Photon-counting CT using multi-material decomposition algorithm enables fat quantification in the presence of iron deposits

Abstract

Purpose: A new generation of CT detectors were recently developed with the ability to measure individual photon's energy and thus provide spectral information. The aim of this work was to assess the performance of simultaneous fat and iron quantification using a clinical photon-counting CT (PCCT) and its comparison to dual-energy CT (DECT), MRS and MRI at 3 T.

Methods: Two 3D printed cylindrical phantoms with 32 samples (n = 12 fat fractions between 0 % and 100 %, n = 20 with mixtures of fat and iron) were scanned with PCCT and DECT scanners for comparison. A three-material decomposition approach was used to estimate the volume fractions of fat (FF), iron and soft tissue. The same phantoms were examined by MRI (6-echo DIXON, a.k.a. Q-DIXON) and MRS (multi-echo STEAM, a.k.a. HISTO) at 3 T for comparison.

Results: PCCT, DECT, MRI and MRS computed FFs showed correlation with reference fat fraction values in samples with no iron (r > 0.98). PCCT decomposition showed slightly weaker correlation with FFref in samples with added iron (r = 0.586) compared to MRI (r = 0.673) and MRS (r = 0.716) methods. On the other hand, it showed no systematic over- or underestimation. Surprisingly, DECT decomposition-derived FF showed strongest correlation (r = 0.758) in these samples, however systematic overestimation was observed. FF values computed by three-material PCCT decomposition, DECT decomposition, MRI and MRS were unaffected by iron concentration.

Conclusions: This in-vitro study shows for the first time that photon-counting computed tomography may be used for quantification of fat content in the presence of iron deposits.

Keywords: Fat fraction; Iron; MRI; Photon-counting CT.