Dose reduction for sinus and temporal bone imaging using photon-counting detector CT with an additional tin filter

Abstract

Objective:

To quantitatively demonstrate radiation dose reduction for sinus and temporal bone exams using high-resolution photon-counting detector (PCD) CT with an additional tin (Sn) filter.

Materials and methods:

A multi-energy CT phantom, an anthropomorphic head phantom and a cadaver head were scanned on a research PCD-CT scanner using ultra-high resolution (UHR) mode at 100 kV tube potential with an additional tin filter (Sn-100 kV), and CTDI_vol of 10 mGy. They were also scanned on a commercial CT scanner with an energy-integrating detector (EID) following standard clinical protocols. Thirty patients referred to clinically indicated sinus exams and two patients referred to temporal bone exams were scanned on the PCD-CT system following their clinical scans on an EID-CT. For the sinus cohort, PCD-CT scans were performed using Sn-100 kV at four dose levels at 10 mGy (n = 9), 8 mGy (n = 7), 7 mGy (n = 7) and 6 mGy (n = 7) and the clinical EID-CT was performed at 120 kV and 13.7 mGy (mean CTDI_vol). For the temporal bone scans, PCD-CT was performed using Sn-100 kV, 10.1 mGy and EID-CT was performed at 120 kV and routine clinical dose (52.6 and 66 mGy). For both PCD-CT and EID-CT, sinus images were reconstructed using H70 kernel at 0.75 mm slice thickness, and temporal bone images were reconstructed using a U70 kernel at 0.6 mm slice thickness. Additionally, iterative reconstruction with a dedicated sharp kernel (V80) was used to obtain high resolution PCD-CT images from a sinus patient scan to demonstrate improved anatomic delineation. Improvements in spatial resolution from the dedicated sharp kernel was quantified using modulation transfer function (MTF) measured using a wire phantom. A neuro-radiologist assessed the H70 sinus images for visualization of critical anatomical structures in low-dose PCD-CT images and routine-dose EID-CT images using a 5-point Likert scale (structural detection obscured and poor diagnostic confidence, score = 1; improved anatomic delineation and diagnostic confidence, score = 5). Image contrast and noise were measured in representative region-of-interests and compared between PCD-CT and EID-CT, and the noise difference between the two acquisitions was used to estimate the dose reduction in the sinus and temporal bone patient cohorts.

Results:

The multi-energy phantom experiment showed a noise reduction of 26% in the Sn-100 kV PCD-CT image, corresponding to a total dose reduction of 56% compared to EID-CT (clinical dose) without compromising image contrast. The PCD-CT images from the head phantom and the cadaver scans demonstrated a dose reduction of 67% and 83%, for sinus and temporal bone exams, respectively compared to EID-CT. In the sinus cohort, PCD-CT demonstrated a mean dose reduction of 67%. The 10 and 8 mGy sinus patient images from PCD-CT were significantly superior to clinical EID-CT for visualization of critical sinus structures (median score = 5 ± 0.82 and p = 0.01 for lesser palatine foramina, median score = 4 ± 0.68 and p = 0.039 for nasomaxillary sutures, and median score = 4 ± 0.96 and p = 0.01 for anterior ethmoid artery canal). The 6 and 7 mGy sinus patient images did not show any significant difference between PCD-CT and EID-CT. Additionally, V80 (sharp kernel, 10% MTF = 18.6 cm⁻¹) PCD-CT images from a sinus patient scan increased the conspicuity of nasomaxillary sutures compared to the clinical EID-CT images. The temporal bone patient images demonstrated a dose reduction of up to 85% compared to clinical EID-CT images while visualization of inner ear structures such as the incudomalleolar joint were similar between EID-CT and PCD-CT.

Conclusions:

Phantom and cadaver studies demonstrated dose reduction using Sn-100 kV PCD-CT compared to current clinical EID-CT while maintaining the desired image contrast. Dose reduction was further validated in sinus and temporal bone patient studies. The UHR capability from PCD-CT allowed improved anatomical delineation for sinus imaging compared to current clinical standard.

1. INTRODUCTION

Computed tomography (CT) remains the primary imaging modality of choice for sinus and temporal bone imaging due to its intrinsic benefits of high spatial resolution, fast scan time and wide availability. It is frequently utilized in trauma and surgical planning where the high spatial resolution greatly facilitates delineation of sub-millimeter anatomical structures often encountered in sinus and temporal bone imaging. Recent technical advancements has enabled further improvements in spatial resolution through the use of smaller detector pixels for high-resolution CT imaging¹. CT using photon-counting detectors (PCDs) has recently demonstrated superior imaging performance^2–11 compared to conventional CT using energy-integrating detectors (EIDs). Primary benefits from PCDs include reduced electronic noise for imaging obese patients¹², improved CT number accuracy¹³, high contrast-to-noise ratio (CNR) for contrast-enhanced CT^{5, 14–17}, and dose-efficient ultra-high resolution (UHR) imaging^{18, 19}. Investigational whole-body PCD-CT systems have demonstrated equivalent or superior imaging performance compared to conventional EID-CT in a range of studies involving phantoms, animals, human cadavers and patient subjects^{8, 20–23}.

Unlike EIDs that use comb/grid filter to reduce the pixel aperture and achieve UHR imaging²⁴, PCD semiconductor technology allows fabrication of smaller detector pixels without the need for reflective inter-pixel septa. The use of comb/grid filters in EID-CT to achieve UHR capability is dose inefficient, since the filters block x-rays that have already passed through the patient. PCDs facilitate dose-efficient UHR acquisitions through the use of smaller detector pixels. For a given spatial resolution defined by the modulation transfer function (MTF), smaller detector pixels (as in PCDs) allow aggressive filtering compared to data acquired using larger detector pixels (as in conventional EIDs). The filtering is controlled in the kernel operation²⁵, and smaller detector pixels enabled by PCDs offer lower CT image noise compared to larger EID pixels at matched spatial resolution. The noise benefit from small pixel PCDs translate to dose benefit, where PCD-CT images can be acquired at a lower dose than EID-CT for matched spatial resolution and image noise^{18, 26}. Further dose reduction could be achieved by shaping the x-ray beam spectra using an additional tin filtration^27–29. The tin filter removes low energy photons that are typically absorbed by the patient, therefore contributing to radiation dose but not much to image formation. Using a lower tube potential, such as 100 kV, also reduces the high-energy photons that result in less optimal contrast. The combination of low kV and tin filter results in x-ray beam spectra narrowed at an optimal energy window to improve dose efficiency. The purpose of this study was to quantitatively evaluate the combined benefits of UHR PCD-CT and the additional tin filtration for dose reduction in sinus and temporal bone exams, and assess dose reduction from PCD-CT in phantoms, cadaver and patient scans compared to a commercial EID-CT system.

2. MATERIALS AND METHODS

The Health Insurance Portability and Accountability Act (HIPAA) – compliant prospective study was approved by our institutional review board and written informed consent was obtained from study participants prior to their PCD-CT scans.

2.1. Whole-body photon-counting detector CT

The PCD-CT system (Somatom CounT, Siemens Healthcare, Forchheim, Germany) used in this study is based on a modified dual source CT scanner (Definition Flash, Siemens Healthcare). The scanner is comprised of two subsystems; A-subsystem includes a conventional EID with a 50 cm scan field-of-view (FOV), B-subsystem includes a cadmium telluride PCD with a 27.5 cm scan FOV. The different PCD acquisition modes and their properties are tabulated in Table 1. Additional technical details of the scanner can be found elsewhere^{18, 23, 26}.

Table 1:

PCD-CT acquisition modes

Mode	Collimation	No. of Energy thresholds	Rotation time	Focal spot size
Macro	32 × 0.50 mm	2 (TL, TH)	0.5 or 1.0s	0.9 × 1.1 mm2
Chess	32 × 0.50 mm	4 (TL1, TL2, TH1, TH2)	0.5 or 1.0s	0.9 × 1.1 mm2
UHR	32 × 0.25 mm	2 (TL, TH)	1.0s	0.7 × 0.9 mm2
Sharp	48 × 0.25 mm	2 (TL, TH)	0.5 or 1.0s	0.9 × 1.1 mm2

TL: low-energy threshold, TH: high-energy threshold

An additional tin filter (thickness = 0.4 mm) could be enabled for any of the four acquisition modes. We recently reported³⁰ using tin filter, elevated tube potential (140 kV) and high-energy thresholding to substantially reduce metal artifacts arising from metallic implants³⁰. In this study, we employed the tin (Sn) filter and reduced tube potential of 100 kV to achieve low-dose, ultra-high resolution PCD-CT for sinus and temporal bone imaging while maintaining the desired image quality.

2.2. Sinus and temporal bone - Phantom and cadaver experiments

A multi-energy CT phantom (Gammex, Sun Nuclear, WI), an anthropomorphic head phantom and a cadaver head were scanned using the UHR mode with Sn filter for quantitative evaluation of image quality and radiation dose reduction for sinus and temporal bone exams. The same phantoms and cadaver head were also scanned on a commercial EID-CT (Definition Flash, Siemens Healthcare, Forchheim, Germany) at 120 kV, following the routine clinical protocols used in our institute. Note that an UHR mode with comb filter was used for the temporal bone cadaver scan on the EID-CT. Details of the phantoms and image acquisitions are listed in Table 2. The spectrum for 100 kV with Sn filter (hereafter abbreviated as Sn-100 kV) has a mean energy close to that of a 120 kV spectrum without tin filter, as shown in Fig 1. Therefore, Sn-100 kV was selected for PCD-CT acquisitions for comparison against 120 kV EID-CT acquisitions which are used in clinical exams for sinus and temporal bone imaging. The Sn-100 kV PCD-CT scans were acquired using maximum mAs, which correspond to a CT volume dose index (CTDI_vol 16 cm) of about 10 mGy. This dose level is lower than the clinical dose used for sinus and temporal bone imaging (13.5 mGy and 60 mGy, respectively).

Table 2:

CT acquisition and reconstruction parameters used in the phantom and cadaver experiments

	Multi-energy CT phantom		Anthropomorphic head phantom		Cadaver head
Acquisition System	EID-CT1	PCD-CT2	EID-CT1	PCD-CT2	EID-CT1	PCD-CT2
Protocol	Sinus	Sinus UHR	Sinus	Sinus UHR	T-bone UHR*	T-bone UHR
Tube potential (kV)	120	100	120	100	120	100
Additional Tin filter	No	Yes, 0.4 mm	No	Yes, 0.4 mm	No	Yes, 0.4 mm
Energy thresholds (keV)	-	25, 65	-	25, 65	-	25, 65
Tube-current-time product (mAs)	88	545	71	500	215	500
Pitch	0.55	0.55	0.60	0.60	0.60	0.60
CTDIvol (16 cm)	13.5	10.8	13.5	10.1	48	10.1
Collimation (mm)	64 × 0.6	32 × 0.25	64 × 0.6	32 × 0.25	8 × 0.6	32 × 0.25
Rotation time (s)	1	1	1	1	1	1
Reconstruction kernel	H70	H70	H70	H70	U70	U70
Slice thickness (mm)	0.75	0.75	0.75	0.75	0.6	0.6
Reconstruction FOV (mm)	275		275		80
Image matrix size	512 × 512		1024 × 1024		512 × 512

¹Definition Flash

²Somatom CounT

^*EID-based UHR is enabled using comb filter

Fig 1:

X-ray source spectra for 120 kV without tin filtration (red dashed line) and 100 kV with 0.4 mm tin filtration (black line). The tin filter removes low energy photons from the 100 kV spectrum causing an increase in the mean energy (72.6 to 75.2 keV).

The reconstruction kernel selection was based on routine clinical protocols at our institute, and matched between PCD-CT and EID-CT scans (see Table 2). All images were reconstructed using a weighted-filtered-back projection (WFBP) technique³¹. A sharp head kernel H70 (10% MTF = 13.4 cm⁻¹) was used for the multi-energy phantom and the sinus head phantom scans, while a dedicated sharp kernel U70 (10% MTF = 16 cm⁻¹) was used for the cadaver head (temporal bone) scan.

Mean CT number and image noise (standard deviation) were measured using circular region-of-interests (ROIs) placed in a uniform region of the PCD CT low-energy threshold images (25–100 keV) and EID-CT image. The potential dose reduction from PCD-CT using UHR mode and Sn filter relative to routine 120 kV EID-CT can be estimated based on the relationship between image noise and radiation dose. This is given by:

$$\eta =1-\left[\frac{\text{DPCD}}{{\text{D}}_{\text{FID}}}\times \frac{{\sigma }^2\text{PCD}}{{\sigma }^2\text{FD}}\right]$$ (Eq. 1)

Where $\eta$ is the expected dose reduction factor when image noise is matched, D is the CTDI_vol (mGy) used in the CT acquisitions from both EID and PCD scanners, and σ is the noise measured in the reconstructed images.

2.3. Spatial resolution measurement

To assess spatial resolution, MTF was measured by scanning a 50-micron diameter tungsten wire phantom on the PCD-CT system and the EID-CT system, following the standard protocols. The scanning and reconstruction parameters, such as kV, mAs, kernel, and slice thickness are tabulated in Table 3. In addition to the standard sharp reconstruction kernels, H70/J70, that are available on the EID-CT and PCD-CT for sinus image reconstructions, dedicated kernels U80/V80 that are only available on PCD-CT were also used for the MTF measurements. The reconstruction kernels were kept consistent with phantom and patient scans.

Table 3:

PCD-CT acquisition settings for MTF measurements

System	kV/mAs	Reconstruction kernel	Slice thickness (mm)	CTDIvol
PCD-CT	Sn-100kV/500	H70/J70 U80/V80	0.75 mm	10.1 mGy
EID-CT	120kV/71	H70/J70	0.75 mm	13.5 mGy

Note: The reconstruction kernel field (e.g. H70/J70) represents the same kernel in wFBP/SAFIRE reconstruction mode. The two kernels have identical spatial resolution. U80/V80 are not available on EID-CT for the sinus scan.

2.4. Demonstration of dose reduction in patient images using PCD-CT

Patients referred to clinically indicated sinus (n = 30) and temporal bone (n = 2) CT exams were recruited. After their clinical EID-CT scans (performed on one of the three scanners: Definition Flash, Definition AS+, Somatom Force - Siemens Healthcare, Germany), a research PCD-CT (Somatom CounT) scan was performed on the same anatomical region of interest. The acquisition settings and reconstruction parameters for the patient scans are shown in Table 4 and Table 5. The sinus cohort comprised of PCD-CT scans obtained at four progressively lower dose levels (10, 8, 7 and 6 mGy) compared to the current clinical dose (13.5 mGy). The two temporal bone patient cases were included to evaluate additional dose reduction from PCD-CT Sn-100kV compared to the comb-filter based EID-CT typically used in clinical practice for temporal bone imaging. PCD-CT temporal bone scans were acquired at the maximum dose (10 mGy) available for Sn-100 kV configuration on PCD-CT while the clinical EID-CT scans were acquired at 52 and 66 mGy routine dose.

Table 4:

Sinus and temporal patient CT acquisitions and reconstruction parameters

Sinus exam (n = 30 patient cases)
Acquisition platform (number of patient cases)	Definition Flash (9) Somatom Force (13) Definition AS+ (8)	Somatom CounT (30)
Protocol/mode	Sinus/standard	Sinus/UHR
Collimation (mm)	128 × 0.6 (Flash, AS+) 192 × 0.6 (Force)	32 × 0.25
Tube potential (kV)	120	100
Additional tin filter	No	Yes, 0.4 mm tin
Mean tube-current-time product in mAs and (number of patient cases)	95 (30)	533 (10) 400 (7) 350 (7) 300 (7)
Mean CTDIvol (16 cm)	13.70	10.6 8.0 7.0 6.0
Pitch	0.6	0.6
Rotation time (s)	1	1
Energy thresholds (keV)	-	25, 65
Temporal bone exam (n = 2 patient cases)
Acquisition platform (number of patient cases)	Definition Flash (2)	Somatom CounT (2)
Protocol/mode	T-bone UHR*	T-bone UHR
Collimation (mm)	16 × 0.6 (Flash)	32 × 0.25
Tube potential (kV)	120	100
Additional tin filter	No	Yes, 0.4 mm tin
Mean tube-current-time product in mAs and (number of cases)	300 (2)	500 (2)
Mean CTDIvol (16 cm)	66	10.1
Pitch	0.8	0.6
Rotation time (s)	1	1
Energy thresholds (keV)	-	25, 65

^*EID-based UHR is enabled using comb filter

Table 5:

Image reconstruction parameters for sinus and temporal bone patient study

	Sinus				Temporal bone
Scanner platform	EID-CT		PCD-CT		EID-CT	PCD-CT
Image matrix size	1024 × 1024	512 × 512	1024 × 1024	512 × 512	512 × 512	512 × 512
Reconstruction method	wFBP	SAFIRE	wFBP	wFBP	wFBP
Reconstruction kernel	H70/Hr69	J70	H70/Hr69	V80	U70
SAFIRE strength	-	3	-	5	-
Slice thickness (mm)	0.75	0.6	0.75	0.5	0.60
Reconstruction field-of-view (mm)	275*	150	275*	150	80

^*Images used in the reader study

SAFIRE = Sinogram Affirmed Iterative Reconstruction (Siemens Healthcare, Germany)

wFBP = weighted filtered back projection

A board-certified neuro-radiologist (D.R.D.) with over 20 years of experience evaluated the sinus images from both EID-CT and PCD-CT in a blinded fashion. A 5-point Likert scale (1 = worse visualization and confidence compared to clinical routine; 2 = worse than clinical routine, no change in diagnostic confidence; 3 = similar to routine clinical sinus images; 4 = preferred, no change in diagnostic confidence; 5 = improved detection of critical structures and improved diagnostic confidence) was used to assess the visualization of critical sinus structures that included the sphenoid ostia, lesser palatine foramina, nasomaxillary sutures and anterior ethmoid artery canal. For each patient, EID-CT and PCD-CT image display was randomized and scored independently. The temporal bone patient images from EID-CT and PCD-CT scanners were qualitatively assessed for visualization of inner ear structures such as stapes and incudomalleolar joint. A comprehensive assessment (like the sinus images) was not performed due to the limited number of patients.

Image noise was measured for both sinus and temporal bone images acquired from PCD-CT and EID-CT with matched reconstruction parameters (kernel, matrix size, slice thickness and FOV). For each patient image volume (PCD-CT or EID-CT), a circular ROI was placed in a homogeneous tissue region, and image noise was measured as the standard deviation of CT numbers. This measurement was repeated in three different slices and the average image noise value was reported. The potential dose reduction was calculated using the relationship between image noise and CT dose given in Eq. 1.

2.5. Demonstration of high spatial resolution in sinus patient images

In addition to the reader evaluation for the sinus patient images from routine dose EID-CT and low dose PCD-CT scans, a dedicated sharp kernel (V80) was used to demonstrate the UHR capability for sinus imaging. This kernel was available for the PCD-CT but not the EID-CT as the larger detector pixel size in EID-CT cannot support the high spatial resolution. For temporal bone images, identical spatial resolution was anticipated for both PCD-CT and EID-CT images since both scans were acquired with UHR mode using matched reconstruction kernel (dedicated sharp U70). Images reconstructed with U70 from the comb filter-based EID-CT exhibit similar spatial resolution as the PCD-UHR U70 images³².

2.5. Statistical analysis

For analysis of the reader evaluation scores, the sinus patient cohort was subdivided into two dose groups: 10 and 8 mGy dose group, and 7 and 6 mGy dose group. For each dose group, Wilcoxon signed rank test was applied to EID-CT vs PCD-CT and a p-value < 0.05 was considered statistically significant. Median and standard deviation from the reader evaluation scores were calculated individually for PCD-CT and EID-CT and compared. Since the image quality for the temporal bone images was only qualitatively assessed in two patients, no formal statistical analysis was performed on this cohort.

RESULTS

3.1. Phantom and cadaver experiments

Results from the multi-energy phantom experiment are shown in Fig 2. PCD-CT images from Sn-100 kV acquisition (Fig 2B) showed lower image noise (Fig 2D) compared to EID-CT (Fig 2A) at 120 kV despite the 20% reduction in dose for the PCD-CT acquisition. The average noise reduction across all the inserts was 26.2% while the image contrast between PCD-CT (Sn-100 kV) and EID-CT (120 kV) were comparable (Fig 2C). Using Eq. 1, the dose reduction from Sn-100 kV PCD-CT was calculated to be 56.5% if matched image noise is targeted. The results from the multi-energy phantom experiment demonstrate that Sn-100 kV PCD-CT could be used as an alternative to 120 kV EID-CT with the potential to reduce dose if matched image noise and contrast is targeted for a given reconstruction kernel.

Fig. 2:

Image contrast and noise evaluation using the multienergy CT phantom scanned using [A] EID-CT 120 kV at CTDI_vol of 13.5 mGy (sinus clinical dose), and [B] PCD-CT, UHR mode, Sn – 100 kV at CTDI_vol of 10.8 mGy. [C] Image contrast between EID-CT 120 kV and PCD-CT Sn-100 kV are comparable while [D] PCD-CT Sn-100 kV showed lower image noise across all material inserts compared to routine EID-CT 120 kV. Phantom inserts include hydroxyapatite 200 and 400 mg/cc (labeled as HA-200 and HA-400 respectively), blood (40 and 70 HU), cortical bone, adipose, and brain grey and white matter (labeled brain-G and brain-W respectively). CT display W/L = [2000/100] HU. EID and PCD CT images reconstructed using H70 kernel, 0.75 mm slice thickness.

EID-CT and PCD-CT images from the anthropomorphic head phantom scans (sinus protocol) and the cadaver scans (temporal bone protocol) are shown in Fig 3. The head phantom image corresponding to PCD-CT Sn-100 kV (10.1 mGy) in Fig. 3B showed 33% lower noise compared to EID-CT at 120 kV (13.5 mGy, clinical dose) in Fig 3A. The total dose reduction that could be achieved from Sn-100 kV PCD-CT relative to EID-CT is 67% for matched image noise. Similarly, the Sn-100 kV PCD-CT temporal bone results (Fig 3D) showed about 9% lower noise despite a five-fold dose reduction during acquisition compared to the EID-CT (Fig 3C). This corresponds to a total dose reduction of 83% that could be achieved for matched image noise between Sn-100 kV PCD-CT and 120 kV EID-CT.

Fig. 3:

[A] Anthropomorphic head phantom images scanned using EID-CT 120 kV at 13.5 mGy (sinus clinical dose), and [B] PCD-CT, UHR mode, Sn – 100 kV at 10.1 mGy dose. The PCD-CT image showed lower noise than EID-CT while the image contrast was maintained. [C] Cadaver head scanned using EID-CT (comb-based UHR) at 120 kV (48 mGy) and [D] the same cadaver head scanned using PCD-CT Sn-100 kV (10.1 mGy). Image noise was lower on the PCD-CT images while the image contrast and conspicuity of incudomalleolar joint (red arrow) were similar between PCD-CT and EID-CT images. CT display W/L = [2000/100] HU for sinus phantom, and [2800, 600] HU for temporal bone cadaver images.

3.2. Patient images from sinus and temporal bone exams

Fig 4 shows CT images from a sinus patient exam obtained with PCD-CT and clinical EID-CT systems. For a given kernel (H70) and slice thickness (0.75 mm), the PCD-CT image at 7 mGy showed lower image noise (by 21%) compared to the EID-CT image at 13.6 mGy. This corresponds to a total dose reduction of 67% dose reduction for UHR PCD-CT Sn-100 kV if matched image noise is targeted using the same reconstruction kernel.

Fig. 4:

Image contrast and noise evaluation in patient sinus images using [A] EID-CT 120 kV at 13.6 mGy (sinus clinical dose), and [B] PCD-CT, UHR mode, Sn – 100 kV at 7.07 mGy dose (48% lower dose). Both images were reconstructed using the same kernel (H70) at 0.75 mm slice thickness. The PCD-CT image showed 21% lower noise compared to EID-CT. CT display W/L = [3200/700] HU

The scores from the sinus reader study showed that both EID-CT and PCD-CT images had clinically acceptable image quality (median scores ranging from 3 – similar to routine clinical images, to 5 – improved detection of sinus structures and improved diagnostic confidence) as summarized in Table 6. The PCD-CT sinus images at 10 and 8 mGy dose levels was significantly superior to routine-dose EID-CT for visualization of lesser palatine foramina (median score = 5 ± 0.82 vs 4 ± 0.86, p = 0.01), nasomaxillary sutures (median score = 4 ± 0.68 vs. 3 ± 0.72, p = 0.0039) and the anterior ethmoid artery canal (median score = 4 ± 0.96 vs. 3 ± 0.10, p = 0.0027). The sphenoid ostia visualization received a median score of 5 ± 0.81 on PCD-CT and 4 ± 0.81 on EID-CT at p > 0.05 (not significant). At 7 and 6 mGy dose levels, the ability to visualize critical sinus structures was not significantly different (p = 0.1 to 0.75) between the PCD-CT and routine-dose EID-CT images. Across all dose levels, PCD-CT received a median score ranging from 3 (similar to routine clinical sinus images) to 5 (improved detection and diagnostic confidence) for visualization of critical sinus structures. The mean dose reduction from Sn-100kV PCD-CT relative to routine EID-CT calculated from the image noise differences across the sinus cohort was 67% (25th – 75th percentile = 64 – 70%), which was consistent with the findings from the sinus phantom experiments.

Table 6:

Summary of reader evaluation scores for sinus PCD-CT and EID-CT patient images

	PCD-CT Sn-100kV (median ± std)	EID-CT (median ± std)	p-value
Sphenoid Ostia
(10, 8) mGy	5 ± 0.81	4 ± 0.81	0.1250
(7, 6) mGy	4 ± 1.17	4 ± 1.22	0.3750
Lesser palatine foramina
(10, 8) mGy	5 ± 0.82	4 ± 0.86	0.0137 *
(7, 6) mGy	4 ± 1.04	4 ± 0.58	0.5078
Nasomaxillary sutures
(10, 8) mGy	4 ± 0.68	3 ± 0.72	0.0039 *
(7, 6) mGy	3 ± 0.76	4 ± 0.53	0.1094
Anterior ethmoid artery canal
(10, 8) mGy	4 ± 0.96	3 ± 0.10	0.0027 *
(7, 6) mGy	4 ± 0.91	3.5 ± 0.85	0.7500

^*= statistically significant (p < 0.05)

Score description: 1= worse visualization and confidence compared to clinical routine; 2= worse than clinical routine, no change in diagnostic confidence; 3= similar to routine clinical sinus images; 4= preferred, no change in diagnostic confidence; 5= improved detection of critical structures and improved diagnostic confidence

Additionally, PCD-CT image reconstructed using the dedicated sharp kernel V80 exhibited superior anatomical delineation compared to conventional EID-CT as shown in Fig 5A and 5B. Improved conspicuity of the nasomaxillary sutures was noticed in the V80 PCD-CT image, compared to the J70 image (sharpest SAFIRE kernel available) from EID-CT. This could be attributed to the higher spatial resolution from the smaller detector pixels and the dedicated sharp kernel available in the UHR mode for PCD-CT as shown in Fig 6. The 10% MTF value for the V80 kernel (PCD-CT) was 18.6 cm⁻¹ compared to 13.3 cm⁻¹ from the J70 kernel, the sharpest head kernel available in EID-CT for sinus imaging.

Fig. 5:

High-resolution images of the sinus patient obtained using [A] EID-CT 120 kV at 13.6 mGy and [B] PCD-CT Sn-100 kV at 7 mGy. The EID-CT image was reconstructed using J70 (sharpest SAFIRE head kernel available on this system) and the PCD-CT image was obtained using a dedicated sharp kernel V80. The image noise in the two images was comparable, but the high-resolution V80 PCD-CT image improved the conspicuity of the nasomaxillary sutures (marked by red arrows). CT display W/L = [1800/10] HU

Fig. 6:

MTF measurements from PCD-CT and EID-CT. The dedicated sharp kernel (U80/V80) available on the PCD-CT system for sinus reconstructions demonstrated higher spatial resolution (10% MTF = 18.6 cm⁻¹), compared to the standard sharp kernel (H70/J70, routine clinical kernel for sinus CT reconstructions) available on EID-CT and PCD-CT systems.

Fig 7 shows the patient images from temporal bone exams acquired using PCD-CT and EID-CT. The Sn-100 kV PCD-CT acquisition facilitated 81 – 85% dose reduction compared to the clinical EID-CT exam (at 52.6 and 66 mGy) at matched image noise without compromising image contrast as shown in the ROI analysis. EID-CT based UHR imaging suffers from poor dose efficiency due to the inclusion of comb filters while PCD-CT UHR imaging is achieved through smaller detector pixels which is a dose-efficient approach to UHR imaging. Visualization of fine temporal bone structures such as the incudomalleolar joint was comparable between the UHR images from EID-CT and PCD-CT in the two patient examples.

Fig. 7:

Temporal bone patient images acquired with [A] EID-CT 120 kV at 66 mGy (clinical dose for temporal bone), and [B] PCD-CT, UHR mode, Sn – 100 kV at 10.1 mGy dose. Both images were reconstructed using a dedicated sharp kernel (U70) and demonstrated comparable image noise and CT number as shown in the ROIs, however the PCD-CT image was acquired at about 6-fold reduced dose compared to the clinical EID-CT image. The visualization of incudomalleolar joint (red arrow) in the low dose PCD-CT image was comparable to that of the clinical EID-CT image. The absence of UHR comb filter in PCD-CT combined with the additional tin filter enabled a substantial dose reduction while preserving the image quality and representation of fine anatomic details. CT display W/L = [3200/700]

3. DISCUSSION

We demonstrated radiation dose reduction for sinus and temporal bone CT exams using a research PCD-CT scanner with UHR mode and an additional tin filter. For sinus protocol, the average dose reduction using Sn-100 kV PCD-CT in the multi-energy CT phantom, anthropomorphic head phantom and patient images were 56%, 67%, and 67%, respectively. The reader study demonstrated non-inferiority of Sn-100kV PCD-CT for assessment of critical sinus structures. For the temporal bone protocol, dose reduction from Sn-100 kV PCD-CT in the cadaver head and patient images were 83% and 81–85%, respectively. The dose reduction factor was consistent between phantom, cadaver and patient studies for sinus and temporal bone exams. The demonstrated dose reduction from PCD-CT is due to the combination of three key factors: (i) The additional tin filter removes low energy x-ray photons from the primary beam that are likely to contribute to patient dose (absorbed by the patient) but produces little signal. (ii) The PCD UHR mode has better intrinsic detector resolution due to smaller detector pixel size. Therefore stronger filter can be applied resulting in reduced noise, compared to data from larger EID pixels for matched spatial resolution. (iii) In PCD-CT, UHR capability is achieved through the design of smaller detector pixels facilitated by the semiconductor technology, thereby eliminating the need for comb/grid filters for UHR imaging. The comb/grid filters are dose-inefficient since they block x-rays that have already passed through the patient but do not contribute to image formation. The dose reduction from PCD-CT in temporal bone exams were more pronounced than sinus exams due to the use of dose-inefficient UHR comb filter in the temporal bone EID-CT acquisitions. For sinus imaging, the UHR mode from PCD-CT offers the flexibility to utilize dedicated sharp kernels (such as V80) which is not possible in current clinical EID-CT exams that are performed using the conventional non-UHR mode. The smaller detector pixel size from PCD-CT allows high resolution imaging without sacrificing geometrical dose efficiency¹⁸. The use of dedicated sharp kernels could yield improved delineation of finer anatomical details encountered in sinus exams as demonstrated in Figure 5. When using the dedicated sharp kernel (V80), we reduced the reconstruction field of view from 275 mm to 150 mm to visualize fine anatomic details such as the nasomaxillary sutures. Small FOV reduces the voxel size and facilitates finer sampling of CT image pixels in order to exploit the improved spatial resolution from the UHR mode and the sharp kernel.

Since the mean energies of Sn-100 kV and 120 kV spectrum are similar, the final image contrast is similar between low energy threshold PCD-CT (Sn-100 kV) and routine EID-CT (120 kV) as demonstrated in the multi-energy phantom experiment with a wide range of tissue surrogates. The dose reduction technique described in this study could have a higher impact for in pediatric population, for instance, in cases of cystic fibrosis or ciliary disorder that often requires CT scans at a very young age and involves repeated CT exams. For sinus applications, chronic sinusitis including allergic fungal sinusitis in adult and pediatric population warrants repeated sinus CT exams for longitudinal evaluation. Similarly, for temporal bone imaging, recurrent cholesteatoma and conductive hearing loss after ossicular chain reconstruction often requires repeat CT exams for clinical evaluation, which could benefit from the proposed dose reduction technique using PCD-CT.

While we demonstrated dose reduction in thirty sinus patients scanned using PCD-CT, the temporal bone cohort was limited to two patient cases to demonstrate potential dose reduction as a preliminary proof-of-concept. Future work involving additional temporal bone patients are underway at our institute for comprehensive evaluation of image quality and diagnostic outcomes to ascertain the overall clinical impact of the proposed dose reduction technique in temporal bone imaging. In this study we focused on sinus and temporal bone exams, but other clinical high-contrast imaging tasks such as chest, colon and musculoskeletal imaging may benefit from the proposed dose reduction technique. We used UHR mode in our studies since sinus and temporal bone exams demand high spatial resolution. However the tin filter can be used with the other PCD-CT modes such as Sharp and Macro for dose reduction.

In conclusion, PCD-CT images acquired at 100 kV with an additional tin filter demonstrated a dose reduction of 67% for sinus imaging, and up to 85% dose reduction for temporal bone imaging compared to current clinical EID-CT. For sinus exams, besides dose reduction, PCD-CT also offers the ability to reconstruct ultra-high resolution images using a dedicated sharp kernel which is not possible with current clinical EID-CT protocols. No loss in image contrast was observed in the PCD-CT Sn-100 kV images compared with 120 kV EID-CT images for both sinus and temporal bone acquisitions.