Application of Vendor-Neutral Iterative Reconstruction Technique to Pediatric Abdominal Computed Tomography

Universal 120-kV Dual-Source Ultra-High Pitch Protocol on the Photon-Counting CT System for Pediatric Abdomen of All Sizes: A Phantom Investigation Comparing With Energy-Integrating CT

OBJECTIVES

The purpose of this study is to determine if a universal 120-kV ultra-high pitch and virtual monoenergetic images (VMIs) protocol on the photon-counting computed tomography (PCCT) system can provide sufficient image quality for pediatric abdominal imaging, regardless of size, compared with protocols using a size-dependent kV and dual-source flash mode on the energy-integrating CT (EICT) system.

MATERIALS AND METHODS

One solid water insert and 3 iodine (2, 5, 10 mg I/mL) inserts were attached or inserted into phantoms of variable sizes, simulating the abdomens of a newborn, 5-year-old, 10-year-old, and adult-sized pediatric patients. Each phantom setting was scanned on an EICT using clinical size-specific kV dual-source protocols with a pitch of 3.0. The scans were performed with fixed scanning parameters, and the CTDIvol values of full dose were 0.30, 0.71, 1.05, and 7.40 mGy for newborn to adult size, respectively. In addition, half dose scans were acquired on EICT. Each phantom was then scanned on a PCCT (Siemens Alpha) using a universal 120-kV protocol with the same full dose and half dose as determined above on the EICT scanner. All other parameters matched to EICT settings. Virtual monoenergetic images were generated from PCCT scans between 40 and 80 keV with a 5-keV interval. Image quality metrics were compared between PCCT VMIs and EICT, including image noise (measured as standard deviation of solid water), contrast-to-noise ratio (CNR) (measured at iodine inserts with solid water as background), and noise power spectrum (measured in uniform phantom regions).

RESULTS

Noise at a PCCT VMI of 70 keV (7.0 ± 0.6 HU for newborn, 14.7 ± 1.6 HU for adult) is comparable (P > 0.05, t test) or significantly lower (P < 0.05, t test) compared with EICT (7.8 ± 0.8 HU for newborn, 15.3 ± 1.5 HU for adult). Iodine CNR from PCCT VMI at 50 keV (50.8 ± 8.4 for newborn, 27.3 ± 2.8 for adult) is comparable (P > 0.05, t test) or significantly higher (P < 0.05, t test) to the corresponding EICT measurements (57.5 ± 6.7 for newborn, 13.8 ± 1.7 for adult). The noise power spectrum curve shape of PCCT VMI is similar to EICT, despite PCCT VMI exhibiting higher noise at low keV levels.

CONCLUSIONS

The universal PCCT 120 kV with ultra-high pitch and postprocessed VMIs demonstrated equivalent or improved performance in noise (70 keV) and iodine CNR (50 keV) for pediatric abdominal CT, compared with size-specific kV images on the EICT.

Diagnostic performance and image quality of an image-based denoising algorithm applied to radiation dose-reduced CT in diagnosing acute appendicitis

PURPOSE

To evaluate diagnostic performance and image quality of ultralow-dose CT (ULDCT) in diagnosing acute appendicitis with an image-based deep-learning denoising algorithm (IDLDA).

METHODS

This retrospective multicenter study included 180 patients (mean ± standard deviation, 29 ± 9 years; 91 female) who underwent contrast-enhanced 2-mSv CT for suspected appendicitis from February 2014 to August 2016. We simulated ULDCT from 2-mSv CT, reducing the dose by at least 50%. Then we applied an IDLDA on ULDCT to produce denoised ULDCT (D-ULDCT). Six radiologists with different experience levels (three board-certified radiologists and three residents) independently reviewed the ULDCT and D-ULDCT. They rated the likelihood of appendicitis and subjective image qualities (subjective image noise, diagnostic acceptability, and artificial sensation). One radiologist measured image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). We used the receiver operating characteristic (ROC) analyses, Wilcoxon's signed-rank tests, and paired t-tests.

RESULTS

The area under the ROC curves (AUC) for diagnosing appendicitis ranged 0.90-0.97 for ULDCT and 0.94-0.97 for D-ULDCT. The AUCs of two residents were significantly higher on D-ULDCT (AUC difference = 0.06 [95% confidence interval, 0.01-0.11; p = .022] and 0.05 [0.00-0.10; p = .046], respectively). D-ULDCT provided better subjective image noise and diagnostic acceptability to all six readers. However, the response of board-certified radiologists and residents differed in artificial sensation (all p ≤ .003). D-ULDCT showed significantly lower image noise, higher SNR, and higher CNR (all p < .001).

CONCLUSION

An IDLDA can provide better ULDCT image quality and enhance diagnostic performance for less-experienced radiologists.

Enhanced visualization in endoleak detection through iterative and AI-noise optimized spectral reconstructions

To assess the image quality parameters of dual-energy computed tomography angiography (DECTA) 40-, and 60 keV virtual monoenergetic images (VMIs) combined with deep learning-based image reconstruction model (DLM) and iterative reconstructions (IR). CT scans of 28 post EVAR patients were enrolled. The 60 s delayed phase of DECTA was evaluated. Objective [noise, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR)] and subjective (overall image quality and endoleak conspicuity - 3 blinded readers assessment) image quality analyses were performed. The following reconstructions were evaluated: VMI 40, 60 keV VMI; IR VMI 40, 60 keV; DLM VMI 40, 60 keV. The noise level of the DLM VMI images was approximately 50% lower than that of VMI reconstruction. The highest CNR and SNR values were measured in VMI DLM images. The mean CNR in endoleak in 40 keV was accounted for as 1.83 ± 1.2; 2.07 ± 2.02; 3.6 ± 3.26 in VMI, VMI IR, and VMI DLM, respectively. The DLM algorithm significantly reduced noise and increased lesion conspicuity, resulting in higher objective and subjective image quality compared to other reconstruction techniques. The application of DLM algorithms to low-energy VMIs significantly enhances the diagnostic value of DECTA in evaluating endoleaks. DLM reconstructions surpass traditional VMIs and IR in terms of image quality.

Ultra-low-dose computed tomography with deep learning reconstruction for craniosynostosis at radiation doses comparable to skull radiographs: a pilot study

BACKGROUND

Craniofacial computed tomography (CT) is the diagnostic investigation of choice for craniosynostosis, but high radiation dose remains a concern.

OBJECTIVE

To evaluate the image quality and diagnostic performance of an ultra-low-dose craniofacial CT protocol with deep learning reconstruction for diagnosis of craniosynostosis.

MATERIALS AND METHODS

All children who underwent initial craniofacial CT for suspected craniosynostosis between September 2021 and September 2022 were included in the study. The ultra-low-dose craniofacial CT protocol using 70 kVp, model-based iterative reconstruction and deep learning reconstruction techniques was compared with a routine-dose craniofacial CT protocol. Quantitative analysis of the signal-to-noise ratio and noise was performed. The 3-dimensional (D) volume-rendered images were independently evaluated by two radiologists with regard to surface coarseness, step-off artifacts and overall image quality on a 5-point scale. Sutural patency was assessed for each of six sutures. Radiation dose was compared between the two protocols.

RESULTS

Among 29 patients (15 routine-dose CT and 14 ultra-low-dose CT), 23 patients had craniosynostosis. The 3-D volume-rendered images of ultra-low-dose CT without deep learning showed decreased image quality compared to routine-dose CT. The 3-D volume-rendered images of ultra-low-dose CT with deep learning reconstruction showed higher noise level, higher surface coarseness but decreased step-off artifacts, comparable signal-to-noise ratio and overall similar image quality compared to the routine-dose CT images. Diagnostic performance for detecting craniosynostosis at the suture level showed no significant difference between ultra-low-dose CT without deep learning reconstruction, ultra-low-dose CT with deep learning reconstruction and routine-dose CT. The estimated effective radiation dose for the ultra-low-dose CT was 0.05 mSv (range, 0.03-0.06 mSv), a 95% reduction in dose over the routine-dose CT at 1.15 mSv (range, 0.54-1.74 mSv). This radiation dose is comparable to 4-view skull radiography (0.05-0.1 mSv) and lower than previously reported effective dose for craniosynostosis protocols (0.08-3.36 mSv).

CONCLUSION

In this pilot study, an ultra-low-dose CT protocol using radiation doses at a level similar to skull radiographs showed preserved diagnostic performance for craniosynostosis, but decreased image quality compared to the routine-dose CT protocol. However, by combining the ultra-low-dose CT protocol with deep learning reconstruction, image quality was improved to a level comparable to the routine-dose CT protocol, without sacrificing diagnostic performance for craniosynostosis.

Computed Tomography 2.0: New Detector Technology, AI, and Other Developments

ABSTRACT

Computed tomography (CT) dramatically improved the capabilities of diagnostic and interventional radiology. Starting in the early 1970s, this imaging modality is still evolving, although tremendous improvements in scan speed, volume coverage, spatial and soft tissue resolution, as well as dose reduction have been achieved. Tube current modulation, automated exposure control, anatomy-based tube voltage (kV) selection, advanced x-ray beam filtration, and iterative image reconstruction techniques improved image quality and decreased radiation exposure. Cardiac imaging triggered the demand for high temporal resolution, volume acquisition, and high pitch modes with electrocardiogram synchronization. Plaque imaging in cardiac CT as well as lung and bone imaging demand for high spatial resolution. Today, we see a transition of photon-counting detectors from experimental and research prototype setups into commercially available systems integrated in patient care. Moreover, with respect to CT technology and CT image formation, artificial intelligence is increasingly used in patient positioning, protocol adjustment, and image reconstruction, but also in image preprocessing or postprocessing. The aim of this article is to give an overview of the technical specifications of up-to-date available whole-body and dedicated CT systems, as well as hardware and software innovations for CT systems in the near future.

Radiation dose reduction using deep learning-based image reconstruction for a low-dose chest computed tomography protocol: a phantom study

Background

The aim of this study was to compare the dose reduction potential and image quality of deep learning-based image reconstruction (DLIR) with those of filtered back-projection (FBP) and iterative reconstruction (IR) and to determine the clinically usable dose of DLIR for low-dose chest computed tomography (LDCT) scans.

Methods

Multi-slice computed tomography (CT) scans of a chest phantom were performed with various tube voltages and tube currents, and the images were reconstructed using seven methods to control the amount of noise reduction: FBP, three stages of IR, and three stages of DLIR. For subjective image analysis, four radiologists compared 48 image data sets with reference images and rated on a 5-point scale. For quantitative image analysis, the signal to noise ratio (SNR), contrast to noise ratio (CNR), nodule volume, and nodule diameter were measured.

Results

In the subjective analysis, DLIR-Low (0.46 mGy), DLIR-Medium (0.31 mGy), and DLIR-High (0.18 mGy) images showed similar quality to the FBP (2.47 mGy) image. Under the same dose conditions, the SNR and CNR were higher with DLIR-High than with FBP and all the IR methods (all P<0.05). The nodule volume and size with DLIR-High were significantly closer to the real volume than with FBP and all the IR methods (all P<0.001).

Conclusions

DLIR can improve the image quality of LDCT compared to FBP and IR. In addition, the appropriate effective dose for LDCT would be 0.24 mGy with DLIR-High.

Ultra-low-dose CT is feasible for torsion measurement of the lower limb in patients with metal implants

OBJECTIVES

Patients who need torsion measurement of the lower limb often have metal implants hindering e.g. MRI. A new ultra-low-dose (ULD-)CT protocol might be feasible for torsion measurement at cost of relatively low radiation exposure.

METHODS

We retrospectively included all patients with clinically indicated torsion measurement in the period July 2019 to June 2021 and metal implants in the scanning field. The ULD-CT protocol comprised automated tube current time product and automated tube voltage with reference settings of 100kV/20mAs (hip), 80kV/20mAs (knee) and 80kV/10mAs (ankle). Femoral neck anteversion, tibial, intra-articular knee and overall leg torsion measurements were performed by two radiologists independently. Diagnostic confidence regarding the delineation of the relevant cortical bone was rated on a 5-point Likert scale (1 = non-diagnostic, 5 = excellent).

RESULTS

102 consecutive patients could be included (BMI 27.38 ± 5.85) with 154 metal implants. Median total dose length product of the ULD-CT-torsion measurement was 16.5mGycm [11-39]. Both readers showed high agreement with a maximum torsional difference of 4.1°. Diagnostic confidence was rated best (5/5) in 92.2% (reader 1) and 93.1% (reader 2) with a worst rating of 3/5.

CONCLUSION

The new ULD-CT protocol is feasible for torsion measurement of the lower limb - even in patients with metal implants.

ADVANCES IN KNOWLEDGE

Metal implants are not an obstacle for ULD-CT torsion measurements of the lower limb.

The Impact of Novel Reconstruction Algorithms on Calcium Scoring: Results on a Dedicated Cardiac CT Scanner

Contemporary reconstruction algorithms yield the potential of reducing radiation exposure by denoising coronary computed tomography angiography (CCTA) datasets. We aimed to assess the reliability of coronary artery calcium score (CACS) measurements with an advanced adaptive statistical iterative reconstruction (ASIR-CV) and model-based adaptive filter (MBAF2) designed for a dedicated cardiac CT scanner by comparing them to the gold-standard filtered back projection (FBP) calculations. We analyzed non-contrast coronary CT images of 404 consecutive patients undergoing clinically indicated CCTA. CACS and total calcium volume were quantified and compared on three reconstructions (FBP, ASIR-CV, and MBAF2+ASIR-CV). Patients were classified into risk categories based on CACS and the rate of reclassification was assessed. Patients were categorized into the following groups based on FBP reconstructions: 172 zero CACS, 38 minimal (1-10), 87 mild (11-100), 57 moderate (101-400), and 50 severe (400<). Overall, 19/404 (4.7%) patients were reclassified into a lower-risk group with MBAF2+ASIR-CV, while 8 additional patients (27/404, 6.7%) shifted downward when applying stand-alone ASIR-CV. The total calcium volume with FBP was 7.0 (0.0-133.25) mm³, 4.0 (0.0-103.5) mm³ using ASIR-CV, and 5.0 (0.0-118.5) mm³ with MBAF2+ASIR-CV (all comparisons p < 0.001). The concomitant use of ASIR-CV and MBAF2 may allow the reduction of noise levels while maintaining similar CACS values as FBP measurements.

Low-contrast-dose liver CT using low monoenergetic images with deep learning-based denoising for assessing hepatocellular carcinoma: a randomized controlled noninferiority trial

OBJECTIVE

Low monoenergetic images obtained using noise-reduction techniques may reduce CT contrast media requirements. We aimed to investigate the effectiveness of low-contrast-dose CT using dual-energy CT and deep learning-based denoising (DLD) techniques in patients at high risk of hepatocellular carcinoma (HCC).

METHODS

We performed a prospective, randomized controlled noninferiority trial at a tertiary hospital between June 2019 and August 2020 (NCT04027556). Patients at high risk of HCC were randomly assigned (1:1) to the standard-contrast-dose group or low-contrast-dose group, which targeted a 40% reduction in contrast medium dose based on lean body weight. HCC conspicuity on arterial phase images was the primary endpoint with a noninferiority margin of 0.2. Images were independently assessed by three radiologists; model-based iterative reconstruction (MBIR) images of the standard-contrast-dose group and low monoenergetic (50-keV) DLD images of the low-contrast-dose group were compared using a generalized estimating equation.

RESULTS

Ninety participants (age 59 ± 10 years; 68 men) were analyzed. Compared with the standard-contrast-dose group (n = 47), 40% less contrast media was used in the low-contrast-dose group (n = 43) (107.0 ± 17.1 mL vs. 64.5 ± 11.3 mL, p < 0.001). In the arterial phase, HCC conspicuity on 50-keV DLD images in the low-contrast-dose group was noninferior to that of MBIR images in the standard-contrast-dose group (2.92 vs. 2.56; difference, 0.36; 95% confidence interval, -0.13 to ∞; p = 0.013).

CONCLUSIONS

The contrast dose in liver CT can be reduced by 40% without impairing HCC conspicuity when using 50-keV and DLD techniques.

KEY POINTS

• In the arterial phase, hepatocellular carcinoma conspicuity on 50-keV deep learning-based denoising images in the low-contrast-dose group was noninferior to that of model-based iterative reconstruction images in the standard-contrast-dose group. • HCC detection was comparable between 50-keV deep learning-based denoising images in the low-contrast-dose group and model-based iterative reconstruction images in the standard-contrast-dose group.

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.

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.

Accuracy of two deep learning–based reconstruction methods compared with an adaptive statistical iterative reconstruction method for solid and ground-glass nodule volumetry on low-dose and ultra–low-dose chest computed tomography: A phantom study

No published studies have evaluated the accuracy of volumetric measurement of solid nodules and ground-glass nodules on low-dose or ultra–low-dose chest computed tomography, reconstructed using deep learning–based algorithms. This is an important issue in lung cancer screening. Our study aimed to investigate the accuracy of semiautomatic volume measurement of solid nodules and ground-glass nodules, using two deep learning–based image reconstruction algorithms (Truefidelity and ClariCT.AI), compared with iterative reconstruction (ASiR-V) in low-dose and ultra–low-dose settings. We performed computed tomography scans of solid nodules and ground-glass nodules of different diameters placed in a phantom at four radiation doses (120 kVp/220 mA, 120 kVp/90 mA, 120 kVp/40 mA, and 80 kVp/40 mA). Each scan was reconstructed using Truefidelity, ClariCT.AI, and ASiR-V. The solid nodule and ground-glass nodule volumes were measured semiautomatically. The gold-standard volumes could be calculated using the diameter since all nodule phantoms are perfectly spherical. Subsequently, absolute percentage measurement errors of the measured volumes were calculated. Image noise was also calculated. Across all nodules at all dose settings, the absolute percentage measurement errors of Truefidelity and ClariCT.AI were less than 11%; they were significantly lower with Truefidelity or ClariCT.AI than with ASiR-V (all P<0.05). The absolute percentage measurement errors for the smallest solid nodule (3 mm) reconstructed by Truefidelity or ClariCT.AI at all dose settings were significantly lower than those of this nodule reconstructed by ASiR-V (all P<0.05). Furthermore, the lowest absolute percentage measurement errors for ground-glass nodules were observed with Truefidelity or ClariCT.AI at all dose settings. The absolute percentage measurement errors for ground-glass nodules reconstructed with Truefidelity at ultra–low-dose settings were significantly lower than those of all sizes of ground-glass nodules reconstructed with ASiR-V (all P<0.05). Image noise was lowest with Truefidelity (all P<0.05). In conclusion, the deep learning–based algorithms were more accurate for volume measurements of both solid nodules and ground-glass nodules than ASiR-V at both low-dose and ultra–low-dose settings.

Deep Learning-Based Image Conversion Improves the Reproducibility of Computed Tomography Radiomics Features: A Phantom Study

OBJECTIVES

This study aimed to evaluate the usefulness of deep learning-based image conversion to improve the reproducibility of computed tomography (CT) radiomics features.

MATERIALS AND METHODS

This study was conducted using an abdominal phantom with liver nodules. We developed an image conversion algorithm using a residual feature aggregation network to reproduce radiomics features with CT images under various CT protocols and reconstruction kernels. External validation was performed using images from different scanners, consisting of 8 different protocols. To evaluate the variability of radiomics features, regions of interest (ROIs) were drawn by targeting the liver parenchyma, vessels, paraspinal area, and liver nodules. We extracted 18 first-order, 68 second-order, and 688 wavelet radiomics features. Measurement variability was assessed using the concordance correlation coefficient (CCC), compared with the ground-truth image.

RESULTS

In the ROI-based analysis, there was an 83.3% improvement of CCC (80/96; 4 ROIs with 3 categories of radiomics features and 8 protocols) in synthetic images compared with the original images. Among them, the 56 CCC pairs showed a significant increase after image synthesis. In the radiomics feature-based analysis, 62.0% (3838 of 6192; 774 radiomics features with 8 protocols) features showed increased CCC after image synthesis, and a significant increase was noted in 26.9% (1663 of 6192) features. In particular, the first-order feature (79.9%, 115/144) showed better improvement in terms of the reproducibility of radiomics feature than the second-order (59.9%, 326/544) or wavelet feature (61.7%, 3397/5504).

CONCLUSIONS

Our study demonstrated that a deep learning model for image conversion can improve the reproducibility of radiomics features across various CT protocols, reconstruction kernels, and CT scanners.

Image quality assessment of pediatric chest and abdomen CT by deep learning reconstruction

BACKGROUND

Efforts to reduce the radiation dose have continued steadily, with new reconstruction techniques. Recently, image denoising algorithms using artificial neural networks, termed deep learning reconstruction (DLR), have been applied to CT image reconstruction to overcome the drawbacks of iterative reconstruction (IR). The purpose of our study was to compare the objective and subjective image quality of DLR and IR on pediatric abdomen and chest CT images.

METHODS

This retrospective study included pediatric body CT images from February 2020 to October 2020, performed on 51 patients (34 boys and 17 girls; age 1-18 years). Non-contrast chest CT (n = 16), contrast-enhanced chest CT (n = 12), and contrast-enhanced abdomen CT (n = 23) images were included. Standard 50% adaptive statistical iterative reconstruction V (ASIR-V) images were compared to images with 100% ASIR-V and DLR at medium and high strengths. Attenuation, noise, contrast to noise ratio (CNR), and signal to noise (SNR) measurements were performed. Overall image quality, artifacts, and noise were subjectively assessed by two radiologists using a four-point scale (superior, average, suboptimal, and unacceptable). A phantom scan was performed including the dose range of the clinical images used in our study, and the noise power spectrum (NPS) was calculated. Quantitative and qualitative parameters were compared using repeated-measures analysis of variance (ANOVA) with Bonferroni correction and Wilcoxon signed-rank tests.

RESULTS

DLR had better CNR and SNR than 50% ASIR-V in both pediatric chest and abdomen CT images. When compared with 50% ASIR-V, high strength DLR was associated with noise reduction in non-contrast chest CT (33.0%), contrast-enhanced chest CT (39.6%), and contrast-enhanced abdomen CT (38.7%) with increases in CNR at 149.1%, 105.8%, and 53.1% respectively. The subjective assessment of overall image quality and the noise was also better on DLR images (p < 0.001). However, there was no significant difference in artifacts between reconstruction methods. From NPS analysis, DLR methods showed a pattern of reducing the magnitude of noise while maintaining the texture.

CONCLUSION

Compared with 50% ASIR-V, DLR improved pediatric body CT images with significant noise reduction. However, artifacts were not improved by DLR, regardless of strength.

Advances in deep learning for computed tomography denoising

Computed tomography (CT) has seen a rapid increase in use in recent years. Radiation from CT accounts for a significant proportion of total medical radiation. However, given the known harmful impact of radiation exposure to the human body, the excessive use of CT in medical environments raises concerns. Concerns over increasing CT use and its associated radiation burden have prompted efforts to reduce radiation dose during the procedure. Therefore, low-dose CT has attracted major attention in the radiology, since CT-associated x-ray radiation carries health risks for patients. The reduction of the CT radiation dose, however, compromises the signal-to-noise ratio, which affects image quality and diagnostic performance. Therefore, several denoising methods have been developed and applied to image processing technologies with the goal of reducing image noise. Recently, deep learning applications that improve image quality by reducing the noise and artifacts have become commercially available for diagnostic imaging. Deep learning image reconstruction shows great potential as an advanced reconstruction method to improve the quality of clinical CT images. These improvements can provide significant benefit to patients regardless of their disease, and further advances are expected in the near future.

Deep Learning Algorithm for Simultaneous Noise Reduction and Edge Sharpening in Low-Dose CT Images: A Pilot Study Using Lumbar Spine CT

Objective

The purpose of this study was to assess whether a deep learning (DL) algorithm could enable simultaneous noise reduction and edge sharpening in low-dose lumbar spine CT.

Materials and Methods

This retrospective study included 52 patients (26 male and 26 female; median age, 60.5 years) who had undergone CT-guided lumbar bone biopsy between October 2015 and April 2020. Initial 100-mAs survey images and 50-mAs intraprocedural images were reconstructed by filtered back projection. Denoising was performed using a vendor-agnostic DL model (ClariCT.AI™, ClariPI) for the 50-mAS images, and the 50-mAs, denoised 50-mAs, and 100-mAs CT images were compared. Noise, signal-to-noise ratio (SNR), and edge rise distance (ERD) for image sharpness were measured. The data were summarized as the mean ± standard deviation for these parameters. Two musculoskeletal radiologists assessed the visibility of the normal anatomical structures.

Results

Noise was lower in the denoised 50-mAs images (36.38 ± 7.03 Hounsfield unit [HU]) than the 50-mAs (93.33 ± 25.36 HU) and 100-mAs (63.33 ± 16.09 HU) images (p < 0.001). The SNRs for the images in descending order were as follows: denoised 50-mAs (1.46 ± 0.54), 100-mAs (0.99 ± 0.34), and 50-mAs (0.58 ± 0.18) images (p < 0.001). The denoised 50-mAs images had better edge sharpness than the 100-mAs images at the vertebral body (ERD; 0.94 ± 0.2 mm vs. 1.05 ± 0.24 mm, p = 0.036) and the psoas (ERD; 0.42 ± 0.09 mm vs. 0.50 ± 0.12 mm, p = 0.002). The denoised 50-mAs images significantly improved the visualization of the normal anatomical structures (p < 0.001).

Conclusion

DL-based reconstruction may enable simultaneous noise reduction and improvement in image quality with the preservation of edge sharpness on low-dose lumbar spine CT. Investigations on further radiation dose reduction and the clinical applicability of this technique are warranted.

Dose reduction potential of vendor-agnostic deep learning model in comparison with deep learning-based image reconstruction algorithm on CT: a phantom study

Objectives

To compare the dose reduction potential (DRP) of a vendor-agnostic deep learning model (DLM, ClariCT.AI) with that of a vendor-specific deep learning–based image reconstruction algorithm (DLR, TrueFidelity™).

Methods

Computed tomography (CT) images of a multi-sized image quality phantom (Mercury v4.0) were acquired under six radiation dose levels (0.48/0.97/1.93/3.87/7.74/15.47 mGy) and were reconstructed using filtered back projection (FBP) and three strength levels of the DLR (low/medium/high). The FBP images were denoised using the DLM. For all DLM and DLR images, the detectability index (d′) (a task-based detection performance metric) was obtained, under various combinations of three target sizes (10/5/1 mm), five inlets (CT value difference with the background; −895/50/90/335/1000 HU), five phantom diameters (36/31/26/21/16 cm), and six radiation dose levels. Dose reduction potential (DRP) measures the dose reduction made by using DLM or DLR, while yielding d′ equivalent to that of FBP at full dose.

Results

The DRPs of the DLM, DLR-low, DLR-medium, and DLR-high were 86% (81–88%), 60% (46–67%), 76% (60–81%), and 87% (78–92%), respectively. For 10-mm targets, the DRP of the DLM (87%) was higher than that of all DLR algorithms (58–86%). However, for smaller targets (5 mm/1 mm), the DRPs of the DLR-high (89/88%) were greater than those of the DLM (87/84%).

Conclusion

The dose reduction potential of the vendor-agnostic DLM was shown to be comparable to that of the vendor-specific DLR at high strength and superior to those of the DLRs at medium and low strengths.

Key Points

• DRP of the vendor-agnostic model was comparable to that of high-strength vendor-specific model and superior to those of medium- and low-strength models.

• Under various radiation dose levels, the deep learning model shows higher detectability indexes compared to FBP.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00330-021-08199-9.

Image quality of ultralow-dose chest CT using deep learning techniques: potential superiority of vendor-agnostic post-processing over vendor-specific techniques

OBJECTIVE

To compare the image quality between the vendor-agnostic and vendor-specific algorithms on ultralow-dose chest CT.

METHODS

Vendor-agnostic deep learning post-processing model (DLM), vendor-specific deep learning image reconstruction (DLIR, high level), and adaptive statistical iterative reconstruction (ASiR, 70%) algorithms were employed. One hundred consecutive ultralow-dose noncontrast CT scans (CTDI_vol; mean, 0.33 ± 0.056 mGy) were reconstructed with five algorithms: DLM-stnd (standard kernel), DLM-shrp (sharp kernel), DLIR, ASiR-stnd, and ASiR-shrp. Three thoracic radiologists blinded to the reconstruction algorithms reviewed five sets of 100 images and assessed subjective noise, spatial resolution, distortion artifact, and overall image quality. They selected the most preferred algorithm among five image sets for each case. Image noise and signal-to-noise ratio were measured. Edge-rise-distance was measured at a pulmonary vessel, i.e., the distance between two points where attenuation was 10% and 90% of maximal intravascular intensity. The skewness of attenuation was calculated in homogeneous areas.

RESULTS

DLM-stnd, followed by DLIR, showed the best subjective noise on both lung and mediastinal windows, while DLIR yielded the least measured noise (ps < .0001). Compared to DLM-stnd, DLIR showed inferior subjective spatial resolution on lung window and higher edge-rise-distance (ps < .0001). Additionally, DLIR showed the most frequent distortion artifacts and deviated skewness (ps < .0001). DLM-stnd scored the best overall image quality, followed by DLM-shrp and DLIR (mean score 3.89 ± 0.19, 3.68 ± 0.24, and 3.53 ± 0.33; ps < .001). Two among three readers preferred DLM-stnd on both windows.

CONCLUSION

Although DLIR provided the best quantitative noise profile, DLM-stnd showed the best overall image quality with fewer artifacts and was preferred by two among three readers.

KEY POINTS

• A vendor-agnostic deep learning post-processing algorithm applied to ultralow-dose chest CT exhibited the best image quality compared to vendor-specific deep learning algorithm and ASiR techniques. • Two out of three readers preferred a vendor-agnostic deep learning post-processing algorithm in comparison to vendor-specific deep learning algorithm and ASiR techniques. • A vendor-specific deep learning reconstruction algorithm yielded the least image noise, but showed significantly more frequent specific distortion artifacts and increased skewness of attenuation compared to a vendor-agnostic algorithm.

Noise reduction approach in pediatric abdominal CT combining deep learning and dual-energy technique

OBJECTIVES

To evaluate the image quality of low iodine concentration, dual-energy CT (DECT) combined with a deep learning-based noise reduction technique for pediatric abdominal CT, compared with standard iodine concentration single-energy polychromatic CT (SECT).

METHODS

From December 2016 to May 2017, DECT with 300 mg•I/mL contrast medium was performed in 29 pediatric patients (17 boys, 12 girls; age, 2-19 years). The DECT images were reconstructed using a noise-optimized virtual monoenergetic reconstruction image (VMI) with and without a deep learning method. SECT images with 350 mg•I/mL contrast medium, performed within the last 3 months before the DECT, served as reference images. The quantitative and qualitative parameters were compared using paired t tests and Wilcoxon signed-rank tests, and the differences in radiation dose and total iodine administration were assessed.

RESULTS

The linearly blended DECT showed lower attenuation and higher noise than SECT. The 60-keV VMI showed an increase in attenuation and higher noise than SECT. The combined 60-keV VMI plus deep learning images showed low noise, no difference in contrast-to-noise ratios, and overall image quality or diagnostic image quality, but showed a higher signal-to-noise ratio in the liver and lower enhancement of lesions than SECT. The overall image and diagnostic quality of lesions were maintained on the combined noise reduction approach. The CT dose index volume and total iodine administration in DECT were respectively 19.6% and 14.3% lower than those in SECT.

CONCLUSION

Low iodine concentration DECT, combined with deep learning in pediatric abdominal CT, can maintain image quality while reducing the radiation dose and iodine load, compared with standard SECT.

KEY POINTS

• An image noise reduction approach combining deep learning and noise-optimized virtual monoenergetic image reconstruction can maintain image quality while reducing radiation dose and iodine load. • The 60-keV virtual monoenergetic image reconstruction plus deep learning images showed low noise, no difference in contrast-to-noise ratio, and overall image quality, but showed a higher signal-to-noise ratio in the liver and a lower enhancement of lesion than single-energy polychromatic CT. • This combination could offer a 19.6% reduction in radiation dose and a 14.3% reduction in iodine load, in comparison with a control group that underwent single-energy polychromatic CT with the standard protocol.

Incremental Image Noise Reduction in Coronary CT Angiography Using a Deep Learning-Based Technique with Iterative Reconstruction

Objective

To assess the feasibility of applying a deep learning-based denoising technique to coronary CT angiography (CCTA) along with iterative reconstruction for additional noise reduction.

Materials and Methods

We retrospectively enrolled 82 consecutive patients (male:female = 60:22; mean age, 67.0 ± 10.8 years) who had undergone both CCTA and invasive coronary artery angiography from March 2017 to June 2018. All included patients underwent CCTA with iterative reconstruction (ADMIRE level 3, Siemens Healthineers). We developed a deep learning based denoising technique (ClariCT.AI, ClariPI), which was based on a modified U-net type convolutional neural net model designed to predict the possible occurrence of low-dose noise in the originals. Denoised images were obtained by subtracting the predicted noise from the originals. Image noise, CT attenuation, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were objectively calculated. The edge rise distance (ERD) was measured as an indicator of image sharpness. Two blinded readers subjectively graded the image quality using a 5-point scale. Diagnostic performance of the CCTA was evaluated based on the presence or absence of significant stenosis (≥ 50% lumen reduction).

Results

Objective image qualities (original vs. denoised: image noise, 67.22 ± 25.74 vs. 52.64 ± 27.40; SNR [left main], 21.91 ± 6.38 vs. 30.35 ± 10.46; CNR [left main], 23.24 ± 6.52 vs. 31.93 ± 10.72; all p < 0.001) and subjective image quality (2.45 ± 0.62 vs. 3.65 ± 0.60, p < 0.001) improved significantly in the denoised images. The average ERDs of the denoised images were significantly smaller than those of originals (0.98 ± 0.08 vs. 0.09 ± 0.08, p < 0.001). With regard to diagnostic accuracy, no significant differences were observed among paired comparisons.

Conclusion

Application of the deep learning technique along with iterative reconstruction can enhance the noise reduction performance with a significant improvement in objective and subjective image qualities of CCTA images.