Improved visualization of malpositioned middle ear prostheses using photon-counting detector CT: a cadaver study

Abstract

Background

Evaluation of middle ear prostheses requires high-resolution imaging to assess position and structural integrity.

Purpose

To assess the diagnostic performance of photon-counting-detector (PCD)-CT for middle ear prosthesis imaging in cadaveric specimens.

Materials and Methods

Nine different middle ear prostheses were implanted in 5 fresh-frozen whole-head specimens, both in correct anatomic position and displaced to mimic malposition. Specimens were scanned on PCD-CT (NAEOTOM Alpha, Siemens) and energy-integrating-detector (EID)-CT (SOMATOM Force, Siemens), reconstructed at 0.2 mm and 0.4 mm, respectively. Eighteen cases (9 prostheses, 2 positions) per scanner were evaluated by 2 neuroradiologists (36 readings), blinded to scanner and position, who scored prosthesis position, reader confidence, and overall image quality. Agreement with the reference standard and between readers was compared.

Results

PCD-CT showed higher diagnostic confidence compared with EID-CT, while accuracy and inter-reader agreement did not demonstrate significant differences between scanners. Correct readings, PCD-CT versus EID-CT: lateral contact, 31/36 (86.1%; 95% CI, 58.3-94.4) vs 29/36 (80.6%; 95% CI, 55.6-91.7), P = .500; medial contact, 30/36 (83.3%; 95% CI, 55.6-94.4) vs 26/36 (72.2%; 95% CI, 50.0-83.3), P = .250. PCD-CT identified all cases of complete contact loss (score 3), while EID-CT missed 1 reading per contact. Inter-reader agreement may be higher with PCD-CT vs EID-CT (lateral: κ_w: 0.804 [95% CI, 0.578-0.941] vs 0.571 [95% CI, 0.302-0.778], P = .250; medial: 0.684 [95% CI, 0.191-0.934] vs 0.481 [95% CI, 0.024-0.835], P = .250). Reader confidence was significantly higher with PCD-CT; mean paired differences for Readers 1 and 2 were 9.4 (95% CI, 1.8-17.1; Holm-adjusted P = .019) and 23.1 points (95% CI, 13.4-32.7; P < .001) for lateral contact, and 16.4 (95% CI, 10.0-22.8; P < .001) and 17.8 points (95% CI, 6.8-28.8; P = .007) for medial contact. Readers preferred PCD-CT image quality in 16/18 (88.9%; 95% CI, 65.3-98.6) and 17/18 cases (94.4%; 95% CI, 72.7-99.9), respectively.

Conclusion

PCD-CT was associated with significantly higher reader confidence and preferred image quality, without significant differences in diagnostic accuracy or inter-reader agreement. These findings suggest that PCD-CT is a promising modality for postoperative middle ear evaluations and warrants further investigations.

Abbreviations 

PCD = Photon-counting-detector; EID = Energy-integrating-detector; FOV = Field-of-view; CTDIvol = Volume CT dose index; PORP = Partial ossicular replacement prosthesis; TORP = Total ossicular replacement prosthesis; κ_w = Linearly weighted kappa

Summary 

In a small sample of cadaveric specimens, photon-counting detector CT compared favorably to energy-integrating detector CT in detection of malpositioned middle ear prostheses, diagnostic agreement, and reader confidence.

Key Results 

• Assessment of prostheses contacts in 9 different prostheses implanted in 5 cadaveric heads demonstrated agreement with the reference standard: 86.1% for PCD-CT vs 80.6% for EID-CT for lateral contact (P = .500), and 83.3% vs 72.2% for medial contact (P = .250).

• Inter-reader agreement was similar between PCD-CT and EID-CT (P = .250).

• Diagnostic confidence was higher with PCD-CT (all P ≤ .019).

• PCD-CT image quality was preferred in 88.9% (95% CI, 65.3%-98.6%) of cases.

Introduction

Ossicular chain reconstruction with metallic-based prostheses is commonly used to improve conductive hearing loss among patients with ossicular chain erosion, discontinuity, or fixation caused by chronic middle ear disease, otosclerosis, trauma, or congenital abnormalities.^1,2 However, prosthesis malposition or displacement is a well-recognized postoperative complication.^3,4 A review of 80 published articles on titanium partial and total ossicular replacement prostheses reported a mean extrusion or dislocation rate of 5.2% (range 0%-35%).⁵ Moreover, within the revision stapes surgery literature, prosthesis dislocation has been reported as the most frequent intraoperative finding, accounting for approximately 56% of revision cases in a recent multicenter study.⁴ Postoperative imaging plays a critical role in patient monitoring, especially in cases of complications or recurrent/persistent conductive hearing loss.^6,7 In clinical practice, CT imaging is commonly performed to evaluate prosthesis position relative to remnant ossicles, tympanic membrane, and the inner ear.^6–8

A major challenge of imaging these prostheses is that the small size of the structures requires high spatial resolution. Conventional energy-integrating detector (EID)-CT faces limitations when imaging small prostheses within the complex anatomy of the temporal bone, particularly regarding contact with adjacent ossicles or the tympanic membrane.^9,10 Lower spatial resolution and partial-volume effects restrict its ability to identify fine structures, creating diagnostic challenges that can affect clinical decision-making.^6,8 Over the past decade, photon-counting detector (PCD)-CT has emerged as a promising technology with several key advantages over EID-CT, including higher spatial resolution, lower noise, better contrast-to-noise ratio (CNR), increased dose efficiency, and reduced blooming artifact and beam-hardening.^11–18

Initial studies have shown that PCD-CT improves visualization of the temporal bone and ear anatomy, offering superior spatial resolution and lower radiation dose compared to conventional EID-CT scanners.^19–23 While these studies primarily focused on anatomical visualization, image quality, and dose reduction, there are limited data specifically regarding PCD-CT imaging of ossicular prostheses. Although case examples and a recent in vivo patient comparison have shown improved visualization of prostheses,^20,23–25 these have characterized image quality and spatial resolution rather than diagnostic accuracy against a known reference standard. A systematic evaluation of diagnostic performance for prosthesis position and complications has been lacking because patient studies cannot independently verify true prosthesis position or perform identical repeat scans across different CT systems, while phantom studies could not replicate real-world clinical challenges such as prosthesis interaction with surrounding structures.^13,25,26

To address this gap, a model replicating real clinical practice with a reliable reference standard is needed. Fresh-frozen whole human head cadavers address these methodological limitations and serve as a valuable model for accurate assessment of middle ear prostheses. Cadaveric models offer both anatomical realism and the rigorous reference standard verification necessary for diagnostic accuracy studies.^21,26 The aim of this study was to evaluate the visualization and diagnostic performance of PCD-CT compared to EID-CT for ossicular and stapes prostheses in fresh-frozen human cadaveric heads.

Materials and methods

Specimen preparation

This study was performed with the approval of our institution’s biospecimen committee. Five fresh-frozen whole human head specimens were obtained through the institution’s anatomy laboratory, selected based on availability and not screened for otologic pathology before inclusion. No gross middle ear abnormalities were identified during surgical exploration or subsequent imaging. Five stapes prostheses, 2 total ossicular replacement prostheses (TORP), and 2 partial ossicular prostheses (PORP) were evaluated. The prostheses were made of titanium, platinum, hydroxyapatite, and composite combinations (detailed in Table 1).

Table 1.

Characteristics of implanted prostheses.

	Prosthesis type	Name	Length	Reference number	Material	Type	Shaft diameter/head size	Picture
1	Stapes	Robinson Bucket Handle	3.75 mm	420-375	Titanium	Standard	0.4 mm
2	Stapes	Lippy Bucket Handle	4.50 mm	421-450	Titanium	Notched	0.4 mm
3	Stapes	Eclipse Piston	4.0 mm	467-400	Nitinol/Flpl	Piston	0.5 mm
4	Stapes	MEITI Piston	3.75 mm	418-375	Platinum/Titanium	Piston	0.5 mm
5	Stapes	Eclipse Piston	4.50 mm	467-450	Nitinol/Flpl	Piston	0.5 mm
6	TORP	Centered ALTO total	2.50-7.00 mm (adjustable)	604	Titanium	–	3.00 mm
7	TORP	Wildcat Pro	3.00 mm	709-300	HA/Titanium	Centered	1.00 × 1.75 mm (distal end)
8	PORP	Precise centered partial	3.50 mm	765-350	Titanium	Centered	0.2 mm (Shaft) 1.45 mm (Cup)
9	PORP	Centered ALTO partial	1.50 to 5.60 mm (Adjustable)	655	Titanium	–	3.00 mm (Head) 1.45 mm (Cup)

Abbreviations: Flpl = fluoroplastic; HA = hydroxyapatite; PORP = partial ossicular replacement prosthesis; TORP = total ossicular replacement prosthesis.

Each prosthesis was first implanted in its correct anatomical position by fellowship-trained otologic surgeons and then intentionally displaced to mimic common types of prosthesis malposition, thereby providing an image series with correct and displaced conditions for assessment.

CT image acquisition and reconstruction

All specimens were scanned on a PCD-CT scanner (NAEOTOM Alpha, Siemens). For comparison, the same specimens were also scanned on an EID-CT scanner from the same manufacturer (SOMATOM Force, Siemens). Clinical temporal bone protocols were used on both scanners with matched tube potential (120 kV), volume CT dose index (CTDI_vol), and field-of-view (FOV, 80 mm) (parameters summarized in Table 2). Images were reconstructed at the thinnest available slice thickness for each scanner: 0.2 mm for PCD-CT and 0.4 mm for EID-CT, using clinical kernels (Hr84 for PCD-CT and Ur77 for EID-CT).

Table 2.

CT acquisition and reconstruction parameters.

	PCD-CT a	EID-CTb
Scanner model	PCD-CT (NAEOTOM Alpha, Siemens)	Third-generation dual source (SOMATOM Force, Siemens)
Tube potential (kV)	120	120
CTDIvol (mGy)	Matched between 2 scanners for each specimen. Varies among specimens (automatically adjusted by scanner based on size), range: 48.8-64.2.
Scan mode	Spiral	Spiral
Pitch	0.5	0.35
Rotation time (s)	1	1
Collimation (mm)	120 × 0.2	64 × 0.6
FOV (mm)	80	80
Slice thickness (mm)	0.2	0.4
Increment (mm)	0.1	0.3
Reconstruction mode/strength	QIR/3	ADMIRE/3
Reconstruction kernel	Hr84	Ur77

^aPhoton-counting-detector.

^bEnergy-integrating-detector.

Image evaluation

Eighteen image series (9 prostheses × 2 positions) were independently evaluated by 2 board-certified, fellowship-trained neuroradiologists, with 5 and 7 years of post-fellowship experience. Each scanner had 9 image series with prostheses in the correct position and 9 with prostheses displaced. To minimize bias, readers were blinded to scanner type and prosthesis positioning (correct vs displaced). For each case, PCD-CT and EID-CT image series were presented side-by-side in a paired-comparison format. The left-right position of each scanner type was randomized to prevent readers from identifying which image corresponded to which scanner. The presentation order of specimens was also randomized. Readers independently evaluated each image series for each specimen using a standardized assessment protocol. For each image, prosthesis position was scored for lateral and medial contact using a 3-point ordinal scale: 1 = good contact, 2 = tangential contact, 3 = no contact. For each contact assessment, readers provided their diagnostic confidence using a continuous scale from 0 to 100, where 0 indicated no confidence and 100 indicated complete confidence in their contact classification. Finally, readers evaluated overall image quality in the paired comparisons using a 5-point Likert scale (1 = EID significantly better, 2 = EID slightly better, 3 = no difference between scanners, 4 = PCD slightly better, 5 = PCD significantly better). All image evaluations were performed on a diagnostic workstation with standard bone window settings (width/level: 4000/1000 HU) and controlled ambient lighting conditions to ensure consistency.

Reference standard

The reference standard for contact classification was established by an experienced neuroradiologist (35 years of post-fellowship experience) and an otologic surgeon (14 years of experience, including fellowship training) using intraoperative microscopic photographs obtained during prosthesis placement. PCD-CT images were reviewed in conjunction to confirm that no displacement of the prosthesis had occurred between surgery and scanning. Each case was assessed by the reference-standard panel by consensus, and each contact was classified using the same 3-point scale.

Statistical analysis

The primary objectives were to assess the accuracy of contact assessment and reader confidence, with secondary analyses evaluating overall image quality. Exact agreement was defined as equality between the reader-assigned score and the reference standard score. Agreement between individual reader scores and the reference standard was quantified using linearly weighted kappa (κ_w), reported per reader and pooled across both readers. Inter-reader agreement for contact scoring was assessed using linearly weighted kappa (κ_w). Kappa values were interpreted according to the criteria of Landis and Koch²⁷: ≤0.20, slight; 0.21 to <0.41, fair; 0.41 to <0.61, moderate; 0.61 to <0.81, substantial; and ≥0.81, almost perfect agreement. Ninety-five percent confidence levels (CIs) for κ_w estimations and pooled agreement proportions were obtained using bias-corrected and accelerated cluster bootstrap resampling with 10 000 iterations, with the prosthesis as the clustering unit. Reader-specific proportions were reported with Clopper-Pearson exact 95% CIs for descriptive purposes. Scanner differences in reader-specific agreement were assessed using the exact McNemar test, while pooled agreement and inter-reader agreement were tested with exact prosthesis-level permutation tests. Reader diagnostic confidence was compared between scanner types for each reader using a 2-sided paired t-test, with 95% CIs reported for paired mean differences. Wilcoxon signed-rank tests were performed as sensitivity analyses. Overall image quality ratings were compared with neutral score 3 using Wilcoxon signed-rank tests. The Holm method was applied to adjust for multiple comparisons within the accuracy and confidence analyses. Reader-specific McNemar, paired t-test, and Wilcoxon analyses preserved within-series pairing but did not additionally account for prosthesis or specimen level clustering and were therefore considered exploratory. A P value <.05 was considered statistically significant. All analyses were performed using Python (version 3.12).

Results

PCD-CT was preferred for prosthesis visualization and provided significantly higher diagnostic confidence than EID-CT for both readers (Figures 1-4). Radiation dose was matched between the 2 scanners for each specimen, with CTDI_vol ranging from 48.8 to 64.2 mGy across specimens depending on specimen size (Table 2). Figure 1 represents a comparative visualization of PCD-CT versus EID-CT images from a sample stapes prosthesis. Figures 2 and 3 compare PCD-CT and EID-CT for assessing stapes prosthesis positioning. As shown in Figure 2, in anatomically correct placement, PCD-CT provided better delineation of the stapes hook embracing the lenticular process of incus. In the dislocation case shown in Figure 3, both modalities demonstrated loss of contact, and both readers assigned a score of 3 for lateral contact. Figure 4 represents a case where the reference standard categorized medial contact as good contact (score 1).

Figure 1.

Comparison of prosthesis visualization between PCD-CT (photon-counting-detector CT) and EID-CT (energy-integrating-detector CT) imaging. Representative images of the prosthesis acquired using PCD-CT (A, C) and EID-CT (B, D), shown in axial-oblique planes in the first row and sagittal-obliques in the second row. The images show a 4.0-mm Eclipse Piston stapes prosthesis (solid arrows). The prosthesis is lateralized outside of the oval window (dashed arrow) and displaced posteriorly toward the pyramidal eminence. The reference standard classified lateral contact as score 3 (contact loss), and both readers assigned a contact score of 3 on both PCD-CT and EID-CT. Reader confidence was 100% for PCD-CT and 90% for EID-CT for reader 1 and 100% for PCD-CT and 60% for EID-CT for reader 2.

Figure 2.

Comparison of PCD-CT (A, C) and EID-CT (B, D) in the assessment of stapes prosthesis positioning. Reformatted axial-oblique (A, B) and coronal-oblique (C, D) images show a 4.5-mm Eclipse Piston stapes prosthesis (solid straight arrows). Curved arrows indicate the prosthesis hook around the long process of the incus, which is represented by dashed arrows. The reference standard classified the lateral contact as good contact (score 1). Reader confidence was 100% for PCD-CT and 80% for EID-CT for reader 1 and 80% for PCD-CT and 40% for EID-CT for reader 2.

Figure 3.

Visualization of dislocated stapes prosthesis shown in Figure 2. Coronal-oblique images from PCD-CT (A, C) and EID-CT (B, D) show the same 4.5-mm Eclipse Piston stapes prosthesis (marked with solid arrows), with no contact between the prosthesis hook and the long process of the incus (shown by dashed arrows). The reference standard classified lateral contact as contact loss (score 3), and both readers assigned a score of 3 on both PCD-CT and EID-CT. Reader confidence was 100% for PCD-CT for both readers, compared to 80% for reader 1 and 70% for reader 2 on EID-CT.

Figure 4.

Comparison of medial prosthesis contact on PCD-CT and EID-CT. The first row shows reformatted axial images from PCD-CT (A) and EID-CT (B), and the second row shows coronal images for PCD-CT (C) and EID-CT (D). The images represent a 3.5-mm precise centered partial ossicular replacement prosthesis (PORP; solid arrows). Dashed arrows indicate the medial contact. The reference standard classified medial contact as good contact (score 1). Both readers assigned scores of 1 on PCD-CT but considered the contact tangential and assigned a score of 2 on EID-CT. Reader confidence was 75% for PCD-CT and 50% for EID-CT for reader 1, and 80% for PCD-CT and 50% for EID-CT for reader 2.

In lateral contact assessment, PCD-CT achieved agreement with the reference standard in 31 out of 36 readings (86.1%; 95% CI, 58.3-94.4; κ_w = 0.830 [95% CI, 0.496-0.964]), while EID-CT achieved agreement in 29 out of 36 readings (80.6%; 95% CI, 55.6-91.7; κ_w = 0.714 [95% CI, 0.400-0.861]). The difference in pooled agreement was not statistically significant (P = .500). For medial contact assessment, PCD-CT achieved agreement with the reference standard in 30 out of 36 readings (83.3%; 95% CI, 55.6-94.4; κ_w = 0.761 [95% CI, 0.335-0.964]), whereas EID-CT achieved it in 26 out of 36 readings (72.2%; 95% CI, 50.0-83.3; κ_w = 0.605 [95% CI, 0.308-0.855]). The difference was not statistically significant (P = .250).

Individual reader accuracy was numerically higher with PCD-CT for most contact assessments, though no difference reached statistical significance. For lateral contact assessment, Reader 1 achieved 94.4% accuracy (17/18; 95% CI, 72.7-99.9) with both modalities (κ_w = 0.929 [95% CI, 0.581-1.000 for PCD-CT, and 95% CI, 0.576-1.000 for EID-CT]). No discordant pairs were observed, so the McNemar comparison was not informative. Reader 2’s accuracy was higher with PCD-CT (14/18; 77.8%; 95% CI, 52.4-93.6; κ_w = 0.741 [95% CI, 0.400-0.933]) than with EID-CT (12/18; 66.7%; 95% CI, 41.0%-86.7%; κ_w = 0.500 [95% CI, 0.100-0.767]; McNemar: unadjusted P = .500; Holm-adjusted P = 1.000). For medial contact assessment, Reader 1’s accuracy increased from 66.7% (12/18; 95% CI, 41.0-86.7) with EID-CT (κ_w = 0.518 [95% CI, 0.145-0.902]) to 77.8% (14/18; 95% CI, 52.4-93.6) with PCD-CT (κ_w = 0.690 [95% CI, 0.169-1.000]; McNemar: unadjusted P = .500; Holm-adjusted P = 1.000), and Reader 2’s accuracy improved from 77.8% (14/18; 95% CI, 52.4-93.6; κ_w = 0.690 [95% CI, 0.368-0.929]) to 88.9% (16/18; 95% CI, 65.3-98.6; κ_w = 0.836 [95% CI, 0.407-1.000]; McNemar: unadjusted P = .500; Holm-adjusted P = 1.000).

PCD-CT was numerically better at detecting contact loss (score 3). For lateral contact, both readers correctly identified all 3 cases (6/6 readings) of contact loss with PCD-CT, but using EID-CT 1 of the readers missed 1 case. For medial contact, both readers correctly identified both cases (4/4 readings) with PCD-CT, while with EID-CT, 1 of the readers again missed 1 case. Due to the limited number of score 3 cases, these findings were analyzed descriptively, and no formal comparisons were conducted. Analysis of discordant ratings revealed that discrepancies occurred primarily in distinguishing between good contact (score 1) and tangential contact (score 2). Detailed ratings from each reader, compared to the reference standard, are presented in Table 3. Patterns of agreement and discordance for both readers are further summarized using confusion matrices arranged by scanner and contact type, as presented in Table 4.

Table 3.

Confusion matrices comparing reader-assigned scores with the reference standard for lateral and medial prosthesis contact across 18 cases.

Contact Type	Reference standard	Reader 1						Reader 2
		PCD-CTa			EID-CTb			PCD-CT			EID-CT
		Score 1	Score 2	Score 3	Score 1	Score 2	Score 3	Score 1	Score 2	Score 3	Score 1	Score 2	Score 3
Lateral	Score 1	9	1	0	9	1	0	8	2	0	7	3	0
	Score 2	0	5	0	0	5	0	0	3	2	1	3	1
	Score 3	0	0	3	0	0	3	0	0	3	1	0	2
	Total cases	18 (PCD-CT)			18 (EID-CT)			18 (PCD-CT)			18 (EID-CT)
Medial	Score 1	8	4	0	7	5	0	10	2	0	8	4	0
	Score 2	0	4	0	0	4	0	0	4	0	0	4	0
	Score 3	0	0	2	0	1	1	0	0	2	0	0	2
	Total cases	18 (PCD-CT)			18 (EID-CT)			18 (PCD-CT)			18 (EID-CT)

Diagonal cells indicate agreement with the reference standard, while off-diagonal cells represent discordant ratings. Scoring: 1 = good contact, 2 = tangential contact, 3 = no contact.

^aPhoton-counting-detector.

^bEnergy-integrating-detector.

Table 4.

Confusion matrices comparing reader-assigned scores with the reference standard and showing patterns of agreement and discordance for lateral and medial contact, according to scanner type, photon-counting-detector (PCD)-CT, and energy-integrating-detector (EID)-CT, with results from both readers pooled.

Contact type	Reference standard	PCD-CT (both readers)			EID-CT (both readers)
		Score 1	Score 2	Score 3	Score 1	Score 2	Score 3
Lateral	Score 1	17	3	0	16	4	0
	Score 2	0	8	2	1	8	1
	Score 3	0	0	6	1	0	5
	Total	36 Readings			36 Readings
Medial	Score 1	18	6	0	15	9	0
	Score 2	0	8	0	0	8	0
	Score 3	0	0	4	0	1	3
	Total	36 Readings			36 Readings

Diagonal cells indicate agreement with the reference standard, while off-diagonal cells represent discordant ratings. Scoring: 1 = good contact, 2 = tangential contact, 3 = no contact.

Inter-reader agreement for contact scoring was numerically higher with PCD-CT than with EID-CT. For lateral contact, agreement was substantial with PCD-CT (κ_w = 0.804; 95% CI, 0.578-0.941) and moderate with EID-CT (κ_w = 0.571; 95% CI, 0.302-0.778), although the difference was not statistically significant (Δκ_w = 0.233; P = .250). For medial contact, agreement was substantial with PCD-CT (κ_w = 0.684; 95% CI, 0.191-0.934) and moderate with EID-CT (κ_w = 0.481; 95% CI, 0.024-0.835), with no statistically significant differences demonstrated between the 2 scanners (Δκ_w = 0.203; P = .250) (Figure 5).

Figure 5.

Linearly weighted kappa coefficients (κ_w) for inter-reader agreement in lateral and medial contact assessment with PCD-CT and EID-CT. For lateral contact, κ_w was 0.804 (95% CI, 0.578-0.941) for PCD-CT and 0.571 (95% CI, 0.302-0.778) for EID-CT. For medial contact, κ_w was 0.684 (95% CI, 0.191-0.934) for PCD-CT and 0.481 (95% CI, 0.024-0.835) for EID-CT. The dashed line shows the threshold for substantial agreement (κ_w = 0.61).

Reader confidence was significantly higher with PCD-CT compared to EID-CT for both lateral and medial contact assessments. For lateral contact, Reader 1 reported a mean confidence of 88.9 for PCD-CT compared to 79.4 for EID-CT, corresponding to a mean paired difference of 9.4 points (95% CI, 1.8-17.1; Holm-adjusted P = .019), and Reader 2 reported 86.9 compared to 63.9, corresponding to a mean difference of 23.1 points (95% CI, 13.4-32.7; Holm-adjusted P < .001). For medial contact, readers again reported higher confidence for PCD-CT. Reader 1 reported mean scores of 92.5 for PCD-CT compared to 76.1 for EID-CT, corresponding to a mean difference of 16.4 points (95% CI, 10.0-22.8; Holm-adjusted P < .001). Reader 2 reported mean scores of 93.3 and 75.6, corresponding to a mean paired difference of 17.8 points (95% CI, 6.8-28.8; Holm-adjusted P = .007). Wilcoxon signed-rank sensitivity analyses produced consistent results (all P ≤ .027).

For overall image quality, PCD-CT was also rated superior to EID-CT by both readers in paired comparisons. Reader 1 rated PCD-CT significantly better (score 5) in 16 out of 18 cases (88.9%; 95% CI, 65.3-98.6) and EID-CT slightly better (score 2) in 2 cases (11.1%; 95% CI, 1.4-34.7). Reader 2 rated PCD-CT significantly better (score 5) in 13 cases (72.2%; 95% CI, 46.5-90.3), PCD-CT slightly better (score 4) in 4 cases (22.2%; 95% CI, 6.4-47.6), and no difference (score 3) in 1 case (5.6%; 95% CI, 0.1-27.3); overall, Reader 2 preferred PCD-CT in 17 out of 18 cases (94.4%; 95% CI, 72.7-99.9). For both readers, ratings were significantly higher than the neutral score 3, showing a consistent preference for PCD-CT (both P < .001). The cases where EID-CT was preferred or rated equivalent to PCD-CT were those with severe metal artifacts, where both imaging techniques showed substantial artifacts but neither provided clearly superior visualization. In cases without metal artifacts, PCD-CT was uniformly preferred due to its superior spatial resolution and anatomical detail.

Discussion

This study assessed the visualization and diagnostic performance of PCD-CT compared to EID-CT for the evaluation of ossicular and stapes prostheses in cadaveric human heads. PCD-CT was consistently preferred for visualization of prosthesis positioning and adjacent anatomy. Given the small cohort size, the reader study demonstrated no statistically significant differences in assessing prosthesis contact points between the 2 technologies and in inter-reader agreement. However, diagnostic confidence was significantly higher on PCD-CT images, highlighting the potential advantages of PCD-CT technology.

The higher resolution and diagnostic confidence observed with PCD-CT compared with conventional EID-CT scanners may be explained by fundamental differences in detector design. PCD systems directly convert X-ray photons into electronic signals with a semiconductor, whereas EID systems use a 2-step process in which X-ray photons are first converted into visible light by a scintillator and then into electrical signal by photodiodes. This direct conversion design eliminates the reflective septa required between EID detector elements, allowing smaller detector pixels and improving both spatial resolution and geometric dose efficiency.^11,21,28,29 In addition, individual photon counting and energy thresholding in PCD systems can suppress low-energy electronic noise and improve contrast-to-noise ratio.^14,30 These advantages are more relevant in middle ear imaging for depicting small, high-contrast osseous and metallic structures, such as middle ear prostheses and their interactions with surrounding temporal bone anatomy.

Our findings are consistent with prior work demonstrating improved visualization of temporal bone anatomy and ossicular structures with PCD-CT.^11,19,20,31 Hermans et al¹⁹ demonstrated that PCD-CT visualization of temporal bone anatomy was rated significantly higher than EID-CT across 71 exams (36 EID-CT vs 35 PCD-CT exams) at 22% lower radiation exposure. Benson et al²⁰ reported an average 31% lower radiation dose with PCD-CT compared with EID-CT. In prior phantom- and patient-based work, results showed that PCD-CT provides superior spatial resolution and clearer visualization of ossicular prostheses compared to EID-CT.^25,26 This study builds upon those findings by suggesting that these technical advantages may improve visualization and diagnostic confidence in a simulated surgical model, thereby bridging the gap between technical image quality assessment and clinically relevant interpretation.

These results have several clinical implications. Accurate postoperative assessment of ossicular prostheses is critical when surgical outcomes are unsatisfactory or complications are suspected. When postoperative hearing is suboptimal and imaging is inconclusive, the surgeon may recommend middle ear exploration and definitive intervention is determined intraoperatively. Reliance on surgical exploration for diagnosis and treatment underscores the limitations of conventional imaging. In a multicenter cohort of 308 revision procedures, preoperative CT could not identify the cause of failure in 46.1% of cases, corresponding to a sensitivity of 53.9%.⁴ Furthermore, some explorations are non-informative. Among 129 revision stapes surgeries, 6% revealed normal anatomy and apparently satisfactory prosthesis placement, with no structural abnormality to account for the deficit.³² Such procedures expose patients to the risks of revision surgery without a demonstrable benefit. As imaging can determine whether revision surgery is necessary, higher diagnostic certainty can lead to better management, potentially decreasing unnecessary revision surgery or avoiding unnecessary delay in revision surgery when indicated.

This study has several limitations. First, cadaveric specimens do not fully replicate in vivo conditions, as factors like motion, fluid, or postoperative changes were not represented in this model and may affect imaging performance. Second, the sample size and range of prosthesis types were narrow, potentially limiting the generalizability of the findings. The small number of prostheses and specimens also resulted in wide confidence intervals and limited statistical power. Therefore, these findings should be considered exploratory. While reader assessments of higher image quality and diagnostic confidence are promising, larger sample sizes will be required to determine whether PCD-CT provides superior diagnostic performance. Third, although readers were blinded to scanner type and the paired images were presented in random order, experienced readers may have recognized differences in image appearance, potentially introducing residual reader bias. The controlled design, paired comparisons with the same specimens, the blinded reader assessments, and availability of a reference standard help to mitigate these limitations. Future research to validate the findings in postoperative patient cohorts is needed to demonstrate that the technical advantages observed here translate into better clinical outcomes.

Conclusions

PCD-CT provided significantly greater confidence and preferred image quality for evaluating the position and contact points of middle ear prostheses compared to EID-CT under matched radiation dose. These exploratory findings support further development of PCD-CT as a potentially better modality for postoperative assessment of middle ear prostheses.

Author contributions

Maryam Sadeghian (Conceptualization, Data curation, Formal analysis, Methodology, Validation, Visualization, Writing—original draft, Writing—review & editing), John Benson (Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Visualization, Writing—review & editing), Paul Farnsworth (Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Visualization, Writing—review & editing), John Lane (Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Writing—review & editing), Kevin Treb (Methodology, Writing—review & editing), Karl Khandalavala (Methodology, Writing—review & editing), Eric Babajanian (Methodology, Writing—review & editing), Michael Bruesewitz (Methodology, Writing—review & editing), Joseph Swicklik (Methodology, Writing—review & editing), Fang-Shu Ou (Formal analysis, Writing—review & editing), Cynthia McCollough (Funding acquisition, Investigation, Methodology, Writing—review & editing), Matthew Carlson (Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Writing—review & editing), and Shuai Leng (Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing—original draft, Writing—review & editing)

Supplementary material

Supplementary materialis available at Radiology Advances online.

Funding

Research reported in this work was supported by the National Institutes of Health under award number R01DC022228. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Health. Research reported in this work was also supported by the Mayo Clinic CT Clinical Innovation Center and the Mayo Clinic Graduate School of Biomedical Sciences.

Conflicts of interest

Please see ICMJE form(s) for author conflicts of interest. These have been provided as supplementary materials. C.H.M. is the recipient of a research grant to the institution from Siemens Healthineers. The other authors have no relevant conflicts of interest to disclose.

Data availability

The data underlying this article will be made publicly available in accordance with NIH data sharing requirements. Data will be deposited in a publicly accessible repository upon completion of embargo period; the repository name and access link will be provided upon publication.