A Physics-Informed Deep Neural Network for Harmonization of CT Images

OBJECTIVE

Computed Tomography (CT) quantification is affected by the variability in image acquisition and rendition. This paper aimed to reduce this variability by harmonizing the images utilizing physics-based deep neural networks (DNNs).

METHODS

An adversarial generative network was trained on virtual CT images acquired under various imaging conditions using a virtual imaging platform with 40 computational patient models. These models featured anthropomorphic lungs with different levels of pulmonary diseases, including nodules and emphysema. Imaging was conducted using a validated CT simulator at two dose levels and varying reconstruction kernels. The trained model was tested on an independent virtual test dataset and two clinical datasets.

RESULTS

On the virtual test set, the harmonizer improved the structural similarity index from 79.3 16.4% to 95.8 1.7%, normalized mean squared error from 16.7 9.7% to 9.2 1.7%, and peak signal-to-noise ratio from 27.7 3.7 dB to 32.2 1.6 dB. Moreover, the harmonized images yielded more precise quantification of emphysema-based imaging biomarkers for lung attenuation, LAA -950 from 5.6 8.7% to 0.23 0.16%, Perc 15 from 43.4 45.4 HU to 20.0 7.5 HU, and Lung Mass from 0.3 0.3 g to 0.1 0.2 g. In clinical data, the harmonizer reduced biomarker variability by an average of 70%. For lung nodules, harmonized images improved the detectability index by 6.5-fold and DNN-based precision by 6%.

CONCLUSION

The proposed harmonizer significantly enhances image quality and quantification accuracy in CT imaging.

SIGNIFICANCE

The study demonstrated the potential utility of image harmonization for consistent CT image quality and reliable quantification, which is crucial for clinical applications and patient management.

Learning with privileged knowledge of multiple kernels via joint prediction for CT Kernel conversion

BACKGROUND

Most existing models for CT kernel conversion take images reconstructed with a single predetermined source kernel as input and convert them to images that are reconstructed with a target kernel. However, these models can achieve even better performance if they leverage complementary information obtained from images reconstructed with multiple different kernels. In many clinical practice scenarios, only images with one kernel can be acquired.

PURPOSE

We propose a privileged knowledge learning framework that learns privileged knowledge of other source kernels available only in the training data (called privileged information) to guide the conversion from a specific single source kernel to the target kernel, via a joint prediction (JP) task.

METHODS

We construct an ensemble of kernel-specific (KS) tasks where a KS network (KSNet) takes images reconstructed with a specific source kernel as input and converts them to images reconstructed with the target kernel. Then, a JP task is designed to provide extra regularization, which helps each KSNet learn more informative feature representations for kernel conversion, such as detail and structure representations. Meanwhile, we use a cross-shaped window-based attention mechanism in the JP task to highlight the most relevant features to strengthen privileged knowledge learning, thereby alleviating the problems of redundant noise unrelated to images reconstructed with target kernel and inconsistent features that arise from images reconstructed with different kernels. All KSNets can be trained collaboratively by using a JP task to improve the performance of each individual KSNet.

RESULTS

We extensively evaluate our method on a clinical dataset with scanners from three manufacturers, that is, Siemens, GE and Philips. The experimental results demonstrate that our privileged knowledge learning framework is effective in improving CT kernel conversion.

CONCLUSIONS

Through both quantitative and qualitative research, our privileged knowledge learning framework improves the kernel conversion results, thereby contributing to the improvement of diagnostic accuracy and the advancement of comparative research in quantitative measurements.

Deep learning in computed tomography super resolution using multi-modality data training

BACKGROUND

One of the limitations in leveraging the potential of artificial intelligence in X-ray imaging is the limited availability of annotated training data. As X-ray and CT shares similar imaging physics, one could achieve cross-domain data sharing, so to generate labeled synthetic X-ray images from annotated CT volumes as digitally reconstructed radiographs (DRRs). To account for the lower resolution of CT and the CT-generated DRRs as compared to the real X-ray images, we propose the use of super-resolution (SR) techniques to enhance the CT resolution before DRR generation.

PURPOSE

As spatial resolution can be defined by the modulation transfer function kernel in CT physics, we propose to train a SR network using paired low-resolution (LR) and high-resolution (HR) images by varying the kernel's shape and cutoff frequency. This is different to previous deep learning-based SR techniques on RGB and medical images which focused on refining the sampling grid. Instead of generating LR images by bicubic interpolation, we aim to create realistic multi-detector CT (MDCT) like LR images from HR cone-beam CT (CBCT) scans.

METHODS

We propose and evaluate the use of a SR U-Net for the mapping between LR and HR CBCT image slices. We reconstructed paired LR and HR training volumes from the same CT scans with small in-plane sampling grid size of0.20 × 0.20 mm 2 $0.20 \times 0.20 \, {\rm mm}^2$ . We used the residual U-Net architecture to train two models. SRUNR e s K $^K_{Res}$ : trained with kernel-based LR images, and SRUNR e s I $^I_{Res}$ : trained with bicubic downsampled data as baseline. Both models are trained on one CBCT dataset (n = 13 391). The performance of both models was then evaluated on unseen kernel-based and interpolation-based LR CBCT images (n = 10 950), and also on MDCT images (n = 1392).

RESULTS

Five-fold cross validation and ablation study were performed to find the optimal hyperparameters. Both SRUNR e s K $^K_{Res}$ and SRUNR e s I $^I_{Res}$ models show significant improvements (p-value < $<$ 0.05) in mean absolute error (MAE), peak signal-to-noise ratio (PSNR) and structural similarity index measures (SSIMs) on unseen CBCT images. Also, the improvement percentages in MAE, PSNR, and SSIM by SRUNR e s K $^K_{Res}$ is larger than SRUNR e s I $^I_{Res}$ . For SRUNR e s K $^K_{Res}$ , MAE is reduced by 14%, and PSNR and SSIMs increased by 6 and 8%, respectively. To conclude, SRUNR e s K $^K_{Res}$ outperforms SRUNR e s I $^I_{Res}$ , which the former generates sharper images when tested with kernel-based LR CBCT images as well as cross-modality LR MDCT data.

CONCLUSIONS

Our proposed method showed better performance than the baseline interpolation approach on unseen LR CBCT. We showed that the frequency behavior of the used data is important for learning the SR features. Additionally, we showed cross-modality resolution improvements to LR MDCT images. Our approach is, therefore, a first and essential step in enabling realistic high spatial resolution CT-generated DRRs for deep learning training.

Adaptation to CT Reconstruction Kernels by Enforcing Cross-Domain Feature Maps Consistency

Deep learning methods provide significant assistance in analyzing coronavirus disease (COVID-19) in chest computed tomography (CT) images, including identification, severity assessment, and segmentation. Although the earlier developed methods address the lack of data and specific annotations, the current goal is to build a robust algorithm for clinical use, having a larger pool of available data. With the larger datasets, the domain shift problem arises, affecting the performance of methods on the unseen data. One of the critical sources of domain shift in CT images is the difference in reconstruction kernels used to generate images from the raw data (sinograms). In this paper, we show a decrease in the COVID-19 segmentation quality of the model trained on the smooth and tested on the sharp reconstruction kernels. Furthermore, we compare several domain adaptation approaches to tackle the problem, such as task-specific augmentation and unsupervised adversarial learning. Finally, we propose the unsupervised adaptation method, called F-Consistency, that outperforms the previous approaches. Our method exploits a set of unlabeled CT image pairs which differ only in reconstruction kernels within every pair. It enforces the similarity of the network's hidden representations (feature maps) by minimizing the mean squared error (MSE) between paired feature maps. We show our method achieving a 0.64 Dice Score on the test dataset with unseen sharp kernels, compared to the 0.56 Dice Score of the baseline model. Moreover, F-Consistency scores 0.80 Dice Score between predictions on the paired images, which almost doubles the baseline score of 0.46 and surpasses the other methods. We also show F-Consistency to better generalize on the unseen kernels and without the presence of the COVID-19 lesions than the other methods trained on unlabeled data.

Emphysema quantification using low-dose computed tomography with deep learning-based kernel conversion comparison

OBJECTIVE

This study determined the effect of dose reduction and kernel selection on quantifying emphysema using low-dose computed tomography (LDCT) and evaluated the efficiency of a deep learning-based kernel conversion technique in normalizing kernels for emphysema quantification.

METHODS

A sample of 131 participants underwent LDCT and standard-dose computed tomography (SDCT) at 1- to 2-year intervals. LDCT images were reconstructed with B31f and B50f kernels, and SDCT images were reconstructed with B30f kernels. A deep learning model was used to convert the LDCT image from a B50f kernel to a B31f kernel. Emphysema indices (EIs), lung attenuation at 15th percentile (perc15), and mean lung density (MLD) were calculated. Comparisons among the different kernel types for both LDCT and SDCT were performed using Friedman's test and Bland-Altman plots.

RESULTS

All values of LDCT B50f were significantly different compared with the values of LDCT B31f and SDCT B30f (p < 0.05). Although there was a statistical difference, the variation of the values of LDCT B50f significantly decreased after kernel normalization. The 95% limits of agreement between the SDCT and LDCT kernels (B31f and converted B50f) ranged from - 2.9 to 4.3% and from - 3.2 to 4.4%, respectively. However, there were no significant differences in EIs and perc15 between SDCT and LDCT converted B50f in the non-chronic obstructive pulmonary disease (COPD) participants (p > 0.05).

CONCLUSION

The deep learning-based CT kernel conversion of sharp kernel in LDCT significantly reduced variation in emphysema quantification, and could be used for emphysema quantification.

KEY POINTS

• Low-dose computed tomography with smooth kernel showed adequate performance in quantifying emphysema compared with standard-dose CT. • Emphysema quantification is affected by kernel selection and the application of a sharp kernel resulted in a significant overestimation of emphysema. • Deep learning-based kernel normalization of sharp kernel significantly reduced variation in emphysema quantification.

CT Image Conversion among Different Reconstruction Kernels without a Sinogram by Using a Convolutional Neural Network

OBJECTIVE

The aim of our study was to develop and validate a convolutional neural network (CNN) architecture to convert CT images reconstructed with one kernel to images with different reconstruction kernels without using a sinogram.

MATERIALS AND METHODS

This retrospective study was approved by the Institutional Review Board. Ten chest CT scans were performed and reconstructed with the B10f, B30f, B50f, and B70f kernels. The dataset was divided into six, two, and two examinations for training, validation, and testing, respectively. We constructed a CNN architecture consisting of six convolutional layers, each with a 3 × 3 kernel with 64 filter banks. Quantitative performance was evaluated using root mean square error (RMSE) values. To validate clinical use, image conversion was conducted on 30 additional chest CT scans reconstructed with the B30f and B50f kernels. The influence of image conversion on emphysema quantification was assessed with Bland-Altman plots.

RESULTS

Our scheme rapidly generated conversion results at the rate of 0.065 s/slice. Substantial reduction in RMSE was observed in the converted images in comparison with the original images with different kernels (mean reduction, 65.7%; range, 29.5-82.2%). The mean emphysema indices for B30f, B50f, converted B30f, and converted B50f were 5.4 ± 7.2%, 15.3 ± 7.2%, 5.9 ± 7.3%, and 16.8 ± 7.5%, respectively. The 95% limits of agreement between B30f and other kernels (B50f and converted B30f) ranged from -14.1% to -2.6% (mean, -8.3%) and -2.3% to 0.7% (mean, -0.8%), respectively.

CONCLUSION

CNN-based CT kernel conversion shows adequate performance with high accuracy and speed, indicating its potential clinical use.

Impact of a single distance phase retrieval algorithm on spatial resolution in X-ray inline phase sensitive imaging

A single-projection based phase retrieval method based on the phase attenuation duality principle (PAD) was used to compare the spatial resolution of the acquired phase sensitive and PAD processed phase retrieved images. An inline phase sensitive prototype was used to acquire the phase sensitive images. The prototype incorporates a micro-focus x-ray source and a flat panel detector with a 50 μm pixel pitch. A phantom composed of a 2 cm thick 50-50 adipose-glandular mimicking slab sandwiched with a 0.82 cm thick slanted PMMA sharp edge was used. Phase sensitive image of the phantom was acquired at 120 kV, 3.35 mAs with a 16 μm tube focal spot size under a geometric magnification (M) of 2.5. The PAD based method was applied to the acquired phase sensitive image for the retrieval of phase values. With necessary data processing, modulation transfer function (MTF) curves were determined for the estimation and comparison of the spatial resolution. The PAD processed phase retrieved values of the phantom were in good agreement with the theoretically calculated values. Phase sensitive images showed higher spatial resolution at all spatial frequencies compared to the phase retrieved images. It was noted that the high-frequency signal components in the retrieved image were suppressed that resulted in lower MTF values. When compared to the phase sensitive image, the cutoff resolution (10% MTF) for phase retrieved image dropped 32% from 15.6 lp/mm (32μm) to 10.6 lp/mm (47μm). The resolution offered by this phase sensitive prototype is radiographically enough to detect breast cancer.

Deep learning-enabled accurate normalization of reconstruction kernel effects on emphysema quantification in low-dose CT

Lung densitometry is being frequently adopted in CT-based emphysema quantification, yet known to be affected by the choice of reconstruction kernel. This study presents a two-step deep learning architecture that enables accurate normalization of reconstruction kernel effects on emphysema quantification in low-dose CT. Deep learning is used to convert a CT image of a sharp kernel to that of a standard kernel with restoration of truncation artifacts and smoothing-free pixel size normalization. We selected 353 scans reconstructed by both standard and sharp kernels from four different CT scanners from the United States National Lung Screening Trial program database. A truncation artifact correction model was constructed with a combination of histogram extrapolation and a deep learning model trained with truncated and non-truncated image sets. Then, we performed frequency domain zero-padding to normalize reconstruction field of view effects while preventing image smoothing effects. The kernel normalization model has a U-Net based architecture trained for each CT scanner dataset. Three lung density measurements including relative lung area under 950 HU (RA950), lower 15th percentile threshold (perc15), and mean lung density were obtained in the datasets from standard, sharp, and normalized kernels. The effect of kernel normalization was evaluated with pair-wise differences in lung density metrics. The mean of pair-wise differences in RA950 between standard and sharp kernel reconstructions was reduced from 10.75% to  -0.07% using kernel normalization. The difference for perc15 decreased from  -31.03 HU to  -0.30 HU after kernel normalization. Our study demonstrated the feasibility of applying deep learning techniques for normalizing CT kernel effects, thereby reducing the kernel-induced variability in lung density measurements. The deep learning model could increase the accuracy of emphysema quantification, thereby allowing reliable surveillance of emphysema in lung cancer screening even when follow-up CT scans are acquired with different reconstruction kernels.

Deep Learning-based Image Conversion of CT Reconstruction Kernels Improves Radiomics Reproducibility for Pulmonary Nodules or Masses

Background Intratumor heterogeneity in lung cancer may influence outcomes. CT radiomics seeks to assess tumor features to provide detailed imaging features. However, CT radiomic features vary according to the reconstruction kernel used for image generation. Purpose To investigate the effect of different reconstruction kernels on radiomic features and assess whether image conversion using a convolutional neural network (CNN) could improve reproducibility of radiomic features between different kernels. Materials and Methods In this retrospective analysis, patients underwent non-contrast material-enhanced and contrast material-enhanced axial chest CT with soft kernel (B30f) and sharp kernel (B50f) reconstruction using a single CT scanner from April to June 2017. To convert different kernels without sinogram, the CNN model was developed using residual learning and an end-to-end way. Kernel-converted images were generated, from B30f to B50f and from B50f to B30f. Pulmonary nodules or masses were semiautomatically segmented and 702 radiomic features (tumor intensity, texture, and wavelet features) were extracted. Measurement variability in radiomic features was evaluated using the concordance correlation coefficient (CCC). Results A total of 104 patients were studied, including 54 women and 50 men, with pulmonary nodules or masses (mean age, 63.2 years ± 10.5). The CCC between two readers using the same kernel was 0.92, and 592 of 702 (84.3%) of the radiomic features were reproducible (CCC ≥ 0.85); using different kernels, the CCC was 0.38 and only 107 of 702 (15.2%) of the radiomic features were reliable. Texture features and wavelet features were predominantly affected by reconstruction kernel (CCC, from 0.88 to 0.61 for texture features and from 0.92 to 0.35 for wavelet features). After applying image conversion, CCC improved to 0.84 and 403 of 702 (57.4%) radiomic features were reproducible (CCC, 0.85 for texture features and 0.84 for wavelet features). Conclusion Chest CT image conversion using a convolutional neural network effectively reduced the effect of two different reconstruction kernels and may improve the reproducibility of radiomic features in pulmonary nodules or masses. © RSNA, 2019 Online supplemental material is available for this article. See also the editorial by Park in this issue.

Utility of second-generation single-energy metal artifact reduction in helical lung computed tomography for patients with pulmonary arteriovenous malformation after coil embolization

PURPOSE

The quality of images acquired using single-energy metal artifact reduction (SEMAR) on helical lung computed tomography (CT) in patients with pulmonary arteriovenous malformation (PAVM) after coil embolization was retrospectively evaluated.

MATERIALS AND METHODS

CT images were reconstructed with and without SEMAR. Twenty-seven lesions [20 patients (2 males, 18 females), mean age 61.2 ± 11.0 years; number of embolization coils, 9.8 ± 5.0] on contrast-enhanced CT and 18 lesions of non-enhanced lung CT concurrently performed were evaluated. Regions of interest were positioned around the coils and mean standard deviation value was compared as noise index. Two radiologists visually evaluated metallic coil artifacts using a four-point scale: 4 = minimal; 3 = mild; 2 = strong; 1 = extensive.

RESULTS

Noise index was significantly improved with SEMAR versus without SEMAR (median [interquartile range]; 194.4 [161.6-211.9] Hounsfield units [HU] vs. 243.9 [220.4-286.0] HU; p < 0.001). Visual score was significantly improved with SEMAR versus without SEMAR (Reader 1, 3 [3] vs.1 [1]; Reader 2, 3 [3] vs.1 [1]; p < 0.001). Significant differences were similarly demonstrated on lung CT (p < 0.001).

CONCLUSION

SEMAR provided clear chest CT images in patients who underwent PAVM coil embolization.

Technical Note: Image filtering to make computer-aided detection robust to image reconstruction kernel choice in lung cancer CT screening

PURPOSE

In lung cancer computed tomography (CT) screening, the performance of a computer-aided detection (CAD) system depends on the selection of the image reconstruction kernel. To reduce this dependence on reconstruction kernels, the authors propose a novel application of an image filtering method previously proposed by their group.

METHODS

The proposed filtering process uses the ratio of modulation transfer functions (MTFs) of two reconstruction kernels as a filtering function in the spatial-frequency domain. This method is referred to as MTFratio filtering. Test image data were obtained from CT screening scans of 67 subjects who each had one nodule. Images were reconstructed using two kernels: fSTD (for standard lung imaging) and fSHARP (for sharp edge-enhancement lung imaging). The MTFratio filtering was implemented using the MTFs measured for those kernels and was applied to the reconstructed fSHARP images to obtain images that were similar to the fSTD images. A mean filter and a median filter were applied (separately) for comparison. All reconstructed and filtered images were processed using their prototype CAD system.

RESULTS

The MTFratio filtered images showed excellent agreement with the fSTD images. The standard deviation for the difference between these images was very small, ∼6.0 Hounsfield units (HU). However, the mean and median filtered images showed larger differences of ∼48.1 and ∼57.9 HU from the fSTD images, respectively. The free-response receiver operating characteristic (FROC) curve for the fSHARP images indicated poorer performance compared with the FROC curve for the fSTD images. The FROC curve for the MTFratio filtered images was equivalent to the curve for the fSTD images. However, this similarity was not achieved by using the mean filter or median filter.

CONCLUSIONS

The accuracy of MTFratio image filtering was verified and the method was demonstrated to be effective for reducing the kernel dependence of CAD performance.

A method for evaluating the performance of computer-aided detection of pulmonary nodules in lung cancer CT screening: detection limit for nodule size and density

OBJECTIVE

We propose the application of virtual nodules to evaluate the performance of computer-aided detection (CAD) of lung nodules in cancer screening using low-dose CT.

METHODS

The virtual nodules were generated based on the spatial resolution measured for a CT system used in an institution providing cancer screening and were fused into clinical lung images obtained at that institution, allowing site specificity. First, we validated virtual nodules as an alternative to artificial nodules inserted into a phantom. In addition, we compared the results of CAD analysis between the real nodules (n = 6) and the corresponding virtual nodules. Subsequently, virtual nodules of various sizes and contrasts between nodule density and background density (ΔCT) were inserted into clinical images (n = 10) and submitted for CAD analysis.

RESULTS

In the validation study, 46 of 48 virtual nodules had the same CAD results as artificial nodules (kappa coefficient = 0.913). Real nodules and the corresponding virtual nodules showed the same CAD results. The detection limits of the tested CAD system were determined in terms of size and density of peripheral lung nodules; we demonstrated that a nodule with a 5-mm diameter was detected when the nodule had a ΔCT > 220 HU.

CONCLUSION

Virtual nodules are effective in evaluating CAD performance using site-specific scan/reconstruction conditions. Advances in knowledge: Virtual nodules can be an effective means of evaluating site-specific CAD performance. The methodology for guiding the detection limit for nodule size/density might be a useful evaluation strategy.

Practical approaches to approximating MTF and NPS in CT with an example application to task-based observer studies

PURPOSE

To investigate two methods of approximating the Modulation Transfer Function (MTF) and Noise Power Spectrum (NPS) in computed tomography (CT) for a range of scan parameters, from limited image acquisitions.

METHODS

The two methods consist of 1) using a linear systems approach to approximate the NPS for different filtered backprojection (FBP) kernels with a filter function derived from the kernel ratio of determined MTFs and 2) using an empirical fitted model to approximate the MTF and NPS. In both cases a scaling function accounts for variations in mAs and kV. The two methods of approximating the MTF/NPS are further investigated by comparing image quality figure of merits (FOM) d' and AUC calculated using approximations of the MTF/NPS and MTF/NPS that have been determined for different mAs/kV levels and reconstruction kernels.

RESULTS

The greatest RMSE for NPS approximated for a range of mAs/kVp/convolution kernels using both methods and compared to determined NPS was 0.05 of the peak value. The RMSE for FOM with the kernel ratio method were at most 0.1 for d' and 0.01 for the AUC. Using the empirical model method, the RMSE for FOM were at most 0.02 for d' and 0.001 for the AUC.

CONCLUSIONS

The two methods proposed in this paper can provide a convenient way of approximating the MTF and NPS for use in, among other things, mathematical observer studies. Both methods require a relatively small number of direct determinations of NPS from scan acquisitions to model the NPS/MTF for arbitrary mAs and kV.

Image quality of mixed convolution kernel in thoracic computed tomography

The mixed convolution kernel alters his properties geographically according to the depicted organ structure, especially for the lung. Therefore, we compared the image quality of the mixed convolution kernel to standard soft and hard kernel reconstructions for different organ structures in thoracic computed tomography (CT) images.Our Ethics Committee approved this prospective study. In total, 31 patients who underwent contrast-enhanced thoracic CT studies were included after informed consent. Axial reconstructions were performed with hard, soft, and mixed convolution kernel. Three independent and blinded observers rated the image quality according to the European Guidelines for Quality Criteria of Thoracic CT for 13 organ structures. The observers rated the depiction of the structures in all reconstructions on a 5-point Likert scale. Statistical analysis was performed with the Friedman Test and post hoc analysis with the Wilcoxon rank-sum test.Compared to the soft convolution kernel, the mixed convolution kernel was rated with a higher image quality for lung parenchyma, segmental bronchi, and the border between the pleura and the thoracic wall (P < 0.03). Compared to the hard convolution kernel, the mixed convolution kernel was rated with a higher image quality for aorta, anterior mediastinal structures, paratracheal soft tissue, hilar lymph nodes, esophagus, pleuromediastinal border, large and medium sized pulmonary vessels and abdomen (P < 0.004) but a lower image quality for trachea, segmental bronchi, lung parenchyma, and skeleton (P < 0.001).The mixed convolution kernel cannot fully substitute the standard CT reconstructions. Hard and soft convolution kernel reconstructions still seem to be mandatory for thoracic CT.

Normalizing computed tomography data reconstructed with different filter kernels: effect on emphysema quantification

Objectives

To propose and evaluate a method to reduce variability in emphysema quantification among different computed tomography (CT) reconstructions by normalizing CT data reconstructed with varying kernels.

Methods

We included 369 subjects from the COPDGene study. For each subject, spirometry and a chest CT reconstructed with two kernels were obtained using two different scanners. Normalization was performed by frequency band decomposition with hierarchical unsharp masking to standardize the energy in each band to a reference value. Emphysema scores (ES), the percentage of lung voxels below -950 HU, were computed before and after normalization. Bland-Altman analysis and correlation between ES and spirometry before and after normalization were compared. Two mixed cohorts, containing data from all scanners and kernels, were created to simulate heterogeneous acquisition parameters.

Results

The average difference in ES between kernels decreased for the scans obtained with both scanners after normalization (7.7 ± 2.7 to 0.3 ± 0.7; 7.2 ± 3.8 to -0.1 ± 0.5). Correlation coefficients between ES and FEV₁, and FEV₁/FVC increased significantly for the mixed cohorts.

Conclusions

Normalization of chest CT data reduces variation in emphysema quantification due to reconstruction filters and improves correlation between ES and spirometry.

Key Points

• Emphysema quantification is sensitive to the reconstruction kernel used.

• Normalization allows comparison of emphysema quantification from images reconstructed with varying kernels.

• Normalization allows comparison of emphysema quantification obtained with scanners from different manufacturers.

• Normalization improves correlation of emphysema quantification with spirometry.

• Normalization can be used to compare data from different studies and centers.

Accurate determination of CT point-spread-function with high precision

The measurement of modulation transfer functions (MTFs) in computed tomography (CT) is often performed by scanning a point source phantom such as a thin wire or a microbead. In these methods the region of interest (ROI) is generally placed on the scanned image to crop the point source response. The aim of the present study was to examine the effect of ROI size on MTF measurement, and to optimize the ROI size. Using a 4 multidetector‐row CT, MTFs were measured by the wire and bead methods for three types of reconstruction kernels designated as ‘smooth', ‘standard', and ‘edge‐enhancement’ kernels. The size of a square ROI was changed from 30 to 50 pixels (approximately 2.9 to 4.9 mm). The accuracies of the MTFs were evaluated using the verification method. The MTFs measured by the wire and bead methods were dependent on ROI size, particularly in MTF measurement for the ‘edge‐enhancement’ kernel. MTF accuracy evaluated by the verification method changed with ROI size, and we were able to determine the optimum ROI size for each method (wire/bead) and for each kernel. Using these optimal ROI sizes, the MTF obtained by the wire method was in strong agreement with the MTF obtained by the bead method in each kernel. Our data demonstrate that the difficulties in obtaining accurate MTFs for some kernels such as edge‐enhancement can be overcome by incorporating the verification method into the wire and bead methods, allowing optimization of the ROI size to accurately determine the MTF.

PACS numbers: 87.57.‐s, 87.57.cf, 87.57.Q‐

Accuracy of lung nodule density on HRCT: analysis by PSF-based image simulation

A computed tomography (CT) image simulation technique based on the point spread function (PSF) was applied to analyze the accuracy of CT‐based clinical evaluations of lung nodule density. The PSF of the CT system was measured and used to perform the lung nodule image simulation. Then, the simulated image was resampled at intervals equal to the pixel size and the slice interval found in clinical high‐resolution CT (HRCT) images. On those images, the nodule density was measured by placing a region of interest (ROI) commonly used for routine clinical practice, and comparing the measured value with the true value (a known density of object function used in the image simulation). It was quantitatively determined that the measured nodule density depended on the nodule diameter and the image reconstruction parameters (kernel and slice thickness). In addition, the measured density fluctuated, depending on the offset between the nodule center and the image voxel center. This fluctuation was reduced by decreasing the slice interval (i.e., with the use of overlapping reconstruction), leading to a stable density evaluation. Our proposed method of PSF‐based image simulation accompanied with resampling enables a quantitative analysis of the accuracy of CT‐based evaluations of lung nodule density. These results could potentially reveal clinical misreadings in diagnosis, and lead to more accurate and precise density evaluations. They would also be of value for determining the optimum scan and reconstruction parameters, such as image reconstruction kernels and slice thicknesses/intervals.

PACS numbers: 87.57.‐s, 87.57.cf, 87.57.Q‐

Application of CT-PSF-based computer-simulated lung nodules for evaluating the accuracy of computer-aided volumetry

With the wide dissemination of computed tomography (CT) screening for lung cancer, measuring the nodule volume accurately with computer-aided volumetry software is increasingly important. Many studies for determining the accuracy of volumetry software have been performed using a phantom with artificial nodules. These phantom studies are limited, however, in their ability to reproduce the nodules both accurately and in the variety of sizes and densities required. Therefore, we propose a new approach of using computer-simulated nodules based on the point spread function measured in a CT system. The validity of the proposed method was confirmed by the excellent agreement obtained between computer-simulated nodules and phantom nodules regarding the volume measurements. A practical clinical evaluation of the accuracy of volumetry software was achieved by adding simulated nodules onto clinical lung images, including noise and artifacts. The tested volumetry software was revealed to be accurate within an error of 20 % for nodules >5 mm and with the difference between nodule density and background (lung) (CT value) being 400-600 HU. Such a detailed analysis can provide clinically useful information on the use of volumetry software in CT screening for lung cancer. We concluded that the proposed method is effective for evaluating the performance of computer-aided volumetry software.