HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Quarantelli, M. Articles by Salvatore, M. Search for Related Content PubMed PubMed Citation Articles by Quarantelli, M. Articles by Salvatore, M.

Integrated Software for the Analysis of Brain PET/SPECT Studies with Partial-Volume-Effect Correction

¹ Biostructure and Bioimaging Institute, National Council for Research, Naples, Italy
² INSERM Unit 320 and E0218, Cyceron, Caen, France
³ Neurobiology Research Unit, Rigshospitalet, Copenhagen, Denmark
⁴ PET Centre, University Medical School of Debrecen, Hungary
⁵ Diagnostic Imaging, University Federico II of Naples, Italy
⁶ Department of Neurology, University of Cambridge, United Kingdom

View larger version (35K): [in a new window]   FIGURE 1. Representative slice at the level of the centra semiovalia from a 30-y-old healthy volunteer. (A) Segmented MRI used for PET simulation. GM ROIs are color coded (medial and lateral sections are pooled; left and right frontal lobe GM is green and red; left and right parietal GM is yellow and blue; WM voxels used for definition of mean WM value for MG-PVEc are gray; remaining WM is white). (B) Corresponding simulated PET slice. Also represented are the largest errors introduced in registration (with simultaneous display of segmented MRI and simulated PET with 6-pixel misalignment along the x-axis) (C); segmentation (with GM volumes of -9.7% and 10.1%) (D and E, respectively); and resolution estimate (with virtual PET as provided for feedback by the program, 8.7- and 13.2-mm FWHM) (F and G, respectively). Note that for registration and resolution estimate errors, detection of the error is possible by simple visual assessment.

View larger version (27K): [in a new window]   FIGURE 2. Same PET image as in Figure 1B, corrected for PVE according to M-PVEc (A), MG-PVEc (B), and mMG-PVEc (C) under optimal conditions. Color scale ranges from 0% to 150% GMR. A substantial residual underestimation of GM values remained after M-PVEc and was still present at the GM/WM interface when using MG-PVEc, because of the overestimation of WM, and was further decreased when using mMG-PVEc, as can be noted by the reduction of red pixels (GMR, 50%–75%) at the inner GM in C.

View larger version (33K): [in a new window]   FIGURE 3. Frequency histograms of GMR (A and B), WMR (C and D), and GCV (E and F) for uncorrected (PET) and corrected (M, MG, R, mMG) values. Data are plotted for the results of validation on the full dataset ([A, C, and E] misregistration 6 pixels; FWHM estimate ranging from -16.9% to 26.8% of the optimal value; GM volume changes resulting from segmentation error ranging from -13.7% to 14.1%) and on the dataset derived by the introduction of a reduced span of errors ([B, D, and F] misregistration 4 pixels; FWHM estimate between -12.0% and 16.3% of the optimal value; GM volume changes resulting from segmentation error between -5.3% and 5.5%).

View larger version (25K): [in a new window]   FIGURE 4. Effect of the 3 tested errors on the accuracy of the estimates of GM (left) and WM (right) ROIs before (PET) and after correction with the 4 PVEc techniques implemented.

View larger version (25K): [in a new window]   FIGURE 5. Effect of the 3 tested errors on the precision of the estimates of uncorrected GM values (PET) and after correction with the 4 PVEc techniques implemented. The coefficient of variation of the 15 GM ROIs (GCV) is plotted vs. the error introduced.