A New Frontier in Temporal Bone Imaging: Photon-Counting Detector CT Demonstrates Superior Visualization of Critical Anatomic Structures at Reduced Radiation Dose

Abstract

Background and purpose: Photon-counting detector CT is a new technology with a limiting spatial resolution of ≤150 μm. In vivo comparisons between photon-counting detector CT and conventional energy-integrating detector CT are needed to determine the clinical impact of photon counting-detector CT in temporal bone imaging.

Materials and methods: Prospectively recruited patients underwent temporal bone CT examinations on an investigational photon-counting detector CT system after clinically indicated temporal bone energy-integrating detector CT. Photon-counting detector CT images were obtained at an average 31% lower dose compared with those obtained on the energy-integrating detector CT scanner. Reconstructed images were evaluated in axial, coronal, and Pöschl planes using the smallest available section thickness on each system (0.4 mm on energy-integrating detector CT; 0.2 mm on photon-counting detector CT). Two blinded neuroradiologists compared images side-by-side and scored them using a 5-point Likert scale. A post hoc reassignment of readers' scores was performed so that the scores reflected photon-counting detector CT performance relative to energy-integrating detector CT.

Results: Thirteen patients were enrolled, resulting in 26 image sets (left and right sides). The average patient age was 63.6 [SD, 13.4] years; 7 were women. Images from the photon-counting detector CT scanner were significantly preferred by the readers in all reconstructed planes (P < .001). Photon-counting detector CT was rated superior for the evaluation of all individual anatomic structures, with the oval window (4.79) and incudostapedial joint (4.75) receiving the highest scores on a Likert scale of 1-5.

Conclusions: Temporal bone CT images obtained on a photon-counting detector CT scanner were rated as having superior spatial resolution and better critical structure visualization than those obtained on a conventional energy-integrating detector scanner, even with a substantial dose reduction.

© 2022 by American Journal of Neuroradiology.

Figures

FIG 1.

Readers’ score distribution for spatial resolution and visualization of critical anatomic structures in different reformatted planes (A) and mean readers’ scores for individual anatomic structures and overall image quality (B). All scores were based on a 5-point Likert scale, comparing PCD-CT with EID-CT: 1 = inferior resolution with degraded visualization, 2 = slightly inferior resolution without affecting visualization, 3 = equivalent resolution and visualization, 4 = slightly superior resolution without affecting visualization, and 5 = superior spatial resolution with improved visualization.

FIG 2.

Pöschl reformatted images in a patient with superior semicircular canal dehiscence, shown on EID-CT (left) and PCD-CT (right) images. The PCD-CT image (B) clearly demonstrates 2 discrete regions of dehiscence (curved arrows). These regions are also identifiable on conventional EID-CT (A), though the intact adjacent bone is less well-visualized. The integrity of the roof of the superior semicircular canal was not formally evaluated in the readers’ study but is shown for illustrative purposes.

FIG 3.

Ossicular anatomy, shown on conventional EID-CT (upper row) and PCD-CT (lower row). Reformatted images along the plane of the tensor tympani (A and E) demonstrate the tensor tympani (TT) extending to the upper handle of the malleus (HA); the lateral process (L) of the malleus is also clearly visible. An image reformatted along the long plane of the malleus (B and F) shows its handle (HA), lateral process (L), neck (N), and head (H). An image reformatted along the length of the stapes (C and G) clearly demonstrates the suprastructure (SS) and both crura. A “molar tooth” reformatted image (D and H) shows the HA of the malleus, as well as the body (B) and long process (LP) of the incus. These additional reformatted images were generated by a nonreviewer radiologist to demonstrate certain anatomic features but were not used by the readers to score image quality.

FIG 4.

The incudostapedial joint (arrows), shown on EID-CT (left) and PCD-CT (right) images. The joint was one of several anatomic structures specifically graded using a 5-point Likert score, with higher scores favoring the quality of the PCD-CT images. The images reformatted in this plane were generated by a nonreviewer radiologist to demonstrate certain anatomic features but were not used by the readers to score image quality.

FIG 5.

A stapes piston prosthesis shown on EID-CT (upper row) and PCD-CT (lower row) images. The prothesis is shown on oblique axial (A and C) and oblique coronal (B and D) reformats along the plane of the piston. In both planes, the stapes prosthesis was better delineated on PCD-CT due to reduced partial volume averaging from 0.2 mm sections. These additional reformatted images were generated by a nonreviewer radiologist to demonstrate certain anatomic features but were not used by the readers to score image quality.

FIG 6.

Images reformatted along the long axis of the stapes in a patient with fenestral otosclerosis shown on EID-CT (left) and conventional PCD-CT (right) images. The arrow points to the insertion of the anterior crus of the stapes into the region involved by otosclerosis. These additional reformatted images were generated by a nonreviewer radiologist to demonstrate certain anatomic features but were not used by the readers to score image quality.