Evaluation of conventional imaging performance in a research whole-body CT system with a photon-counting detector array

Abstract

This study evaluated the conventional imaging performance of a research whole-body photon-counting CT system and investigated its feasibility for imaging using clinically realistic levels of x-ray photon flux. This research system was built on the platform of a 2nd generation dual-source CT system: one source coupled to an energy integrating detector (EID) and the other coupled to a photon-counting detector (PCD). Phantom studies were conducted to measure CT number accuracy and uniformity for water, CT number energy dependency for high-Z materials, spatial resolution, noise, and contrast-to-noise ratio. The results from the EID and PCD subsystems were compared. The impact of high photon flux, such as pulse pile-up, was assessed by studying the noise-to-tube-current relationship using a neonate water phantom and high x-ray photon flux. Finally, clinical feasibility of the PCD subsystem was investigated using anthropomorphic phantoms, a cadaveric head, and a whole-body cadaver, which were scanned at dose levels equivalent to or higher than those used clinically. Phantom measurements demonstrated that the PCD subsystem provided comparable image quality to the EID subsystem, except that the PCD subsystem provided slightly better longitudinal spatial resolution and about 25% improvement in contrast-to-noise ratio for iodine. The impact of high photon flux was found to be negligible for the PCD subsystem: only subtle high-flux effects were noticed for tube currents higher than 300 mA in images of the neonate water phantom. Results of the anthropomorphic phantom and cadaver scans demonstrated comparable image quality between the EID and PCD subsystems. There were no noticeable ring, streaking, or cupping/capping artifacts in the PCD images. In addition, the PCD subsystem provided spectral information. Our experiments demonstrated that the research whole-body photon-counting CT system is capable of providing clinical image quality at clinically realistic levels of x-ray photon flux.

Figure 1

(A): Illustration of the research PCCT system. (B): Diagram of a macro pixel of the PCD.

Figure 2

Illustration of the macro mode (A) and the chess mode (B). The dashed curves represent typical CT x-ray spectra. TL and TH stand for ‘threshold low’ and ‘threshold high’, respectively.

Figure 3

Illustration of VOIs used for uniformity measurement.

Figure 4

(A): CT number accuracy of water from both the head and abdominal protocols. The CT numbers of water met regulatory requirements in all configurations. (B): CT numbers of bone and iodine from the abdominal protocol. The CT numbers of iodine and bone depended on the energy threshold or bin.

Figure 5

CT number uniformity measured using both the head protocol (A) and the abdominal protocol (B). All data met regulatory requirements.

Figure 6

Measurements of the spatial resolution of the research PCCT system using the macro mode with pitch 0.5 and reconstruction kernel D30. (A): In-plane spatial resolution. (B): Longitudinal spatial resolution corresponding to the smallest slice thicknesses, i.e. 0.6 mm for the EID subsystem and 0.5 mm for the PCD subsystem. (C): Longitudinal spatial resolution corresponding to slice thickness of 1 mm. (D): Longitudinal spatial resolution corresponding to slice thickness of 2 mm. MTF: modulation transfer function. SSP: slice sensitivity profile.

Figure 7

In-plane noise power spectra.

Figure 8

Assessment of the impact of high photon flux for the EID subsystem. (A): Relationship between CT number of water and tube current. (B): Relationship between noise and tube current.

Figure 9

Assessment of the impact of high photon flux for the macro mode of the PCD subsystem. (A): Relationship between CT number of water and tube current. (B): Relationship between noise and tube current. (C): Relationship between the ratio of the noise in bin 1 to that in bin 2 and tube current.

Figure 10

Reconstructions of the CIRS phantom (mimicking a 15 year old human being) with a bone-mimicking vial (left) and an iodine vial (right). Display window: W/L = 700/100 HU. Slice thickness: 2 mm.

Figure 11

Reconstructions of the CIRS phantom (mimicking a large adult) with a bone-mimicking vial (left) and an iodine vial (right). Display window: W/L = 200/0 HU. Slice thickness: 2 mm.

Figure 12

Images of the cadaver head from the PCD chess mode. Left: 220 mAs; right: 550 mAs. Rotation time: 1 s. Arrows demonstrate beam-hardening artifacts from adjacent dense bone, which are decreased in those images using predominantly higher energy photons. Display window: W/L = 300/− 20 HU.

Figure 13

Images of the whole-body cadaver obtained from the PCD macro mode. From A to C: threshold low, bin 1, and bin 2. Arrows demonstrate beam-hardening artifacts from adjacent dense bone, which are decreased in the bin 2 (high energy) images. Display window: W/L = 300/− 20 HU.