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

This Article Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Bond, S. Articles by Declerck, J. PubMed Articles by Bond, S. Articles by Declerck, J.

Automatic registration of cardiac SPECT-CT for attenuation correction

¹ Siemens Molecular Imaging, Oxford, United Kingdom; ² Siemens Molecular Imaging, Hoffman Estates, Illinois; ³ University of Michigan, Ann Arbor, Michigan

1651

Objectives: Misalignment of images in cardiac SPECT-CT imaging may lead to erroneous Attenuation Correction (AC) and mis-diagnosis. Such misalignment is usually corrected by hand prior to clinical assessment. The aim of the study is to automatically align CT to SPECT prior to AC in order to reduce operator variability.

Methods: 60 normal patients have had SPECT-CT, consisting of a SPECT at stress, and 3 CTs, at BreathHold Inspiration (BHI), BreathHold Expiration (BHE) and Shallow Free Breathing (SFB). Each CT has been manually aligned to the SPECT by an expert, using translations only, and translation and rotation, if translation alone did not provide satisfactory alignment. An automatic registration, using mutual information, was used to align the datasets, first using translation only, and then deformable. The registrations were assessed visually, and translations were compared to manual alignments. In addition, we took a subset of 30 cases and performed 17-segment polar plot analysis using 4D M-SPECT on each registered image: for each patient, consistency of scores across the images was measured by counting the number of segments with different scores.

Results: The translation gave a visually acceptable result for 95% of cases for both translation and deformable (96% BHE, 90% BHI, 98% SFB). Comparison with the manual registrations showed a mean difference of 12mm. Consistency across the 3 AC images showed that there were differences in 29% of segments without registration, 11% with manual, 12% with automatic translation and 10% with automatic deformable.

Conclusions: Our analysis has shown that automatic registration is a viable option for the task of aligning CT and SPECT for AC, with a consistency comparable to that of using manual alignment.

This Article Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Bond, S. Articles by Declerck, J. PubMed Articles by Bond, S. Articles by Declerck, J.