Photon-counting computed tomography for phantom-less quantitative measures of musculoskeletal tissues

The advent of photon-counting computed tomography (PCCT) provides new opportunities to quantitatively measure musculoskeletal tissues such as bone, muscle and adipose because of the intrinsic use of spectral imaging. We aimed to evaluate the accuracy of measuring these tissues by PCCT under a range of scan protocols and compared our results to the current standard dual-energy CT (DECT). Phantoms containing inserts ranging from 50 to 200 mg/cm3 of calcium hydroxyapatite (HA) for estimating bone mineral density (BMD), and another phantom containing inserts for muscle and adipose tissues were scanned on PCCT and DECT at 120 and 140 kVp. We created virtual monoenergetic images (VMI) at energy levels from 40 keV to 190 keV for quantitative analyses. The averaged linear attenuation of phantom inserts was compared to theoretical values calculated from standardized attenuation profiles. Material decomposition using VMIs was compared to known HA concentration inserts to determine optimal image pairs for BMD measurement, notably without the need of phantom calibration. For most VMI energy levels the attenuation error was <1% for BMD at both 120 kVp and 140 kVp by PCCT compared to errors of <2% by DECT. The linear attenuation errors were <2.5% for muscle and <3.0% for adipose and results were similar for PCCT and DECT. Generally, errors were highest for low energy VMIs. Material decomposition using VMI pairs with a low energy at 50 or 60 keV and high energy between 150 and 190 keV produced calibration phantom-free estimates of BMD with <1% error. Results were similar for PCCT and DECT at 120 and 140 kVp. PCCT provides an accurate estimate of bone, muscle and adipose attenuation, and using material decomposition, estimations of BMD can be obtained without the need of phantom calibration.