Noise reduction profile: A new method for evaluation of noise reduction techniques in CT

Advances in spatial resolution and radiation dose reduction using super-resolution deep learning-based reconstruction for abdominal computed tomography: A phantom study

RATIONALE AND OBJECTIVES

This study evaluated the performance of super-resolution deep learning-based reconstruction (SR-DLR) and compared with it that of hybrid iterative reconstruction (HIR) and normal-resolution DLR (NR-DLR) for enhancing image quality in computed tomography (CT) images across various field of view (FOV) sizes, radiation doses, and noise reduction strengths.

MATERIALS AND METHODS

A Catphan phantom equipped with an external body ring was used. CT images were reconstructed using filtered back-projection (FBP), HIR, NR-DLR, and SR-DLR across three noise reduction strengths: mild, standard, and strong. The noise power spectrum (NPS) was obtained from the FBP, HIR, NR-DLR, and SR-DLR images at various FOVs, radiation doses, and noise reduction strengths. The noise magnitude ratio (NMR) and central frequency ratio (CFR) were calculated from the HIR, NR-DLR, and SR-DLR images relative to the FBP images using NPS. The high-contrast value was obtained from the amplitude values of the peaks and valleys of profile curve and the task-based transfer function were also analyzed.

RESULTS

SR-DLR consistently demonstrated superior noise reduction capabilities, with NMR of 0.29-0.36 at reduced dose and 0.35-0.45 at standard dose, outperforming HIR and showing comparable efficiency to NR-DLR. The high-contrast values for SR-DLR were highest at mild and standard levels for both low and standard doses (0.610 and 0.726 at mild and 0.725 and 0.603 at standard levels). At the standard dose, the spatial resolution of SR-DLR was significantly improved, regardless of the noise reduction strength and FOV.

CONCLUSION

SR-DLR images achieved more substantial noise reduction than HIR and similar noise reduction as NR-DLR reconstructions while also improving spatial resolution.

Assessment and improvement of the quality of radiotherapy treatment planning CT images using a clinically validated phantom based method and a multicentre intercomparison

PURPOSE

To develop a phantom method of image quality assessment for radiotherapy planning CT protocols (head and neck (H&N) and prostate) and validate results against clinical image quality. Test with data from different scanners and suggest protocol adjustments.

METHODS

Macros measured patient water-equivalent diameter and noise from clinical CT images. Target transfer function (TTF), contrast, noise-power spectrum (NPS), detectability index and the edge visibility of a low contrast target were measured using Catphan 604 and bespoke phantoms. Ten centres scanned the phantoms with modified clinical protocols and collected data from patient images using the macros. Clinical experts, ranked the quality of images for contouring and correlated results against phantom metrics.

RESULTS

Clinical image review showed a large range of results from different scanners for H&N scans and fewer differences for prostate. The phantom metrics best correlated with high clinical image scores were, for H&N: high TTF50 (r = 0.73, p = 0.003), contrast (r = 0.58, p = 0.003) and target edge visibility (r = 0.70, p = 0.004); for prostate: high TTF50 (r = 0.83, p = 0.002), low noise (r = 0.37, p = 0.26) and target edge visibility (r = 0.59, p = 0.05). Hence, optimal contrast, resolution and noise are important for good contouring image quality. Reconstruction kernel, field of view and noise, or X-ray tube current and rotation time, are possible parameters for adjustment.

CONCLUSIONS

This phantom method (using Catphan 604) was a good surrogate for clinical quality assessment of CT images for radiotherapy contouring. Results identified the poorest performing scanners, allowing recommendations for image quality improvement and confirming scan protocol optimisation is necessary in some centres.

Adaptive-weighted high order TV algorithm for sparse-view CT reconstruction

BACKGROUND

With the development of low-dose computed tomography (CT), incomplete data reconstruction has been widely concerned. The total variation (TV) minimization algorithm can accurately reconstruct images from sparse or noisy data.

PURPOSE

However, the traditional TV algorithm ignores the direction of structures in images, leading to the loss of edge information and block artifacts when the object is not piecewise constant. Since the anisotropic information can facilitate preserving the edge and detail information in images, we aim to improve the TV algorithm in terms of reconstruction accuracy via this approach.

METHODS

In this paper, we propose an adaptive-weighted high order total variation (awHOTV) algorithm. We construct the second order TV-norm using the second order gradient, adapt the anisotropic edge property between neighboring image pixels, adjust the local image-intensity gradient to keep edge information, and design the corresponding Chambolle-Pock (CP) solving algorithm. Implementing the proposed algorithm, comprehensive studies are conducted in the ideal projection data experiment where the Structural similarity (SSIM), Root Mean Square Error (RMSE), Contrast to noise ratio (CNR), and modulation transform function (MTF) curves are utilized to evaluate the quality of reconstructed images in statism, structure, spatial resolution, and contrast, respectively. In the noisy data experiment, we further use the noise power spectrum (NPS) curve to evaluate the reconstructed images and compare it with other three algorithms.

RESULTS

We use the 2D slice in the XCAT phantom, 2D slice in TCIA Challenge data and FORBILD phantom as simulation phantoms and use real bird data for real verification. The results show that, compared with the traditional TV and FBP algorithms, the awHOTV has better performance in terms of RMSE, SSIM, and Pearson correlation coefficient (PCC) under the projected data with different sparsity. In addition, the awHOTV algorithm is robust against the noise of different intensities.

CONCLUSIONS

The proposed awHOTV method can reconstruct the images with high accuracy under sparse or noisy data. The awHOTV solves the strip artifacts caused by sparse data in the FBP method. Compared with the TV method, the awHOTV can effectively suppress block artifacts and has good detail protection ability.

Noise power spectrum properties of deep learning-based reconstruction and iterative reconstruction algorithms: Phantom and clinical study

PURPOSE

To compare the noise power spectrum (NPS) properties and perform a qualitative analysis of hybrid iterative reconstruction (IR), model-based IR (MBIR), and deep learning-based reconstruction (DLR) at a similar noise level in clinical study and compare these outcomes with those in phantom study.

METHODS

A Catphan phantom with an external body ring was used in the phantom study. In the clinical study, computed tomography (CT) examination data of 34 patients were reviewed. NPS was calculated from DLR, hybrid IR, and MBIR images. The noise magnitude ratio (NMR) and the central frequency ratio (CFR) were calculated from DLR, hybrid IR, and MBIR images relative to filtered back-projection images using NPS. Clinical images were independently reviewed by two radiologists.

RESULTS

In the phantom study, DLR with a mild level had a similar noise level as hybrid IR and MBIR with strong levels. In the clinical study, DLR with a mild level had a similar noise level as hybrid IR with standard and MBIR with strong levels. The NMR and CFR were 0.40 and 0.76 for DLR, 0.42 and 0.55 for hybrid IR, and 0.48 and 0.62 for MBIR. The visual inspection of the clinical DLR image was superior to that of the hybrid IR and MBIR images.

CONCLUSION

Deep learning-based reconstruction improves overall image quality with substantial noise reduction while maintaining image noise texture compared with the CT reconstruction techniques.

Deep Learning Denoising of Low-Dose Computed Tomography Chest Images: A Quantitative and Qualitative Image Analysis

PURPOSE

To assess deep learning denoised (DLD) computed tomography (CT) chest images at various low doses by both quantitative and qualitative perceptual image analysis.

METHODS

Simulated noise was inserted into sinogram data from 32 chest CTs acquired at 100 mAs, generating anatomically registered images at 40, 20, 10, and 5 mAs. A DLD model was developed, with 23 scans selected for training, 5 for validation, and 4 for test.Quantitative analysis of perceptual image quality was assessed with Structural SIMilarity Index (SSIM) and Fréchet Inception Distance (FID). Four thoracic radiologists graded overall diagnostic image quality, image artifact, visibility of small structures, and lesion conspicuity. Noise-simulated and denoised image series were evaluated in comparison with one another, and in comparison with standard 100 mAs acquisition at the 4 mAs levels. Statistical tests were conducted at the 2-sided 5% significance level, with multiple comparison correction.

RESULTS

At the same mAs levels, SSIM and FID between noise-simulated and reconstructed DLD images indicated that images were closer to a perfect match with increasing mAs (closer to 1 for SSIM, and 0 for FID).In comparing noise-simulated and DLD images to standard-dose 100-mAs images, DLD improved SSIM and FID. Deep learning denoising improved SSIM of 40-, 20-, 10-, and 5-mAs simulations in comparison with standard-dose 100-mAs images, with change in SSIM from 0.91 to 0.94, 0.87 to 0.93, 0.67 to 0.87, and 0.54 to 0.84, respectively. Deep learning denoising improved FID of 40-, 20-, 10-, and 5-mAs simulations in comparison with standard-dose 100-mAs images, with change in FID from 20 to 13, 46 to 21, 104 to 41, and 148 to 69, respectively.Qualitative image analysis showed no significant difference in lesion conspicuity between DLD images at any mAs in comparison with 100-mAs images. Deep learning denoising images at 10 and 5 mAs were rated lower for overall diagnostic image quality ( P < 0.001), and at 5 mAs lower for overall image artifact and visibility of small structures ( P = 0.002), in comparison with 100 mAs.

CONCLUSIONS

Deep learning denoising resulted in quantitative improvements in image quality. Qualitative assessment demonstrated DLD images at or less than 10 mAs to be rated inferior to standard-dose images.

A Review of Deep Learning CT Reconstruction: Concepts, Limitations, and Promise in Clinical Practice

Purpose of Review

Deep Learning reconstruction (DLR) is the current state-of-the-art method for CT image formation. Comparisons to existing filter back-projection, iterative, and model-based reconstructions are now available in the literature. This review summarizes the prior reconstruction methods, introduces DLR, and then reviews recent findings from DLR from a physics and clinical perspective.

Recent Findings

DLR has been shown to allow for noise magnitude reductions relative to filtered back-projection without suffering from “plastic” or “blotchy” noise texture that was found objectionable with most iterative and model-based solutions. Clinically, early reader studies have reported increases in subjective quality scores and studies have successfully implemented DLR-enabled dose reductions.

Summary

The future of CT image reconstruction is bright; deep learning methods have only started to tackle problems in this space via addressing noise reduction. Artifact mitigation and spectral applications likely be future candidates for DLR applications.

Fully automated image quality evaluation on patient CT: Multi-vendor and multi-reconstruction study

While the recent advancements of computed tomography (CT) technology have contributed in reducing radiation dose and image noise, an objective evaluation of image quality in patient scans has not yet been established. In this study, we present a patient-specific CT image quality evaluation method that includes fully automated measurements of noise level, structure sharpness, and alteration of structure. This study used the CT images of 120 patients from four different CT scanners reconstructed with three types of algorithm: filtered back projection (FBP), vendor-specific iterative reconstruction (IR), and a vendor-agnostic deep learning model (DLM, ClariCT.AI, ClariPi Inc.). The structure coherence feature (SCF) was used to divide an image into the homogeneous (RH) and structure edge (RS) regions, which in turn were used to localize the regions of interests (ROIs) for subsequent analysis of image quality indices. The noise level was calculated by averaging the standard deviations from five randomly selected ROIs on RH, and the mean SCFs on RS was used to estimate the structure sharpness. The structure alteration was defined by the standard deviation ratio between RS and RH on the subtraction image between FBP and IR or DLM, in which lower structure alterations indicate successful noise reduction without degradation of structure details. The estimated structure sharpness showed a high correlation of 0.793 with manually measured edge slopes. Compared to FBP, IR and DLM showed 34.38% and 51.30% noise reduction, 2.87% and 0.59% lower structure sharpness, and 2.20% and -12.03% structure alteration, respectively, on an average. DLM showed statistically superior performance to IR in all three image quality metrics. This study is expected to contribute to enhance the CT protocol optimization process by allowing a high throughput and quantitative image quality evaluation during the introduction or adjustment of lower-dose CT protocol into routine practice.

Influence of a Deep Learning Noise Reduction on the CT Values, Image Noise and Characterization of Kidney and Ureter Stones

Deep-learning (DL) noise reduction techniques in computed tomography (CT) are expected to reduce the image noise while maintaining the clinically relevant information in reduced dose acquisitions. This study aimed to assess the size, attenuation, and objective image quality of reno-ureteric stones denoised using DL-software in comparison to traditionally reconstructed low-dose abdominal CT-images and evaluated its clinical impact. In this institutional review-board-approved retrospective study, 45 patients with renal and/or ureteral stones were included. All patients had undergone abdominal CT between August 2019 and October 2019. CT-images were reconstructed using the following three methods: filtered back-projection, iterative reconstruction, and PixelShine (DL-software) with both sharp and soft kernels. Stone size, CT attenuation, and objective image quality (signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR)) were evaluated and compared using Bonferroni-corrected Friedman tests. Objective image quality was measured in six regions-of-interest. Stone size ranged between 4.4 × 3.1−4.4 × 3.2 mm (sharp kernel) and 5.1 × 3.8−5.6 × 4.2 mm (soft kernel). Mean attenuation ranged between 704−717 Hounsfield Units (HU) (soft kernel) and 915−1047 HU (sharp kernel). Differences in measured stone sizes were ≤1.3 mm. DL-processed images resulted in significantly higher CNR and SNR values (p < 0.001) by decreasing image noise significantly (p < 0.001). DL-software significantly improved objective image quality while maintaining both correct stone size and CT-attenuation values. Therefore, the clinical impact of stone assessment in denoised image data sets remains unchanged. Through the relevant noise suppression, the software additionally offers the potential to further reduce radiation exposure.

AI Denoising Significantly Enhances Image Quality and Diagnostic Confidence in Interventional Cone-Beam Computed Tomography

(1) To investigate whether interventional cone-beam computed tomography (cbCT) could benefit from AI denoising, particularly with respect to patient body mass index (BMI); (2) From 1 January 2016 to 1 January 2022, 100 patients with liver-directed interventions and peri-procedural cbCT were included. The unenhanced mask run and the contrast-enhanced fill run of the cbCT were reconstructed using weighted filtered back projection. Additionally, each dataset was post-processed using a novel denoising software solution. Place-consistent regions of interest measured signal-to-noise ratio (SNR) per dataset. Corrected mixed-effects analysis with BMI subgroup analyses compared objective image quality. Multiple linear regression measured the contribution of “Radiation Dose”, “Body-Mass-Index”, and “Mode” to SNR. Two radiologists independently rated diagnostic confidence. Inter-rater agreement was measured using Spearman correlation (r); (3) SNR was significantly higher in the denoised datasets than in the regular datasets (p < 0.001). Furthermore, BMI subgroup analysis showed significant SNR deteriorations in the regular datasets for higher patient BMI (p < 0.001), but stable results for denoising (p > 0.999). In regression, only denoising contributed positively towards SNR (0.6191; 95%CI 0.6096 to 0.6286; p < 0.001). The denoised datasets received overall significantly higher diagnostic confidence grades (p = 0.010), with good inter-rater agreement (r ≥ 0.795, p < 0.001). In a subgroup analysis, diagnostic confidence deteriorated significantly for higher patient BMI (p < 0.001) in the regular datasets but was stable in the denoised datasets (p ≥ 0.103).; (4) AI denoising can significantly enhance image quality in interventional cone-beam CT and effectively mitigate diagnostic confidence deterioration for rising patient BMI.