Aim/Introduction: In neurodegenerative dementias the study of brain metabolism, provided by [18]F-FDG PET, can be integrated with brain perfusion by means of pseudo-Continuous Arterial Spin Labelling MR sequences (pCASL). Aim of this study is to validate, on PET images relying on normative data, a region-based pipeline constructed to be independently applied to PET and pCASL data to jointly analyse metabolism and perfusion. Materials and Methods: Thirty-six MCI patients and 107 healthy-controls PET images were considered. Pre-processing included MNI normalization, Grey Matter segmentation through mean Tissue Probability Map, partial volume error (PVE) correction and smoothing. Sixteen ROIs were derived from AAL3 atlas and SUV ratios (SUVr) normalized on cerebellum activity were extracted. For each ROI, SUVr mean and standard deviation in healthy controls were computed. Patient SUVr values falling outside (σ, 1.5σ, 2σ) normality range were considered hypometabolic. Results were compared to: 1) visual analysis (supported by CortexID-Suite); 2) two different SPM statistical analyses (SPM-A: voxel-size 2x2x2 mm3, smoothing Gaussian Kernel FWHM 8x8x8 mm3; SPM-B: voxel-size 1x1x1 mm3¸ PVE correction). This analysis was conducted for each ROI separately. Agreement among methods was assessed with accuracy, sensitivity, specificity and Cohen’s к. Results: By using visual analysis as reference, the (σ, 1.5σ, 2σ) normality ranges obtained, on average on the 16 ROIs, accuracy (76, 78, 76)%, sensitivity (84, 67, 47)%, specificity (69, 85, 94)%, and к (49, 60, 60) %, respectively. With SPM-A as reference, accuracy (71, 79, 83) %, sensitivity (89, 75, 59)%, specificity (63, 80, 93)%, and к (42, 58, 67)% were obtained. With SPB-B as reference, accuracy (76, 77, 73)%, sensitivity (79, 64, 46) %, specificity (71, 88, 97)%, and к (51, 54, 47)% were obtained. SPM-A (SPM-B) compared to visual analysis obtained 79(75)% accuracy, 60(76)% sensitivity, 92(75)% specificity, 57(51)% к. SPM-B compared to SPM-A obtained 79% accuracy, 94% sensitivity, 72% specificity, 58% к. Conclusion: The proposed region-based analysis pipeline with the 1.5σ normality range showed a good agreement with reference methods, in line with intra reference methods agreement levels. It can be therefore considered as a promising tool for future PET-pCASL joint analyses. References: Festari C. et al., Alzheimer’s Dement, 2022; Yan L. et al., NeuroImage Clin, 2017; Musiek E.S. et al., Alzheimer’s Dement., 2013; Guedj E. et al., J. Nucl. Med. Mol. Imaging, 2022; Perani D. et al., NeuroImage Clin, 2014; Caminiti S. P. et al., Eur. J. Nucl. Med. Mol. Imaging, 2021

Cerina, V., DE BERNARDI, E., Crivellaro, C., Morzenti, S., Pozzi, F., Bigiogera, V., et al. (2023). A ROI-based quantitative pipeline for 18F-FDG PET metabolism and pCASL perfusion joint analysis: validation on 18F-FDG PET data. Intervento presentato a: EANM'23 - European Association of Nuclear Medicine, Vienna, Austria [10.1007/s00259-023-06333-x].

A ROI-based quantitative pipeline for 18F-FDG PET metabolism and pCASL perfusion joint analysis: validation on 18F-FDG PET data

Valeria Cerina;Elisabetta De Bernardi;Cinzia Crivellaro;Sabrina Morzenti;Federico E. Pozzi;Vittorio Bigiogera;Lorenzo Jonghi-Lavarini;Rosa M. Moresco;Gianpaolo Basso
2023

Abstract

Aim/Introduction: In neurodegenerative dementias the study of brain metabolism, provided by [18]F-FDG PET, can be integrated with brain perfusion by means of pseudo-Continuous Arterial Spin Labelling MR sequences (pCASL). Aim of this study is to validate, on PET images relying on normative data, a region-based pipeline constructed to be independently applied to PET and pCASL data to jointly analyse metabolism and perfusion. Materials and Methods: Thirty-six MCI patients and 107 healthy-controls PET images were considered. Pre-processing included MNI normalization, Grey Matter segmentation through mean Tissue Probability Map, partial volume error (PVE) correction and smoothing. Sixteen ROIs were derived from AAL3 atlas and SUV ratios (SUVr) normalized on cerebellum activity were extracted. For each ROI, SUVr mean and standard deviation in healthy controls were computed. Patient SUVr values falling outside (σ, 1.5σ, 2σ) normality range were considered hypometabolic. Results were compared to: 1) visual analysis (supported by CortexID-Suite); 2) two different SPM statistical analyses (SPM-A: voxel-size 2x2x2 mm3, smoothing Gaussian Kernel FWHM 8x8x8 mm3; SPM-B: voxel-size 1x1x1 mm3¸ PVE correction). This analysis was conducted for each ROI separately. Agreement among methods was assessed with accuracy, sensitivity, specificity and Cohen’s к. Results: By using visual analysis as reference, the (σ, 1.5σ, 2σ) normality ranges obtained, on average on the 16 ROIs, accuracy (76, 78, 76)%, sensitivity (84, 67, 47)%, specificity (69, 85, 94)%, and к (49, 60, 60) %, respectively. With SPM-A as reference, accuracy (71, 79, 83) %, sensitivity (89, 75, 59)%, specificity (63, 80, 93)%, and к (42, 58, 67)% were obtained. With SPB-B as reference, accuracy (76, 77, 73)%, sensitivity (79, 64, 46) %, specificity (71, 88, 97)%, and к (51, 54, 47)% were obtained. SPM-A (SPM-B) compared to visual analysis obtained 79(75)% accuracy, 60(76)% sensitivity, 92(75)% specificity, 57(51)% к. SPM-B compared to SPM-A obtained 79% accuracy, 94% sensitivity, 72% specificity, 58% к. Conclusion: The proposed region-based analysis pipeline with the 1.5σ normality range showed a good agreement with reference methods, in line with intra reference methods agreement levels. It can be therefore considered as a promising tool for future PET-pCASL joint analyses. References: Festari C. et al., Alzheimer’s Dement, 2022; Yan L. et al., NeuroImage Clin, 2017; Musiek E.S. et al., Alzheimer’s Dement., 2013; Guedj E. et al., J. Nucl. Med. Mol. Imaging, 2022; Perani D. et al., NeuroImage Clin, 2014; Caminiti S. P. et al., Eur. J. Nucl. Med. Mol. Imaging, 2021
abstract + poster
Dementia; Mild Cognitive Impairment; Brain 18F-FDG PET; Arterial Spin Labeling; ROI-based 41 quantitative analysis; Statistical Parametric Mapping (SPM)
English
EANM'23 - European Association of Nuclear Medicine
2023
2023
50
702
702
EP-0685
none
Cerina, V., DE BERNARDI, E., Crivellaro, C., Morzenti, S., Pozzi, F., Bigiogera, V., et al. (2023). A ROI-based quantitative pipeline for 18F-FDG PET metabolism and pCASL perfusion joint analysis: validation on 18F-FDG PET data. Intervento presentato a: EANM'23 - European Association of Nuclear Medicine, Vienna, Austria [10.1007/s00259-023-06333-x].
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/486883
Citazioni
  • Scopus 6
  • ???jsp.display-item.citation.isi??? 3
Social impact