RSNA 2014 

Abstract Archives of the RSNA, 2014


CAS175

Knowledge based Iterative Reconstruction Technique for Left Ventricular Function Assessment: Comparison with Hybrid Iterative Reconstruction and Filtered Back Projection Techniques

Scientific Posters

Presented on November 30, 2014
Presented as part of CAS-SUA: Cardiac Sunday Poster Discussions

Participants

Peiling Li MD, PhD, Presenter: Nothing to Disclose
Xu Ke MD, Abstract Co-Author: Nothing to Disclose
Yan Jiang MD, Abstract Co-Author: Employee, Koninklijke Philips NV
Ying Liu, Abstract Co-Author: Employee, Koninklijke Philips NV
Na Jiang, Abstract Co-Author: Nothing to Disclose
Yu Zhao, Abstract Co-Author: Nothing to Disclose

PURPOSE

To compare the accuracy of contrast computed tomography (CT) for left ventricular function assessment using different reconstruction techniques including filtered back projection (FBP), hybrid iterative reconstruction(iDose4, Philips Healthcare) and knowledge based iterative reconstruction(IMR, Philips Healthcare) with magnetic resonance imaging (MRI).

METHOD AND MATERIALS

A total of 16 patients (48.8±9.4years, 9 male) with coronary artery disease prospectively underwent 256-MDCT and MRI (as reference standard) within 3 days. All CT images were reconstructed with FBP, iDose4 and IMR, respectively. End-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV) and ejection fraction (EF) of left ventricular (LV) were measured on FBP, iDose4, IMR and MRI images. CT image quality assessment among different reconstruction techniques was performed by 2 independent readers according to the features of structure demarcation, sharpness, artifacts and noise using a 5-point scale. Pearson’s correlation, Bland-Altman analysis, McNemar test and paired-t test were used for statistical analysis.

RESULTS

For 256-MDCT, on average effective dose was 8.87±0.83mSv. IMR significantly improved image quality compared to iDose4 and FBP (4.08±0.45, 3.66±0.56, 3.21±0.52, all p<0.01). There was no significantly overestimation and underestimation of EDV, ESV and EF by IMR, iDose4 and FBP compared to MRI (all p>0.05). SV was significantly overestimated by all three reconstruction techniques (FBP 98.3±19.1, iDose4 95.4±18.7, IMR 89.8±18.8 vs MRI 82.7±19.0 ml, all p<0.01), meanwhile, SV value of IMR was significantly higher than iDose4 and FBP (p=0.03,p=0.009,respectively). All three groups of CT images showed high intraclass correlation with MRI.

CONCLUSION

CT quantitative analysis for LV function assessment showed good accuracy compared with MRI. IMR techniques improved image quality and enabled SV values more close to those acquired by MRI.

CLINICAL RELEVANCE/APPLICATION

IMR may improve the accuracy of quantitative analysis for LV function assessment by improving image quality.

Cite This Abstract

Li, P, Ke, X, Jiang, Y, Liu, Y, Jiang, N, Zhao, Y, Knowledge based Iterative Reconstruction Technique for Left Ventricular Function Assessment: Comparison with Hybrid Iterative Reconstruction and Filtered Back Projection Techniques.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045740.html