the unknown scatter correction was incorporated in the joint estimation algorithm. The scaling problem was solved using prior information from the early CT scan. Fourteen patient datasets were added to evaluate the method. These patients went through two separate PET/CT scans. The delayed CT-based AM served as ground truth for the delayed scan. Standard uptake values (SUVmean and SUVmax) of lesion and normal tissue regions of interests (ROIs) in the early and delayed phase and the respective %DSUV (percentage change of SUVmean at two different time points) were analyzed, all with estimated and the true AM. Three radiologists participated in lesion detection tasks with images reconstructed with both AMs and rated scores for detectability. RESULTS: The mean relative difference of SUVmean in lesion and normal liver tissue were 3.30 and 6.69 %. The average lesion-to-background contrast (detectability) with delayed PET images using CT AM was 60 % higher than that of the earlier PET image, and was 64 % higher when using the data-based AM. %DSUV for lesions and liver backgrounds with CT-based AM were - 0.058 ± 0.25 and - 0.33 ± 0.08 while with data-based AM were - 0.00 ± 0.26 and - 0.28 ± 0.08. Only slight significance difference was found between using CT-based AM and using the data-based AM reconstruction delay phase on %DSUV of lesion. The scores associated with the two AMs matched well consistently. CONCLUSIONS: Our method may be used in delayed PET imaging, which allows no secondary CT radiation in delayed phase. The quantitative analysis for lesion detection purpose could be ensured.
Authors: Wouter V Vogel; Jorn A van Dalen; Bas Wiering; Henkjan Huisman; Frans H M Corstens; Theo J M Ruers; Wim J G Oyen Journal: J Nucl Med Date: 2007-05-15 Impact factor: 10.057
Authors: Johan Nuyts; Girish Bal; Frank Kehren; Matthias Fenchel; Christian Michel; Charles Watson Journal: IEEE Trans Med Imaging Date: 2012-09-21 Impact factor: 10.048
Authors: Gerald Antoch; Jörg Stattaus; Andre T Nemat; Simone Marnitz; Thomas Beyer; Hilmar Kuehl; Andreas Bockisch; Jörg F Debatin; Lutz S Freudenberg Journal: Radiology Date: 2003-09-25 Impact factor: 11.105