Coronary stent imaging in photon counting computed Tomography: Optimization of reconstruction kernels in a phantom.

PURPOSE: Imaging stents and in-stent stenosis remains a challenge in coronary computed tomography angiography (CCTA). In comparison to conventional Computed Tomography, Photon Counting CT (PCCT) provides decisive clinical advantages, among other things by providing low dose ultra-high resolution imaging of coronary arteries. This work investigates the image quality in CCTA using clinically established kernels and those optimized for the imaging of cardiac stents in PCCT, both for in-vitro stent imaging in 400 µm standard resolution mode (SRM) and 200 µm Ultra High Resolution Mode (UHR). METHODS: Based on experimental scans, vascular reconstruction kernels (Bv56, Bv64, Bv72) were optimized. In an established phantom, 10 different coronary stents with 3 mm diameter were scanned in the first clinically available PCCT. Scans were reconstructed with clinically established and optimized kernels. Four readers measured visible stent lumen, performed ROI-based density measurements and rated image quality. RESULTS: Regarding the visible stent lumen, UHR is significantly superior to SRM (p < 0.001). In all levels, the optimized kernels are superior to the clinically established kernels (p < 0.001). One optimized kernel showed a significant reduction of noise compared to the clinically established kernels. Overall image quality is improved with optimized kernels. CONCLUSIONS: In a phantom study PCCT UHR with optimized kernels for stent imaging significantly improves the ability to assess the in-stent lumen of small cardiac stents. We recommend using UHR with an optimized sharp vascular reconstruction kernel (Bv72uo) for imaging of cardiac stent.

Deep learning-based harmonization of CT reconstruction kernels towards improved clinical task performance.

OBJECTIVES: To develop a deep learning-based harmonization framework, assessing whether it can improve performance of radiomics models given different kernels in different clinical tasks and additionally generalize to mitigate the effects of new/unobserved kernels on radiomics features. METHODS: Patient data with 2 reconstruction kernels and phantom data with 22 reconstruction kernels were included. Eighty-five patients were studied for lymph node metastasis (LNM) prediction, and 164 patients for differential diagnosis between lung cancer (LC) and pulmonary tuberculosis (TB). Two convolutional neural network (CNN) models were developed to convert images (i) from B70f to B30f (CNNa) and (ii) from B30f to B70f (CNNb). Model performance between the two kernels was evaluated using AUC and compared with other well-known harmonization methods. Patient-normalized feature difference (PNFD) was used to identify the incompatible kernels (i.e., kernel with median PNFD > 1) with baseline (B30f/B70f), and measure the ability of the CNN models to convert the non-comparable kernels. RESULTS: For LC versus pulmonary TB diagnosis, AUCs of CNNa vs. others were 0.85 vs. 0.54-0.74 (p = 0.0001-0.0003), and for CNNb vs. others: 0.87 vs. 0.54-0.86 (p = 0.0001-0.55). For LNM prediction, AUCs of CNNa vs. others were 0.68 vs. 0.56-0.61 (p = 0.10-0.39), and for CNNb vs. others: 0.78 vs. 0.70-0.73 (p = 0.07-0.40). After CNN harmonization, 17 of 20 (85%) of investigated unknown kernels produced comparable radiomics feature values relative to baseline (median PNFD from 1.10-2.31 to 0.23-1.13). CONCLUSION: The CNN harmonization effectively improved performance of radiomics models between reconstruction kernels in different clinical tasks, and reduced feature differences between unknown kernels vs. baseline. KEY POINTS: • The soft (B30f) and sharp (B70f) kernels strongly affect radiomics reproducibility and generalizability. • The convolutional neural network (CNN) harmonization methods performed better than location-scale (ComBat and centering-scaling) and matrix factorization harmonization methods (based on singular value decomposition (SVD) and independent component analysis (ICA)) in both clinical tasks. • The CNN harmonization methods improve feature reproducibility not only between specific kernels (B30f and B70f) from the same scanner, but also between unobserved kernels from different scanners of different vendors.

Influence of monitor display resolution and displayed image size on the spatial resolution of ultra-high-resolution CT images: a phantom study.

To determine the optimal display conditions for ultra-high-resolution computed tomography (UHRCT) images in clinical practice, this study investigated the effects of liquid-crystal display (LCD) resolution and displayed image size on the spatial resolution of phantom images acquired using a UHRCT system. A phantom designed to evaluate the high-contrast resolution was scanned. The scan data were reconstructed into four types of UHRCT image series consisting of the following possible combinations: two types of reconstruction kernels on the filtered back-projection method (for the lung and mediastinum) and two types of matrix sizes (10242 and 20482). These images were displayed under eight types of display conditions: three image sizes displayed on a 2-megapixel (MP) and 3-MP color LCD and two image sizes on an 8-MP color LCD. A total of 32 samples (four image series × eight display conditions) were evaluated by eight observers for high-contrast resolution. The high-contrast resolution of the displayed UHRCT images was significantly affected by the displayed image size, although the largest (full-screen) displayed image size did not necessarily show the maximum high-contrast resolution. When the images were displayed in the full-screen size, LCD resolution affected the high-contrast resolution of only the 20482-matrix-size images reconstructed using the lung kernel. In conclusion, the spatial resolution of UHRCT images may be affected by LCD resolution and displayed image size. To optimize the clinical display conditions for UHRCT images, it is necessary to adopt an LCD with an adequate resolution for each viewing situation.

CT Reconstruction Kernels and the Effect of Pre- and Post-Processing on the Reproducibility of Handcrafted Radiomic Features.

Handcrafted radiomics features (HRFs) are quantitative features extracted from medical images to decode biological information to improve clinical decision making. Despite the potential of the field, limitations have been identified. The most important identified limitation, currently, is the sensitivity of HRF to variations in image acquisition and reconstruction parameters. In this study, we investigated the use of Reconstruction Kernel Normalization (RKN) and ComBat harmonization to improve the reproducibility of HRFs across scans acquired with different reconstruction kernels. A set of phantom scans (n = 28) acquired on five different scanner models was analyzed. HRFs were extracted from the original scans, and scans were harmonized using the RKN method. ComBat harmonization was applied on both sets of HRFs. The reproducibility of HRFs was assessed using the concordance correlation coefficient. The difference in the number of reproducible HRFs in each scenario was assessed using McNemar's test. The majority of HRFs were found to be sensitive to variations in the reconstruction kernels, and only six HRFs were found to be robust with respect to variations in reconstruction kernels. The use of RKN resulted in a significant increment in the number of reproducible HRFs in 19 out of the 67 investigated scenarios (28.4%), while the ComBat technique resulted in a significant increment in 36 (53.7%) scenarios. The combination of methods resulted in a significant increment in 53 (79.1%) scenarios compared to the HRFs extracted from original images. Since the benefit of applying the harmonization methods depended on the data being harmonized, reproducibility analysis is recommended before performing radiomics analysis. For future radiomics studies incorporating images acquired with similar image acquisition and reconstruction parameters, except for the reconstruction kernels, we recommend the systematic use of the pre- and post-processing approaches (respectively, RKN and ComBat).

Evaluating coronary artery calcification with low-dose chest CT reconstructed by different kernels.

PURPOSE: To understand the reliability of low-dose chest computed tomography (LDCT) in coronary artery calcification (CAC) assessment and evaluate the performance of different reconstruction kernels against the standard cardiac computed tomography (CaCT) as reference. MATERIALS AND METHODS: Patients from the NELCIN-B3 screening program who underwent CaCT and LDCT scans were analyzed retrospectively. LDCT were reconstructed with smooth, standard, and sharp kernels (Group B1, B2 and B3) to compare against standard CaCT (Group A). The image quality was evaluated by noise value, signal-to-noise ratio (SNR), and contrast to noise ratio (CNR); moreover, radiation dose was recorded for both scans. Coronary artery calcification scores (CACS) were measured with volume, mass and Agatston standards. Agatston score was divided into four cardiovascular risk categories (0, 1-99, 100-399, and >400). The agreement in CACS and risk classification between LDCT and CaCT was analyzed by intra-group correlation coefficient (ICC) and Kappa test. RESULTS: The sensitivity of diagnosing CAC with LDCT was 98.5% (330/335) regardless of reconstruction kernels. Group B1 demonstrated the highest agreement in raw CACS (ICC volume 0.932; mass 0.904; Agatston 0.906; all p < 0.001) and risk classification (kappa 0.757, 95% CI 0.70-0.82). Smooth-kernel reconstruction achieved lower image noise, better SNR and CNR than other kernels. The effective radiation dose in of LDCT was 41.2% lower than that of the calcium scan (p < 0.001). CONCLUSION: Reconstructing LDCT with a smooth kernel in LDCT could provide a reliable imaging method to detect and quantitatively evaluate CAC, potentially expanding the application of LDCT lung screening to incidental findings of cardiovascular disease.

The effect of CT reconstruction kernels and display window settings on the detection and measurement of pulmonary solid nodules / 西安交通大学学报(医学版)

【Objective】 To assess the effect of reconstruction kernels and window settings on the detection and measurement of pulmonary solid nodules and their measurement variability and repeatability. 【Methods】 We retrospectively recruited 49 patients with pulmonary solid nodules who had undergone low-dose CT scanning. Images were reconstructed using five reconstruction kernels： lung， bone， chest， detail and standard kernels. Two radiologists independently assessed the detection rate， diameter and CT number measurement of nodules under the five kernels and two window settings （lung-window and mediastinal-window）. Bland-Altman plots and relative average deviation （RAD） were used to evaluate the repeatability and variability of nodule diameter and CT number measurement. 【Results】 Seventy-seven nodules were detected on lung-window regardless of reconstruction kernels， while the detection rates （75.3%-98.7%） were significantly different （P<0.001） on the mediastinal-window， with the lung kernel significantly improving the detection of nodules with the diameter below 6 mm. In both display windows， the diameter and CT number measurements among reconstruction kernels were similar except for the lung kernel. The lung-window had better variability in the diameter measurement while mediastinal-window was better in CT number measurement among various reconstruction kernels. Although the variability in the diameter of the nodule on the lung-window and mediastinal-window was similar， there was a significant difference in the variability in the diameter measurement among different reconstruction kernels on the mediastinal-window （P=0.004）. No significant difference in the variability in the CT number measurement was found among the different reconstruction kernels （lung-window P=0.163； mediastinal-window P=0.201）， and the variability in the CT number measurements on the mediastinal-window was smaller than that of the lung-window. Both window displays had acceptable repeatability in diameter and CT number measurement； however， the mediastinal-window was better in CT number measurement. 【Conclusion】 The lung kernel can improve the detection of pulmonary solid nodules below 6 mm， but is limited in the CT number measurement. The lung-window display provides better variability in measuring nodule diameter， while mediastinal-window display is better at measuring CT numbers.

Influence of Radiation Dose and Reconstruction Kernel on Fat Fraction Analysis in Dual-energy CT: A Phantom Study.

BACKGROUND/AIM: The quantitative evaluation of fat tissue, mainly for the determination of liver steatosis, is possible by using dual-energy computed tomography. Different photon energy acquisitions allow for estimation of attenuation coefficients. The effect of variation in radiation doses and reconstruction kernels on fat fraction estimation was investigated. MATERIALS AND METHODS: A six-probe-phantom with fat concentrations of 0%, 20%, 40%, 60%, 80%, and 100% were scanned in Sn140/100 kV with radiation doses ranging between 20 and 200 mAs before and after calibration. Images were reconstructed using iterative kernels (I26,Q30,I70). RESULTS: Fat fractions measured in dual-energy computed tomography (DECT) were consistent with the 20%-stepwise varying actual concentrations. Variation in radiation dose resulted in 3.1% variation of fat fraction. Softer reconstruction kernel (I26) underestimated the fat fraction (-9.1%), while quantitative (Q30) and sharper kernel (I70) overestimated fat fraction (10,8% and 13,1, respectively). CONCLUSION: The fat fraction in DECT approaches the actual fat concentration when calibrated to the reconstruction kerneö. Variation of radiation dose caused an acceptable 3% variation.

Kernel Conversion for Robust Quantitative Measurements of Archived Chest Computed Tomography Using Deep Learning-Based Image-to-Image Translation.

Chest computed tomography (CT) is used to screen for lung cancer and evaluate pulmonary and extra-pulmonary abnormalities such as emphysema and coronary artery calcification, particularly in smokers. In real-world practice, lung abnormalities are visually assessed using high-contrast thin-slice images which are generated from raw scan data using sharp reconstruction kernels with the sacrifice of increased image noise. In contrast, accurate CT quantification requires low-contrast thin-slice images with low noise, which are generated using soft reconstruction kernels. However, only sharp-kernel thin-slice images are archived in many medical facilities due to limited data storage space. This study aimed to establish deep neural network (DNN) models to convert sharp-kernel images to soft-kernel-like images with a final goal to reuse historical chest CT images for robust quantitative measurements, particularly in completed previous longitudinal studies. By using pairs of sharp-kernel (input) and soft-kernel (ground-truth) images from 30 patients with chronic obstructive pulmonary disease (COPD), DNN models were trained. Then, the accuracy of kernel conversion based on the established DNN models was evaluated using CT from independent 30 smokers with and without COPD. Consequently, differences in CT values between new images converted from sharp-kernel images using the established DNN models and ground-truth soft-kernel images were comparable with the inter-scans variability derived from repeated phantom scans (6 times), showing that the conversion error was the same level as the measurement error of the CT device. Moreover, the Dice coefficients to quantify the similarity between low attenuation voxels on given images and the ground-truth soft-kernel images were significantly higher on the DNN-converted images than the Gaussian-filtered, median-filtered, and sharp-kernel images (p < 0.001). There were good agreements in quantitative measurements of emphysema, intramuscular adipose tissue, and coronary artery calcification between the converted and the ground-truth soft-kernel images. These findings demonstrate the validity of the new DNN model for kernel conversion and the clinical applicability of soft-kernel-like images converted from archived sharp-kernel images in previous clinical studies. The presented method to evaluate the validity of the established DNN model using repeated scans of phantom could be applied to various deep learning-based image conversions for robust quantitative evaluation.

Accounting for reconstruction kernel-induced variability in CT radiomic features using noise power spectra.

Large variability in computed tomography (CT) radiomics feature values due to CT imaging parameters can have subsequent implications on the prognostic or predictive significance of these features. Here, we investigated the impact of pitch, dose, and reconstruction kernel on CT radiomic features. Moreover, we introduced correction factors to reduce feature variability introduced by reconstruction kernels. The credence cartridge radiomics and American College of Radiology (ACR) phantoms were scanned on five different scanners. ACR phantom was used for 3-D noise power spectrum (NPS) measurements to quantify correlated noise. The coefficient of variation (COV) was used as the variability assessment metric. The variability in texture features due to different kernels was reduced by applying the NPS peak frequency and region of interest (ROI) maximum intensity as correction factors. Most texture features were dose independent but were strongly kernel dependent, which is demonstrated by a significant shift in NPS peak frequency among kernels. Percentage improvement in robustness was calculated for each feature from original and corrected %COV values. Percentage improvements in robustness of 19 features were in the range of 30% to 78% after corrections. We show that NPS peak frequency and ROI maximum intensity can be used as correction factors to reduce variability in CT texture feature values due to reconstruction kernels.

Modified ideal observer model (MIOM) for high-contrast and high-spatial resolution CT imaging tasks.

PURPOSE: Although a variety of mathematical observer models have been developed to predict human observer performance for low contrast lesion detection tasks, their predictive power for high contrast and high spatial resolution discrimination imaging tasks, including those in CT bone imaging, could be limited. The purpose of this work was to develop a modified observer model that has improved correlation with human observer performance for these tasks. METHODS: The proposed observer model, referred to as the modified ideal observer model (MIOM), uses a weight function to penalize components in the task function that have less contribution to the actual human observer performance for high contrast and high spatial resolution discrimination tasks. To validate MIOM, both human observer and observer model studies were performed, each using exactly the same CT imaging task [discrimination of a connected component in a high contrast (1000 HU) high spatial resolution bone fracture model (0.3 mm)] and experimental CT image data. For the human observer studies, three physicist observers rated the connectivity of the fracture model using a five-point Likert scale; for the observer model studies, a total of five observer models, including both conventional models and the proposed MIOM, were used to calculate the discrimination capability of the CT images in resolving the connected component. Images used in the studies encompassed nine different reconstruction kernels. Correlation between human and observer model performance for these kernels were quantified using the Spearman rank correlation coefficient (ρ). After the validation study, an example application of MIOM was presented, in which the observer model was used to select the optimal reconstruction kernel for a High-Resolution (Hi-Res, GE Healthcare) CT scan technique. RESULTS: The performance of the proposed MIOM correlated well with that of the human observers with a Spearman rank correlation coefficient ρ of 0.88 (P = 0.003). In comparison, the value of ρ was 0.05 (P = 0.904) for the ideal observer, 0.05 (P = 0.904) for the non-prewhitening observer, -0.18 (P = 0.634) for the non-prewhitening observer with eye filter and internal noise, and 0.30 (P = 0.427) for the prewhitening observer with eye filter and internal noise. Using the validated MIOM, the optimal reconstruction kernel for the Hi-Res mode to perform high spatial resolution and high contrast discrimination imaging tasks was determined to be the HD Ultra kernel at the center of the scan field of view (SFOV), or the Lung kernel at the peripheral region of the SFOV. This result was consistent with visual observations of nasal CT images of an in vivo canine subject. CONCLUSION: Compared with other observer models, the proposed modified ideal observer model provides significantly improved correlation with human observers for high contrast and high spatial resolution CT imaging tasks.

Combined multi-kernel head computed tomography images optimized for depicting both brain parenchyma and bone.

BACKGROUND: The hybrid convolution kernel technique for computed tomography (CT) is known to enable the depiction of an image set using different window settings. OBJECTIVE: Our purpose was to decrease the number of artifacts in the hybrid convolution kernel technique for head CT and to determine whether our improved combined multi-kernel head CT images enabled diagnosis as a substitute for both brain (low-pass kernel-reconstructed) and bone (high-pass kernel-reconstructed) images. METHODS: Forty-four patients with nondisplaced skull fractures were included. Our improved multi-kernel images were generated so that pixels of >100 Hounsfield unit in both brain and bone images were composed of CT values of bone images and other pixels were composed of CT values of brain images. Three radiologists compared the improved multi-kernel images with bone images. RESULTS: The improved multi-kernel images and brain images were identically displayed on the brain window settings. All three radiologists agreed that the improved multi-kernel images on the bone window settings were sufficient for diagnosing skull fractures in all patients. CONCLUSIONS: This improved multi-kernel technique has a simple algorithm and is practical for clinical use. Thus, simplified head CT examinations and fewer images that need to be stored can be expected.

Combined multi-kernel chest computed tomography images optimized for depicting both lung and soft tissue.

PURPOSE: To evaluate the quality of our improved multi-kernel chest computed tomography (CT) images. METHODS: A random sample of 50 normal patients was retrospectively selected from those who underwent chest CT scans between January 2010 and July 2010. Normal lung structures were divided into six categories, and two radiologists independently compared with lung images. RESULTS: The improved multi-kernel images were displayed identically to soft tissue images on soft tissue window settings and were evaluated as equal to lung images on lung window settings. CONCLUSIONS: This improved multi-kernel technique required fewer stored images and simplified examinations of chest CT.