Super-resolution deep-learning reconstruction for cardiac CT: impact of radiation dose and focal spot size on task-based image quality

This study aimed to evaluate the impact of radiation dose and focal spot size on the image quality of super-resolution deep-learning reconstruction (SR-DLR) in comparison with iterative reconstruction (IR) and normal-resolution DLR (NR-DLR) algorithms for cardiac CT. Catphan-700 phantom was scanned on a 320-row scanner at six radiation doses (small and large focal spots at 1.4-4.3 and 5.8-8.8 mGy, respectively). Images were reconstructed using hybrid-IR, model-based-IR, NR-DLR, and SR-DLR algorithms. Noise properties were evaluated through plotting noise power spectrum (NPS). Spatial resolution was quantified with task-based transfer function (TTF); Polystyrene, Delrin, and Bone-50% inserts were used for low-, intermediate, and high-contrast spatial resolution. The detectability index (d') was calculated. Image noise, noise texture, edge sharpness of low- and intermediate-contrast objects, delineation of fine high-contrast objects, and overall quality of four reconstructions were visually ranked. Results indicated that among four reconstructions, SR-DLR yielded the lowest noise magnitude and NPS peak, as well as the highest average NPS frequency, TTF50%, d' values, and visual rank at each radiation dose. For all reconstructions, the intermediate- to high-contrast spatial resolution was maximized at 4.3 mGy, while the lowest noise magnitude and highest d' were attained at 8.8 mGy. SR-DLR at 4.3 mGy exhibited superior noise performance, intermediate- to high-contrast spatial resolution, d' values, and visual rank compared to the other reconstructions at 8.8 mGy. Therefore, SR-DLR may yield superior diagnostic image quality and facilitate radiation dose reduction compared to the other reconstructions, particularly when combined with small focal spot scanning.

Can 1.25 mm thin-section images generated with Deep Learning Image Reconstruction technique replace standard-of-care 5 mm images in abdominal CT?

BACKGROUND

CT image reconstruction has evolved from filtered back projection to hybrid- and model-based iterative reconstruction. Deep learning-based image reconstruction is a relatively new technique that uses deep convolutional neural networks to improve image quality.

OBJECTIVE

To evaluate and compare 1.25 mm thin-section abdominal CT images reconstructed with deep learning image reconstruction (DLIR) with 5 mm thick images reconstructed with adaptive statistical iterative reconstruction (ASIR-V).

METHODS

This retrospective study included 52 patients (31 F; 56.9±16.9 years) who underwent abdominal CT scans between August-October 2019. Image reconstruction was performed to generate 5 mm images at 40% ASIR-V and 1.25 mm DLIR images at three strengths (low [DLIR-L], medium [DLIR-M], and high [DLIR-H]). Qualitative assessment was performed to determine image noise, contrast, visibility of small structures, sharpness, and artifact based on a 5-point-scale. Image preference determination was based on a 3-point-scale. Quantitative assessment included measurement of attenuation, image noise, and contrast-to-noise ratios (CNR).

RESULTS

Thin-section images reconstructed with DLIR-M and DLIR-H yielded better image quality scores than 5 mm ASIR-V reconstructed images. Mean qualitative scores of DLIR-H for noise (1.77 ± 0.71), contrast (1.6 ± 0.68), small structure visibility (1.42 ± 0.66), sharpness (1.34 ± 0.55), and image preference (1.11 ± 0.34) were the best (p<0.05). DLIR-M yielded intermediate scores. All DLIR reconstructions showed superior ratings for artifacts compared to ASIR-V (p<0.05), whereas each DLIR group performed comparably (p>0.05, 0.405-0.763). In the quantitative assessment, there were no significant differences in attenuation values between all reconstructions (p>0.05). However, DLIR-H demonstrated the lowest noise (9.17 ± 3.11) and the highest CNR (CNRliver = 26.88 ± 6.54 and CNRportal vein = 7.92 ± 3.85) (all p<0.001).

CONCLUSION

DLIR allows generation of thin-section (1.25 mm) abdominal CT images, which provide improved image quality with higher inter-reader agreement compared to 5 mm thick images reconstructed with ASIR-V.

CLINICAL IMPACT

Improved image quality of thin-section CT images reconstructed with DLIR has several benefits in clinical practice, such as improved diagnostic performance without radiation dose penalties.

Abdominopelvic CT Image Quality: Evaluation of Thin (0.5-mm) Slices Using Deep Learning Reconstruction

BACKGROUND. Because thick-section images (typically 3-5 mm) have low image noise, radiologists typically use them to perform clinical interpretation, although they may additionally refer to thin-section images (typically 0.5-0.625 mm) for problem solving. Deep learning reconstruction (DLR) can yield thin-section images with low noise. OBJECTIVE. The purpose of this study is to compare abdominopelvic CT image quality between thin-section DLR images and thin- and thick-section hybrid iterative reconstruction (HIR) images. METHODS. This retrospective study included 50 patients (31 men and 19 women; median age, 64 years) who underwent abdominopelvic CT between June 15, 2020, and July 29, 2020. Images were reconstructed at 0.5-mm section using DLR and at 0.5-mm and 3.0-mm sections using HIR. Five radiologists independently performed pairwise comparisons (0.5-mm DLR and either 0.5-mm or 3.0-mm HIR) and recorded the preferred image for subjective image quality measures (scale, -2 to 2). The pooled scores of readers were compared with a score of 0 (denoting no preference). Image noise was quantified using the SD of ROIs on regions of homogeneous liver. RESULTS. For comparison of 0.5-mm DLR images and 0.5-mm HIR images, the median pooled score was 2 (indicating a definite preference for DLR) for noise and overall image quality and 1 (denoting a slight preference for DLR) for sharpness and natural appearance. For comparison of 0.5-mm DLR and 3.0-mm HIR, the median pooled score was 1 for the four previously mentioned measures. These assessments were all significantly different (p < .001) from 0. For artifacts, the median pooled score for both comparisons was 0, which was not significant for comparison with 3.0-mm HIR (p = .03) but was significant for comparison with 0.5-mm HIR (p < .001) due to imbalance in scores of 1 (n = 28) and -1 (slight preference for HIR, n = 1). Noise for 0.5-mm DLR was lower by mean differences of 12.8 HU compared with 0.5-mm HIR and 4.4 HU compared with 3.0-mm HIR (both p < .001). CONCLUSION. Thin-section DLR improves subjective image quality and reduces image noise compared with currently used thin- and thick-section HIR, without causing additional artifacts. CLINICAL IMPACT. Although further diagnostic performance studies are warranted, the findings suggest the possibility of replacing current use of both thin- and thick-section HIR with the use of thin-section DLR only during clinical interpretations.

Complex Relationship Between Artificial Intelligence and CT Radiation Dose

Concerns over need for CT radiation dose optimization and reduction led to improved scanner efficiency and introduction of several reconstruction techniques and image processing-based software. The latest technologies use artificial intelligence (AI) for CT dose optimization and image quality improvement. While CT dose optimization has and can benefit from AI, variations in scanner technologies, reconstruction methods, and scan protocols can lead to substantial variations in radiation doses and image quality across and within different scanners. These variations in turn can influence performance of AI algorithms being deployed for tasks such as detection, segmentation, characterization, and quantification. We review the complex relationship between AI and CT radiation dose.

Radiation Dose Reduction for 80-kVp Pediatric CT Using Deep Learning-Based Reconstruction: A Clinical and Phantom Study

BACKGROUND. Deep learning-based reconstruction (DLR) may facilitate CT radiation dose reduction, but a paucity of literature has compared lower-dose DLR images with standard-dose iterative reconstruction (IR) images or explored application of DLR to low-tube-voltage scanning in children. OBJECTIVE. The purpose of this study was to assess whether DLR can be used to reduce radiation dose while maintaining diagnostic image quality in comparison with hybrid IR (HIR) and model-based IR (MBIR) for low-tube-voltage pediatric CT. METHODS. This retrospective study included children 6 years old or younger who underwent contrast-enhanced 80-kVp CT with a standard-dose or lower-dose protocol. Standard images were reconstructed with HIR, and lower-dose images were reconstructed with HIR, MBIR, and DLR. Size-specific dose estimate (SSDE) was calculated for both protocols. Image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were quantified. Two radiologists independently evaluated noise magnitude, noise texture, streak artifact, edge sharpness, and overall quality. Interreader agreement was assessed, and mean values were calculated. To evaluate task-based object detection performance, a phantom was imaged with 80-kVp CT at six doses (SSDE, 0.6-5.3 mGy). Detectability index (d') was calculated from the noise power spectrum and task-based transfer function. Reconstruction methods were compared. RESULTS. Sixty-five children (mean age, 25.0 ± 25.2 months) who underwent CT with standard- (n = 31) or lower-dose (n = 34) protocol were included. SSDE was 54% lower for the lower-dose than for the standard-dose group (1.9 ± 0.4 vs 4.1 ± 0.8 mGy). Lower-dose DLR and MBIR yielded lower image noise and higher SNR and CNR than standard-dose HIR (p < .05). Interobserver agreement on subjective features ranged from a kappa coefficient of 0.68 to 0.78. The readers subjectively scored noise texture, edge sharpness, and overall quality lower for lower-dose MBIR than for standard-dose HIR (p < .001), though higher for lower-dose DLR than for standard-dose HIR (p < .001). In the phantom, DLR provided higher d' than HIR and MBIR at each dose. Object detectability was greater for 2.0-mGy DLR than for 4.0-mGy HIR for low-contrast (3.67 vs 3.57) and high-contrast (1.20 vs 1.04) objects. CONCLUSION. Compared with IR algorithms, DLR results in substantial dose reduction with preserved or even improved image quality for low-tube-voltage pediatric CT. CLINICAL IMPACT. Use of DLR at 80 kVp allows greater dose reduction for pediatric CT than do current IR techniques.

Detecting hip osteoarthritis on clinical CT: a deep learning application based on 2-D summation images derived from CT

We developed and compared deep learning models to detect hip osteoarthritis on clinical CT. The CT-based summation images, CT-AP, that resemble X-ray radiographs can detect radiographic hip osteoarthritis and in the absence of large training data, a reliable deep learning model can be optimized by combining CT-AP and X-ray images.

INTRODUCTION

In this study, we aimed to investigate the applicability of deep learning (DL) to assess radiographic hip osteoarthritis (rHOA) on computed tomography (CT).

METHODS

The study data consisted of 94 abdominopelvic clinical CTs and 5659 hip X-ray images collected from Cohort Hip and Cohort Knee (CHECK). The CT slices were sequentially summed to create radiograph-like 2-D images named CT-AP. X-ray and CT-AP images were classified as rHOA if they had osteoarthritic changes corresponding to Kellgren-Lawrence grade 2 or higher. The study data was split into 55% training, 30% validation, and 15% test sets. A pretrained ResNet18 was optimized for a classification task of rHOA vs. no-rHOA. Five models were trained using (1) X-rays, (2) downsampled X-rays, (3) combination of CT-AP and X-ray images, (4) combination of CT-AP and downsampled X-ray images, and (5) CT-AP images.

RESULTS

Amongst the five models, Model-3 and Model-5 performed best in detecting rHOA from the CT-AP images. Model-3 detected rHOA on the test set of CT-AP images with a balanced accuracy of 82.2% and was able to discriminate rHOA from no-rHOA with an area under the receiver operating characteristic curve (ROC AUC) of 0.93 [0.75-0.99]. Model-5 detected rHOA on the test set at a balanced accuracy of 82.2% and classified rHOA from no-rHOA with an ROC AUC of 0.89 [0.67-0.97].

CONCLUSION

CT-based summation images that resemble radiographs can be used to detect rHOA. In addition, in the absence of large training data, a reliable DL model can be optimized by combining CT-AP and X-ray images.

Deep Learning-based Reconstruction for Lower-Dose Pediatric CT: Technical Principles, Image Characteristics, and Clinical Implementations

Optimizing the CT acquisition parameters to obtain diagnostic image quality at the lowest possible radiation dose is crucial in the radiosensitive pediatric population. The image quality of low-dose CT can be severely degraded by increased image noise with filtered back projection (FBP) reconstruction. Iterative reconstruction (IR) techniques partially resolve the trade-off relationship between noise and radiation dose but still suffer from degraded noise texture and low-contrast detectability at considerably low-dose settings. Furthermore, sophisticated model-based IR usually requires a long reconstruction time, which restricts its clinical usability. With recent advances in artificial intelligence technology, deep learning-based reconstruction (DLR) has been introduced to overcome the limitations of the FBP and IR approaches and is currently available clinically. DLR incorporates convolutional neural networks-which comprise multiple layers of mathematical equations-into the image reconstruction process to reduce image noise, improve spatial resolution, and preserve preferable noise texture in the CT images. For DLR development, numerous network parameters are iteratively optimized through an extensive learning process to discriminate true attenuation from noise by using low-dose training and high-dose teaching image data. After rigorous validations of network generalizability, the DLR engine can be used to generate high-quality images from low-dose projection data in a short reconstruction time in a clinical environment. Application of the DLR technique allows substantial dose reduction in pediatric CT performed for various clinical indications while preserving the diagnostic image quality. The authors present an overview of the basic concept, technical principles, and image characteristics of DLR and its clinical feasibility for low-dose pediatric CT. ^©RSNA, 2021.

The Feasibility of Contrast-to-Noise Ratio on Measurements to Evaluate CT Image Quality in Terms of Low-Contrast Detailed Detectability

Background: To evaluate contrast-to-noise ratio (CNR) measurements in assessing image quality, in the context of the detectability performance of low-contrast detail (LCD), in computed tomography (CT) images, since exposure to elevated ionising-type radiation is considered to present excessive carcinogenic risk, whilst also causing distress in study subjects. Methods: An LCD phantom module (CTP515) was utilised in the study. Three dissimilar contrast items were used to analyse the ramifications of the proportions of an object on the CNR. Three multidetector CT (MDCT) scanners were used, with 16-MDCT, 64-MDCT and 80-MDCT frameworks, respectively. The CT scans were recreated using three dissimilar remaking algorithms—soft, standard and lung. The effects exerted on the CNR by various remodelling algorithms, as well as the contrast of various objects along with the size of the objects, were explored. The Hounsfield units of each chosen object (one unit representing the outer portion of the object) and the background and the standard deviation of the noise parameter were quantified, and algorithms were developed using MATLAB. Results: The CNR information was greatly influenced by changing the image recreation calculations and was very much increased in the soft-tissue recreation images using 16-MDCT and 64-MDCT. The CNR information was also increased more in the optimum recreation images than in the reproduced images from the computational procedure used in the 80-MDCT. The results did not show any remarkable contrasts in the CNR values between the different object sizes. Overall, a higher kVp produced an improved CNR in all the CT scanners. In particular, there were prominent upgrades in the CNR information when the kVp was increased from 80 to 120. Higher mAs levels gave better CNR values overall, especially for greater section thicknesses. Based on the CNR estimations, the 64-MDCT provided the best correlation among the CT scanners. Conclusions: The objective LCD appraisal method, based on CNR measurements, was confirmed as being useful for checking the different impacts of kVp, mAs and section thickness on the nature of the picture. This procedure was similarly viable in assessing the impacts of the different reconstruction calculations and the different differentiation questions on the nature of the image.

Optimizing radiation dose parameters in MDCT arthrography of the shoulder: illustration of basic concepts in a cadaveric study

OBJECTIVE

To determine in a cadaveric study the lowest achievable radiation dose and optimal tube potential generating diagnostic image quality in multidetector computed tomography (MDCT) arthrography of the shoulder.

MATERIALS AND METHODS

Six shoulders from three human cadavers were scanned using a 256-MDCT system after intra-articular injection of diluted iodinated contrast material. Using six decreasing radiation dose levels (CTDI_vol: 20, 15, 10, 8, 6, and 4 mGy) and for each dose level, four decreasing tube potentials (140, 120, 100, and 80 kVp), image noise and contrast-to-noise ratio (CNR) were measured. Two independent and blinded observers assessed the overall diagnostic image quality, subjective amount of noise, and severity of artifacts according to a four-point scale. Influence of those MDCT data acquisition parameters on objective and subjective image quality was analyzed using the Kruskal-Wallis and Wilcoxon signed-rank tests, and pairwise comparisons were performed.

RESULTS

Multidetector CT protocols with radiation doses of 15 mGy or higher, combined with tube potentials of 100 kVp or higher, were equivalent in CNR to the reference 20 mGy-140 kVp protocol (all p ≥ 0.054). Above a CTDI_vol of 10 mGy and a tube potential of 120 kVp, all protocols generated diagnostic image quality and subjective noise equivalent to the 20 mGy-140 kVp protocol (all p ≥ 0.22).

CONCLUSIONS

Diagnostic image quality in MDCT arthrography of the shoulder can be obtained with a radiation dose of 10 mGy at an optimal tube potential of 120 kVp, corresponding to a reduction of up to 50% compared with standard-dose protocols, and as high as 500% compared with reported protocols in the literature.

Diagnostic accuracy of low-dose versus ultra-low-dose CT for lumbar disc disease and facet joint osteoarthritis in patients with low back pain with MRI correlation

OBJECTIVE

To compare the image quality, radiation dose, and diagnostic performance between low-dose (LD) and ultra-low-dose (ULD) lumbar-spine (L-spine) CT with iterative reconstruction (IR) for patients with chronic low back pain (LBP).

METHODS

In total, 260 patients with chronic LBP who underwent L-spine CT between November 2015 and September 2016 were prospectively enrolled. Of these, 143 underwent LD-CT with IR and 117 underwent ULD-CT with IR. The patients were divided according to their body mass index (BMI) into BMI1 (<22.9 kg/m²), BMI2 (23.0-24.9 kg/m²), and BMI3 (≥25 kg/m²) groups. Two blinded radiologists independently evaluated the signal-to-noise ratio (SNR), qualitative image quality, and final diagnoses (lumbar disc disease and facet joint osteoarthritis). L-spine MRIs interpreted by consensus were used as the reference standard. All data were statistically analyzed.

RESULTS

ULD protocol showed significantly lower SNR for all patients (p < 0.001) except the vertebral bodies and lower qualitative image quality for BMI3 patients (p ≤ 0.033). There was no statistically significant difference between ULD (sensitivity, 95.1-98.1%; specificity, 92.5-98.7%; accuracy, 94.6-98.0%) and LD protocols (sensitivity, 95.6-100%; specificity, 95.5-98.9%; accuracy, 97.4-98.1%), (all p≥0.1) in the BMI1 and BMI2; while dose was 60-68% lower with the ULD protocol. Interobserver agreements were excellent or good with regard to image quality and final diagnoses.

CONCLUSIONS

For the BM1 and BMI2 groups, ULD-CT provided an acceptable image quality and exhibited a diagnostic accuracy similar to that of LD-CT. These findings suggest that it is a useful diagnostic tool for patients with chronic LBP who exhibit a BMI of <25 kg/m².

Impact of iterative reconstruction vs. filtered back projection on image quality in 320-slice CT coronary angiography: Insights from the CORE320 multicenter study

Iterative reconstruction has been shown to reduce image noise compared with traditional filtered back projection with quantum denoising software (FBP/QDS+) in CT imaging but few comparisons have been made in the same patients without the influence of interindividual factors. The objective of this study was to investigate the impact of adaptive iterative dose reduction in 3-dimensional (AIDR 3D) and FBP/QDS+-based image reconstruction on image quality in the same patients.We randomly selected 100 patients enrolled in the coronary evaluation using 320-slice CT study who underwent CT coronary angiography using prospectively electrocardiogram triggered image acquisition with a 320-detector scanner. Both FBP/QDS+ and AIDR 3D reconstructions were performed using original data. Studies were blindly analyzed for image quality by measuring the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Image quality was assessed qualitatively using a 4-point scale.Median age was 63 years (interquartile range [IQR]: 56-71) and 72% were men, median body mass index 27 (IQR: 24-30) and median calcium score 222 (IQR: 11-644). For all regions of interest, mean image noise was lower for AIDR 3D vs. FBP/QDS+ (31.69 vs. 34.37, P ≤ .001). SNR and CNR were significantly higher for AIDR 3D vs. FBP/QDS+ (16.28 vs. 14.64, P < .001 and 19.21 vs. 17.06, P < .001, respectively). Subjective (qualitative) image quality scores were better using AIDR 3D vs. FBP/QDS+ with means of 1.6 and 1.74, respectively (P ≤ .001).Assessed in the same individuals, iterative reconstruction decreased image noise and raised SNR/CNR as well as subjective image quality scores compared with traditional FBP/QDS+ in 320-slice CT coronary angiography at standard radiation doses.

Impact of CT parameters on the physical quantities related to image quality for two MDCT scanners using the ACR accreditation phantom: A phantom study

INTRODUCTION

To evaluate the image quality provided by MDCT scanners using an ACR phantom an to find out the relationship between CT parameters and physical quantities related to image quality.

METHODS

A GE Lightspeed VCT and a GE Lightspeed Pro 16 are used. The ACR phantom consists of four modules for evaluating physical parameters. The image quality parameters, such as CT number, linearity, CNR, image uniformity, SNR and at least spatial resolution using MTF, by different sets of image acquisition protocols (IAPs) are characterized. The influences of the IAPs on the physical quantities are also discussed.

RESULTS

The CT numbers behaved linearly relative to material density for all tube voltages. The impact of the tube current on the CT numbers is neglectable. However, the variation of the tube current reflects in the CT number uncertainties. The CNR are altered by changing the IAPs. 50% MTF decreases from 6.2 to 3.6 lp/cm and from 6.5 to 3.7 lp/cm using Lung and Soft kernel for the Lightspeed VCT and Lightspeed Pro 16 scanner, respectively.

CONCLUSION

The dependence of the image quality parameters on reconstruction kernels, tube peak voltages, tube currents and the slice thicknesses has been discussed. The tube peak voltage has the most influence on the CT numbers. The results indicate that the reconstruction kernel has the main impact on the spatial resolution. The spatial resolution dependence on the tube voltages, tube currents and slice thicknesses can be ignored.

Ultra-high resolution C-Arm CT arthrography of the wrist: Radiation dose and image quality compared to conventional multidetector computed tomography

OBJECTIVE

Objective of this phantom and cadaveric study was to compare the effective radiation dose (ED) and image quality (IQ) between C-arm computed tomography (CACT) using an ultra-high resolution 1×1 binning with a standard 16-slice CT (MDCT) arthrography of the wrist.

METHODS

ED was determined with thermoluminescence dosimetry using an anthropomorphic phantom and different patient positions. Imaging was conducted in 10 human cadaveric wrists after tri-compartmental injection of diluted iodinated contrast material and a wire phantom. IQ of MDCT was compared with CACT reconstructed with a soft (CACT1) and sharp (CACT2) kernel. High and low contrast resolution was determined. Three radiologists assessed IQ of wrist structures and occurrence of image artifacts using a 5-point Likert scale.

RESULTS

ED of MDCT was comparable to standard CACT (4.3μSv/3.7μSv). High contrast resolution was best for CACT2, decreased to CACT1 and MDCT. Low contrast resolution increased between CACT2 and MDCT (P<0.001). IQ was best for CACT2 (1.3±0.5), decreased to CACT1 (1.9±0.6) and MDCT (3.5±0.6). Non-compromising artifacts were only reported for CACT.

CONCLUSIONS

The results of this phantom and cadaveric study indicate that ultra-high resolution C-Arm CT arthrography of the wrist bears the potential to outperform MDCT arthrography in terms of image quality and workflow at the cost of mildly increasing image artifacts while radiation dose to the patient is comparably low for both, MDCT and C-Arm CT.

Chest CT at a dose below 0.3 mSv: impact of iterative reconstruction on image quality and lung analysis

BACKGROUND

In chest computed tomography (CT), iterative reconstruction (IR) algorithms maintain diagnostic image quality (IQ) while significantly reducing the dose.

PURPOSE

To evaluate the impact of IR on IQ of chest CT at effective doses below 0.3 mSv.

MATERIAL AND METHODS

Twenty chest CT scans performed at effective dose below 0.3 mSv (CT1) were reconstructed varying three parameters: filtered back-projection and IR iDose(4) algorithms; 512 × 512 and 768 × 768 matrices; and sharp and soft kernels, thus generating eight series per patient. The qualitative evaluation of the IQ was performed by ranking series from 1 to 8 (8 corresponding to the highest rank) which was subsequently compared to quantitative assessment of IQ by using an appropriated merit formula. Intra- and inter-reader IQ ranking reliability was also evaluated using Cohen's kappa. Analysis of lung findings was finally compared between the best CT1 series and the reference CT (CT0).

RESULTS

The best series in terms of qualitative and quantitative IQ was obtained using IR, 512(2) matrix and soft kernel. The best CT1 series detected nodules greater than 4 mm with an almost perfect match with CT0.

CONCLUSION

Chest CT performed at effective doses below 0.3 mSv may be used to confidently diagnose lesions greater than 4 mm using iDose(4), soft kernel and 512 × 512 matrix.

Improved non-calcified plaque delineation on coronary CT angiography by sonogram-affirmed iterative reconstruction with different filter strength and relationship with BMI

PURPOSE

To prospectively compare non-calcified plaque delineation and image quality of coronary computed tomography angiography (CCTA) obtained with sinogram-affirmed iterative reconstruction (IR) with different filter strengths and filtered back projection (FBP).

METHODS

A total of 57 patients [28.1% females; body mass index (BMI) 29.2±6.5 kg/m(2)] were investigated. CCTA was performed using 128-slice dual-source CT. Images were reconstructed with standard FBP and sinogram-affirmed IR using different filter strength (IR-2, IR-3, IR-4) (SAFIRE, Siemens, Germany). Image quality of CCTA and a non-calcified plaque outer border delineation score were evaluated by using a 5-scale score: from 1= poor to 5= excellent. Image noise, contrast-to-noise ratio (CNR) of aortic root, left main (LM) and right coronary artery, and the non-calcified plaque delineation were quantified and compared among the 4 image reconstructions, and were compared between different BMI groups (BMI <28 and ≥28). Statistical analyses included one-way analysis of variance (ANOVA), least significant difference (LSD) and Kruskal-Wallis test.

RESULTS

There were 71.9% patients in FBP, 96.5% in IR-2, 96.5% in IR-3 and 98.2% in IR-4 who had overall CCTA image quality ≥3, and there were statistical differences in CCTA exam image quality score among those groups, respectively (P<0.001). Sixty-one non-calcified plaques were detected by IR-2 to IR-4, out of those 11 (18%) were missed by FBP. Plaque delineation score increased constantly from FBP (2.7±0.4) to IR-2 (3.2±0.3), to IR-3 (3.5±0.3) up to IR-4 (4.0±0.4), while CNRs of the non-calcifying plaque increased and image noise decreased, respectively. Similarly, CNR of aortic root, LM and right coronary artery improved and image noise declined from FBP to IR-2, IR-3 and IR-4. There were no significant differences of image quality and plaque delineation score between low and high BMI groups within same reconstruction (all P>0.05). Significant differences in image quality and plaque delineation scores among different image reconstructions both in low and high BMI groups (all P<0.001) were found. I4f revealed the highest image quality and plaque delineation score.

CONCLUSIONS

IR offers improved image quality and non-calcified plaque delineation as compared with FBP, especially if BMI is increasing. Importantly, 18% of non-calcified plaques were missed with FBP. IR-4 shows the best image quality score and plaque delineation score among the different IR-filter strength.

Comparison of image quality of shoulder CT arthrography conducted using 120 kVp and 140 kVp protocols

OBJECTIVE

To compare the image quality of shoulder CT arthrography performed using 120 kVp and 140 kVp protocols.

MATERIALS AND METHODS

Fifty-four CT examinations were prospectively included. CT scans were performed on each patient at 120 kVp and 140 kVp; other scanning parameters were kept constant. Image qualities were qualitatively and quantitatively compared with respect to noise, contrast, and diagnostic acceptability. Diagnostic acceptabilities were graded using a one to five scale as follows: 1, suboptimal; 2, below average; 3, acceptable; 4, above average; and 5, superior. Radiation doses were also compared.

RESULTS

Contrast was better at 120 kVp, but noise was greater. No significant differences were observed between the 120 kVp and 140 kVp protocols in terms of diagnostic acceptability, signal-to-noise ratio, or contrast-to-noise ratio. Lowering tube voltage from 140 kVp to 120 kVp reduced the radiation dose by 33%.

CONCLUSION

The use of 120 kVp during shoulder CT arthrography reduces radiation dose versus 140 kVp without significant loss of image quality.

Low-dose multidetector computed tomography of the cervical spine: optimization of iterative reconstruction strength levels

BACKGROUND

Iterative reconstruction (IR) techniques reduce image noise in multidetector computed tomography (MDCT) imaging. They can therefore be used to reduce radiation dose while maintaining diagnostic image quality nearly constant. However, CT manufacturers offer several strength levels of IR to choose from.

PURPOSE

To determine the optimal strength level of IR in low-dose MDCT of the cervical spine.

MATERIAL AND METHODS

Thirty consecutive patients investigated by low-dose cervical spine MDCT were prospectively studied. Raw data were reconstructed using filtered back-projection and sinogram-affirmed IR (SAFIRE, strength levels 1 to 5) techniques. Image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured at C3-C4 and C6-C7 levels. Two radiologists independently and blindly evaluated various anatomical structures (both dense and soft tissues) using a 4-point scale. They also rated the overall diagnostic image quality using a 10-point scale.

RESULTS

As IR strength levels increased, image noise decreased linearly, while SNR and CNR both increased linearly at C3-C4 and C6-C7 levels (P < 0.001). For the intervertebral discs, the content of neural foramina and dural sac, and for the ligaments, subjective image quality scores increased linearly with increasing IR strength level (P ≤ 0.03). Conversely, for the soft tissues and trabecular bone, the scores decreased linearly with increasing IR strength level (P < 0.001). Finally, the overall diagnostic image quality scores increased linearly with increasing IR strength level (P < 0.001).

CONCLUSION

The optimal strength level of IR in low-dose cervical spine MDCT depends on the anatomical structure to be analyzed. For the intervertebral discs and the content of neural foramina, high strength levels of IR are recommended.

Effect of reduced radiation exposure and iterative reconstruction on detection of low-contrast low-attenuation lesions in an anthropomorphic liver phantom: an 18-reader study

PURPOSE

To measure the effect of reduced radiation exposure on low-contrast low-attenuation liver lesion detection in an anthropomorphic abdominal phantom by using filtered back projection (FBP) and sinogram-affirmed iterative reconstruction.

MATERIALS AND METHODS

Eighteen radiologists blinded to phantom and study design interpreted randomized image data sets that contained 36 spherical simulated liver lesions of three sizes and three attenuation differences (5-mm diameter: 12, 18, and 24 HU less than the 90-HU background attenuation of the simulated liver insert; 10- and 15-mm diameter: 6, 12, and 18 HU less than the 90-HU background attenuation) scanned with four discrete exposure settings and reconstructed by using FBP and sinogram-affirmed iterative reconstruction. Response assessment included region-level lesion presence or absence on a five-point diagnostic confidence scale. Statistical evaluation included multireader multicase receiver operating characteristic curve analysis, with nonparametric methods and noninferiority analysis at a margin of -0.10.

RESULTS

Pooled accuracy at 75% exposure for both FBP and sinogram-affirmed iterative reconstruction was noninferior to 100% exposure (P = .002 and P < .001, respectively). Subsequent exposure reductions resulted in a significant decrease in accuracy. When the smallest (5-mm-diameter) lesions were excluded from analysis, sinogram-affirmed iterative reconstruction was superior to FBP at 100% exposure (P = .011), and sinogram-affirmed iterative reconstruction at 25% and 50% exposure reduction was noninferior to FBP at 100% exposure (P ≤ .013). Reader confidence was greater with sinogram-affirmed iterative reconstruction than with FBP for 10- and 15-mm lesions (2.94 vs 2.76 and 3.62 vs 3.52, respectively).

CONCLUSION

In this low-contrast low-attenuation liver lesion model, a 25% exposure reduction maintained noninferior diagnostic accuracy. However, detection was inferior with each subsequent exposure reduction, regardless of reconstruction method. Sinogram-affirmed iterative reconstruction and FBP performed equally well at modest exposure reduction (25%-50%). Readers had higher confidence levels with sinogram-affirmed iterative reconstruction for the 10- and 15-mm lesions.

Emergency department CT screening of patients with nontraumatic neurological symptoms referred to the posterior fossa: comparison of thin versus thick slice images

Evaluation of the posterior fossa (PF) on 5-mm-thick helical CT images (current default) has improved diagnostic accuracy compared to 5-mm sequential CT images; however, 5-mm-thick images may not be ideal for PF pathology due to volume averaging of rapid changes in anatomy in the Z-direction. Therefore, we sought to determine if routine review of 1.25-mm-thin helical CT images has superior accuracy in screening for nontraumatic PF pathology. MRI proof of diagnosis was obtained within 6 h of helical CT acquisition for 90 consecutive ED patients with, and 88 without, posterior fossa lesions. Helical CT images were post-processed at 1.25 and 5-mm-axial slice thickness. Two neuroradiologists blinded to the clinical/MRI findings reviewed both image sets. Interobserver agreement and accuracy were rated using Kappa statistics and ROC analysis, respectively. Of the 90/178 (51 %) who were MR positive, 60/90 (66 %) had stroke and 30/90 (33 %) had other etiologies. There was excellent interobserver agreement (κ > 0.97) for both thick and thin slice assessments. The accuracy, sensitivity, and specificity for 1.25-mm images were 65, 44, and 84 %, respectively, and for 5-mm images were 67, 45, and 85 %, respectively. The diagnostic accuracy was not significantly different (p > 0.5). In this cohort of patients with nontraumatic neurological symptoms referred to the posterior fossa, 1.25-mm-thin slice CT reformatted images do not have superior accuracy compared to 5-mm-thick images. This information has implications on optimizing resource utilizations and efficiency in a busy emergency room. Review of 1.25-mm-thin images may help diagnostic accuracy only when review of 5-mm-thick images as current default is inconclusive.

LOCAT (low-dose computed tomography for appendicitis trial) comparing clinical outcomes following low- vs standard-dose computed tomography as the first-line imaging test in adolescents and young adults with suspected acute appendicitis: study protocol for a randomized controlled trial

BACKGROUND

Computed tomography is widely used to diagnose acute appendicitis. Many adolescents and young adults are exposed to the associated radiation. A recent single-institution trial has reported promising results for low-dose computed tomography; however, this technique has not yet been widely adopted. LOCAT (low-dose computed tomography for appendicitis trial), a multi-institution randomized controlled non-inferiority trial, aims to compare low-dose computed tomography and standard-dose computed tomography as the first-line imaging tests for adolescents and young adults, and therefore to test the generalizability of the previous single-institution trial results.

METHODS/DESIGN

Participants with suspected appendicitis are randomly assigned to either the low-dose group (with a typical effective dose of 2 mSv) or the standard-dose group (as used in normal practice at each participating site, typically 8 mSv). The primary end point is the negative appendectomy rate (the percentage of the number of uninflamed appendices that were removed among all non-incidental appendectomies), which is a consequence of false-positive diagnoses, with a non-inferiority margin of 4.5 percentage points. The key secondary end point is the appendiceal perforation rate, which is a consequence of delayed (or false-negative) diagnoses. Participant recruitment will be continued until the number of non-incidental appendectomies for each group exceeds 444. The total number of expected participants approximates 3,000, including those not undergoing appendectomy.

DISCUSSION

In addition to the study protocol, we elaborate on several challenging or potentially debatable components of the study design, including the broad eligibility criteria, choice of the primary end point, potential effect of using advanced imaging techniques on study results, determining and adjusting the radiation doses, ambiguities in reference standards, rationale for the non-inferiority margin, use of the intention-to-treat approach and difficulties in defining adverse events.

TRIAL REGISTRATION

ClinicalTrials.gov NCT01925014.

Development and evaluation of a software tool for the generation of virtual liver lesions in multidetector-row CT datasets

RATIONALE AND OBJECTIVES

Development and evaluation of a software tool for the insertion of simulated hypodense liver lesions in multidetector-row computed tomography (CT) datasets.

MATERIALS AND METHODS

Forty software-generated hypodense liver lesions were inserted at random locations in 20 CT datasets by using the "alpha blending" technique and compared with 40 real metastatic lesions. The location, diameter (5-20 mm) and density of the simulated lesions were individually adjusted to closely resemble real lesions in each patient. Three blinded readers evaluated all 80 lesions twice in a 2-week interval using a five-point Likert confidence scale under standardized conditions. Nonparametric tests were used to statistically evaluate possible differences in scoring between real and simulated lesions. The correctness of the observer rating for real and simulated lesions was compared to chance distribution using the chi-squared statistics. The inter- and intraobserver variability was determined using Kendall's coefficient of concordance.

RESULTS

The observer study did not reveal significant differences between the scoring for real versus simulated lesions for any of the readers (P > .05). The distribution of correct and false scoring of the lesions was not significantly different from chance distribution (P > .05). Inter- and intraobserver agreement was poor (Kendall W coefficient = 0.12/0.13).

CONCLUSION

The proposed algorithm is suitable for creating realistic virtual liver lesions in CT datasets.

Influence of sinogram affirmed iterative reconstruction of CT data on image noise characteristics and low-contrast detectability: an objective approach

Objectives

To utilize a novel objective approach combining a software phantom and an image quality metric to systematically evaluate the influence of sinogram affirmed iterative reconstruction (SAFIRE) of multidetector computed tomography (MDCT) data on image noise characteristics and low-contrast detectability (LCD).

Materials and Methods

A low-contrast and a high-contrast phantom were examined on a 128-slice scanner at different dose levels. The datasets were reconstructed using filtered back projection (FBP) and SAFIRE and virtual low-contrast lesions (-20HU) were inserted. LCD was evaluated using the multiscale structural similarity index (MS-SIM*). Image noise texture and spatial resolution were objectively evaluated.

Results

The use of SAFIRE led to an improvement of LCD for all dose levels and lesions sizes. The relative improvement of LCD was inversely related to the dose level, declining from 208%(±37%), 259%(±30%) and 309%(±35%) at 25mAs to 106%(±6%), 119%(±9%) and 123%(±8%) at 200mAs for SAFIRE filter strengths of 1, 3 and 5 (p<0.05). SAFIRE reached at least the LCD of FBP at a relative dose of 50%. There was no statistically significant difference in spatial resolution. The use of SAFIRE led to coarser image noise granularity.

Conclusion

A novel objective approach combining a software phantom and the MS-SSIM* image quality metric was used to analyze the detectability of virtual low-contrast lesions against the background of image noise as created using SAFIRE in comparison to filtered back-projection. We found, that image noise characteristics using SAFIRE at 50% dose were comparable to the use of FBP at 100% dose with respect to lesion detectability. The unfamiliar imaging appearance of iteratively reconstructed datasets may in part be explained by a different, coarser noise characteristic as demonstrated by a granulometric analysis.

Influence of multiplanar reformations on low-contrast performance in thin-collimated multidetector computed tomography

OBJECTIVES

: To analyze the influence of multiplanar reformations of thin-collimated multidetector computed tomography datasets on low-contrast performance.

MATERIALS AND METHODS

: A low-contrast phantom simulating focal hypodense lesions (-20 HU object contrast) was scanned on a 64-slice spiral CT scanner at 4 different dose levels (25 mAs, 50 mAs, 100 mAs, 200 mAs, and no dose modulation). Other scanner parameters were as follows: tube voltage = 120 kVp, rotation time = 0.8 s, reconstructed slice thickness = 0.625 mm, reconstruction interval = 0.5 mm, reconstruction kernel = standard. Coronal reformations were created using the open-source software OsiriX. A sliding-thin-slab (STS) averaging algorithm was applied to each axial and each reformatted dataset with an increasing slab thickness from 1 to 20 slices. The low-contrast performance of all datasets was calculated semiautomatically using a reader-independent, statistical approach and is expressed as the visibility index. The results were analyzed for differences between the coronal reformations and the original axial datasets. In addition, the statistical approach used herein was validated against a reader study.

RESULTS

: The visibility index of the coronal reformatted datasets over all lesion sizes was inferior when compared with the original axial datasets and reached 75.4% (±11.7%), 79.9% (±16.3%), 79.4% (±5.5%), and 93.7% (±14.6%) for dose levels of 25, 50, 100, and 200 mAs, respectively. The overall mean low-contrast performance was 82.1% of the axial dataset (P < 0.05, except for 200 mAs). The deterioration of low-contrast performance was inversely correlated with lesion size (R = 0.91). The use of the STS averaging algorithm significantly improved image quality for all datasets (112.6%-180.2%) with the beneficial effect being stronger for the coronal reformations. There was no statistically significant difference in the evaluation of low-contrast performance between the statistical approach and the ready study.

CONCLUSION

: Coronal reformations of thin-collimated multidetector computed tomography datasets show a significant reduction of low-contrast performance when compared with the original axial dataset, especially in high noise data. The use of an STS averaging algorithm had a significant benefit for both, coronal and axial orientations. The effect was more pronounced with coronal reformations and should be routinely applied to improve image quality.

The obese emergency patient: imaging challenges and solutions

The dramatic rise in the prevalence of obesity among children and adults in the United States over the last several decades has brought several new challenges to the delivery of healthcare. The increased utilization of and dependence on imaging for accurate and timely diagnosis has placed the radiology department in a unique position in the provision of care for the obese emergency patient. Radiology practices must be cognizant of the imaging challenges presented by the obese patient and adjust their imaging algorithms accordingly to optimize all types of diagnostic studies. The article systematically reviews common pitfalls and offers methods to improve image quality when using radiography, ultrasonography, and computed tomography to image the obese patient population.

Calcium scoring with prospectively ECG-triggered CT: using overlapping datasets generated with MPR decreases inter-scan variability

OBJECTIVE

To examine the feasibility of reducing the inter-scan variability of prospectively ECG-triggered calcium-scoring scans by using overlapping 3-mm datasets generated from multiplanar reformation (MPR) instead of non-overlapping 3-mm or 1.5-mm datasets.

PATIENTS AND METHODS

Seventy-five women (59-79 years old) underwent two sequential prospectively ECG-triggered calcium-scoring scans with 16 mm×1.5mm collimation in one session. Between the two scans patients got off and on the table. We performed calcium scoring (Agatston and mass scores) on the following datasets: contiguous 3-mm sections reconstructed from the raw data (A), contiguous 3-mm sections from MPR (B), overlapping 3-mm sections from MPR (C) and contiguous 1.5-mm sections from the raw data (D). To determine the feasibility of the MPR approach, we compared MPR (B) with direct raw data reconstruction (A). Inter-scan variability was calculated for each type of dataset (A-D).

RESULTS

Calcium scores ranged from 0 to 1455 (Agatston) and 0 to 279 mg (mass) for overlapping 3-mm sections (C). Calcium scores (both Agatston and mass) were nearly identical for MPR (B) and raw data approaches (A), with inter-quartile ranges of 0-1% for inter-scan variability. Median inter-scan variability with contiguous 3-mm sections (B) was 13% (Agatston) and 11% (mass). Median variability was reduced to 10% (Agatston and mass) with contiguous 1.5-mm sections (D) and to 8% (Agatston) and 7% (mass) with overlapping 3-mm MPR (A).

CONCLUSION

Calcium scoring on MPR yields nearly identical results to calcium scoring on images directly reconstructed from raw data. Overlapping MPR from prospectively ECG-triggered scans improve inter-scan variability of calcium scoring without increasing patient radiation dose.