Thank you for visiting us at this year's RSNA!
Scientific Evidence
*This information may not be within the Indicated uses for Canon Medical System devices. The views and findings included here are those of the presenter. Clinical settings and other considerations may impact repeating these results. The results may not have been validated by Canon or submitted to FDA for market clearance.
CANON PCCT: Photon Counting CT
T5A-SPPH-6 - Low Contrast Detectability Comparison between a Prototype Photon-Counting Computed Tomography System and a Conventional CT system across a range of attenuation levels
*Conclusions: The PCCT offers a considerable increase in low contrast performance, particularly in low dose/large size scenarios
*Clinical Relevance/Application: The PCCT counting images offer better general low contrast detectability than the EID system and demonstrates the potential to achieve better performance in low dose scenarios, such as lung cancer screening or with bariatric patients.
W5B-STCE-2 - Quantitative Image Quality Comparison Between Photon Counting and Conventional CT Systems: Contrast-To-Noise Ratio of Iodine and Calcium
*Conclusions: We compared quantitative image quality between a prototype PCCT and a commercial EID-C. Results demonstrate that PCCT significantly improves CNR in both counting and spectral modes, which can potentially lead to reduced input iodine contrast load in clinical practice with equal or better diagnostic capability.
*Clinical Relevance/Application: The initial performance of a prototype PCCT system in both counting and spectral imaging modes demonstrates its potential to achieve better diagnostic performance with reduced iodine contrast load.
T5B-SPPH-6 - A Clinical Prototype CZT-Based Photon-Counting CT Scanner: First Experience in Low-Dose Lung Cancer Screening
*Conclusions: The preliminary investigation of the new CZT PCD-CT prototype scanner showed improvements in CT number stability and CNR low-dose lung cancer screening tasks. This may lead to better detection and characterization of lung pathologies at reduced radiation doses or with better CNR. Further studies are warranted to assess the clinical performance of the prototype scanner.
*Clinical Relevance/Application: Our initial investigation of the new CZT PCD-CT prototype showed improved performance in CT number stability and CNR specifically in low-dose lung CT scans which may lead to better detection and assessment of ground-glass nodules. The improved CNR may be used to lower the radiation dose of screening CT scans which will help include younger patients or patients with less packs per year in lung cancer screening.
CANON AI: PIQE - Precise IQ Engine
S3A-SPCA-1 - Improvement of Coronary Stent CT Imaging with Super-Resolution Deep-Learning Reconstruction: An Initial In Vivo Experience
*Conclusions: SR-DLR improves the objective and subjective qualities of coronary stent imaging compared with HIR, MBIR, and NR-DLR algorithms.
*Clinical Relevance/Application: Novel SR-DLR algorithm allows improvement of coronary stent evaluation because of excellent spatial resolution, lower blooming artifact, and stent-induced beam-hardening artifacts.
PIQE is designed to fully utilize the maximum resolution of the detector.
S3A-SPCA-4 - Super-Resolution Deep Learning Reconstruction for Cardiac CT: Comparison of Stenosis Evaluation Accuracy, Image Quality, Stenosis Confidence Level at Coronary Arteries and CAD-RADS Classification with Hybrid-Type and Model-Based Iterative Reconstructions and Deep Learning
*Conclusions: SR-DLR has a potential to improve stenosis evaluation accuracy, image quality and stenosis confidence level on cardiac CT than others at in vivo and in vitro studies.
*Clinical Relevance/Application: SR-DLR has a potential to improve stenosis evaluation accuracy, image quality and stenosis confidence level than others on cardiac ADCT.
PIQE is designed to fully utilize the maximum resolution of the detector.
M5A-SPPH-7 - Importance of Larger Reconstruction Matrix Sizes for Super-Resolution Deep Learning Reconstruction at In Vivo and In Vitro Studies
*Conclusions: To fully utilize the spatial resolution of SR-DLR, it is suitable to reconstruct images with 1024 matrices.
*Clinical Relevance/Application: The SR-DLR of 1024 matrix images provides higher spatial resolution, which is useful for the visualization of fine structures such as demonstration of plaques in the coronary artery, and fine lesions in diffuse lung disease.
PIQE is designed to fully utilize the maximum resolution of the detector.
R5B-SPPH-1 - Task-Based Image Quality Assessments of Super-Resolution Deep-Learning Reconstruction for Coronary CT Angiography
*Conclusions: Compared with HIR, MBIR, and NR-DLR algorithms, newly introduced SR-DLR improved the task-based image quality and detectability for low- and high-contrast objects relevant to coronary CT angiography.
*Clinical Relevance/Application: SR-DLR provides CT images with excellent spatial resolution, preferable noise texture, and object detectability, potentially facilitating the interpretation of coronary artery diseases.
PIQE is designed to fully utilize the maximum resolution of the detector.
CANON AI: AiCE
M5A-SPPH-6 - Signal-To-Noise Ratio Measurement in CT With Conventional and Textured Phantoms: Comparison of Two Deep-Learning Image Reconstruction Techniques
*Conclusions: Both DLIR techniques sufficiently improved the conventional SNR measured with the rod and water images (88-500%). However, the improvements of SNRNU were low (10-47%), which revealed limited performances for the complex structures included in the textured phantom.
*Clinical Relevance/Application: The SIFs measured using the non-uniform region are possibly more clinically relevant compared to the conventional method using simple rods and a uniform water region.
M5B-SPPH-5 - Differences in Noise Reduction Performance Between Two Deep-Learning CT Image Reconstruction Techniques: A Novel Textured Phantom Approach
*Conclusions: This study has revealed a distinct difference in characteristics between the two DLIR techniques: TF tends to offer less aggressive noise reduction in non-uniform regions and preserve the original signals, whereas AC tends to prioritize noise filtering over edge-preservation, especially at the low-dose condition and with the high noise reduction strength.
*Clinical Relevance/Application: An appropriate combination of strength and dose level should be selected for clinical use of each DLIR technique.
M5B-SPPH-6 - Discriminability of Frequency-Dependent Non-Gaussian Noise Properties in CT and the Limitation of the Noise Power Spectrum to Describe Noise Texture with Non-Linear Reconstruction Algorithms.
*Conclusions: This work demonstrates that the appearance of CT noise texture may be dependent on frequency-dependent higher-order statistics not captured by the NPS. In some cases, noise textures with identical NPS and SD can be distinguished based on non-Gaussian properties. Deep-learning reconstruction has lower levels of excess kurtosis and is less readily discriminable from Gaussian noise, despite having a similar NPS to MBIR.
*Clinical Relevance/Application: Noise texture can influence the acceptability of a reconstruction algorithm to the radiologist and thus full characterization of the noise is key to understanding and describing reconstruction algorithm performance.
S5-SSGI04-1 - Computer-Aided Liver Tumor Evaluations: Comparison of Radiation Dose Reduction Capability among FBP, Hybrid-Type and Mode-Based IRs and Deep Learning Reconstruction in Phantom Study
*Conclusions: DLR is more useful for radiation dose reduction as having superior image quality and measurement accuracy than FBP, hybrid IR or MBIR, when applied computer-aided liver tumor evaluation.
*Clinical Relevance/Application: DLR is more useful for radiation dose reduction as having superior image quality and measurement accuracy than FBP, hybrid IR or MBIR, when applied computer-aided liver tumor evaluation.
M5B-SPPH-1 - Newly Developed Deep Learning Reconstruction (DLR) for Dynamic Contrast-Enhanced CT Angiography: Comparison of Capability for Image Quality Improvements and Abdominal Arteries Assessments among DLR, Hybrid-Type and Model-Based Iterative Reconstructions (IRs) in Patients with Hepatic and Pancreatic Malignant Tumors
*Conclusions: Newly developed DLR for dynamic CE-CTA has equal or superior capability to other methods, especially hybrid-type IR, for improving image quality and vasculature assessment in routine clinical practice.
*Clinical Relevance/Application: Newly developed DLR for dynamic CE-CTA has equal or superior capability to other methods, especially hybrid-type IR, for improving image quality and vasculature assessment in routine clinical practice.
CANON PRECISION
R5B-SPPH-3 - Physical Properties of a New Deep Learning Reconstruction in Ultra-High-Resolution Images: Comparison with Model-Based Iterative Reconstruction
*Conclusions: There was no difference in the resolution characteristics of MBIR and DLR. DLR had better noise characteristics than MBIR. DLR is more useful than MBIR for with U-HR images.
*Clinical Relevance/Application: DLR would be more useful than MBIR for obtaining U-HR images. In contrast to a normal CT image (matrix size: 512 × 512, slice thickness: 0.5 mm), a high-resolution images (matrix size: 1024 × 1024, slice thickness: 0.25 mm) has a smaller voxel size. Therefore, if we consider only the statistics of photon detection, the noise in the image is about four times larger. In other words, in order to achieve a noise level equivalent to that of a conventional image, the dose needs to be increased by a factor of 16. Fine sampling requires a combined image noise suppression technique, and as mentioned above, DLR maybe effective in reducing image noise. To achieve higher spatial resolution and less image noise, DLR would be more useful than MBIR.
W7-SSCH07-4 - High-Spatial-Resolution CT from Normal-Resolution to Super High-Resolution data: Influence on Pathological Invasiveness Prediction of Lung Adenocarcinoma with Machine-Learning Radiomics Software
*Conclusions: Spatial resolution significantly influences on the software for pathological invasiveness prediction of lung adenocarcinoma on high-spatial-resolution CT. When applied SHR data, prediction capability is considered superior to NR data as well as radiologists’ evaluation.
*Clinical Relevance/Application: SHR data from high-spatial-resolution CT may significantly improve pathological invasiveness prediction of lung adenocarcinoma by machine-learning radiomics software, leading to support of radiologists.
W5B-SPPH-8 - Robustness of Texture Biomarkers of Bone to Spatially Variant Blur in Ultra High Resolution CT
*Conclusions: Due to the non-stationary spatial resolution associated with view sampling, texture features of bone derived from UHR CT exhibit substantial dependence on spatial location, especially for fast acquisitions. Ongoing work investigates the impact of this inter-scan variability on radiomic models of bone health based on assays of texture biomarkers.
*Clinical Relevance/Application: Development of radiomic models of bone health from Ultra-High Resolution CT (UHR CT) data requires consideration of inter-scan texture variability due to spatially variant view sampling blur.
S3B-SPCH-2 - Quantitative Evaluation of Airway Disease in Rheumatoid Arthritis Using Ultra-high-resolution CT
*Conclusions: AWV%rl is a useful parameter to evaluate airway disease in RA.
*Clinical Relevance/Application: Bronchiolitis and bronchiectasis are significant features of the RA lung. U-HRCT enables clinicians to extract peripheral bronchi and measure AWVrl%, which is easier than measuring intra-bronchial area and wall-thickness.
S2-SSGI02-3 - Synergy of a Novel Oral Contrast Agent and High-Resolution CT for Improved Bowel Fold Spatial Resolution: A Phantom Study
*Conclusions: Compared to air, neutral, and positive contrast, DBCM improves simulated bowel fold delineation at soft tissue W/L settings, including for different kernel reconstructions, CTDIvol, and at normal and high-resolution CT acquisitions.
*Clinical Relevance/Application: A dark contrast agent may improve visualization of fine bowel folds at conventional and high-resolution CT and warrants further studies in clinical applications such as CT colonography or CT enterography.
T5A-SPGI-10 - Radiomic Features of Hepatocellular Carcinoma at Ultra-High-Resolution CT with a 1024 Matrix: Dependencies on Matrix Size and Reconstruction Algorithm
*Conclusions: Both matrix size and reconstruction algorithm affect the radiomic features in hepatocellular carcinoma at UHR CT. Above all, 1024-matrix size had a significant effect on approximately 90 % of the first-order features and texture features.
*Clinical Relevance/Application: It should be noted that radiomics from UHR CT with a 1024 matrix have different properties than those from conventional CT with a 512 matrix.
M5A-SPGI-6 - Ultra-high resolution computed tomography with deep-learning-reconstruction: Diagnostic ability in the assessment of gastric cancer and the depth of invasion
*Conclusions: U-HRCT with improved AiCE-BS is capable of depicting the three-layered structure of the gastric wall more clearly than the other two reconstruction methods, and is useful for the visualization of gastric cancers as well as for the assessment of depth invasion.
*Clinical Relevance/Application: U-HRCT using new DLR method will improve overall image quality, delineation of the three-layered structure of gastric wall, visibility of gastric cancer, and diagnostic ability for depth invasion.
T5A-SPPH-7 - Cardiac-Induced Motion of the Abdominal Organs and Its Effect on Ultra-High-Resolution CT
*Conclusions: The spatial resolution of UHR CT can easily be impacted by cardiac-induced motion in the abdominal organs. Here we established that motion-correction or abdominal CT gating strategies may be necessary to fully take advantage of this new technology.
*Clinical Relevance/Application: With the new advances in the spatial resolution of CT, the old sub-pixel biological motions now amount to up 10 pixels with the new resolution. Therefore, if left uncorrected these motions will blur the UHR images. Here we present our measurements of the pancreatic displacement during a cardiac cycle, and recommend ways to correct it.
W3-SSHN03-2 - Super-High-Resolution CT with Oral Spacer vs Conventional Radiological Examinations: Capability for Image Quality Improvement and Accurate Depth of Invasion Measurement in Oral Tongue Squamous Cell Carcinoma
*Conclusions: In OSCC patients, SHR-CT with oral spacer can improve image quality and more accurately measure DOI than MRI. When compared with MRI, our proposed technique can reduce DOI measurement error from pathological result.
*Clinical Relevance/Application: In OSCC patients, CE-SHR-CT with oral spacer can improve image quality and more accurately measured DOI than CE-T1WI.
R5A-SPPH-5 - Patient-Specific Analysis of Dose Metrics and Organ Doses Based on Monte Carlo Simulation in Multiphase Abdominal CT Examinations
*Conclusions: Organ doses for organs within or partially included in scan range in multiphase abdominal CT examinations will widely vary depending on the types of CT scanners and patient size. Organ-specific SSDE can be useful for total organ dose estimations for organs within scan range in the CT examinations.
*Clinical Relevance/Application: The reduction of organ doses for individual patients in multiphase abdominal CT examinations will require the optimization of scanning parameters including tube current modulation for each CT scanner.
CANON: CT Guided Interventions
W7-SSPH11-3 - Physician Dose Reduction from Optimized Interventional MDCT Fluoroscopy Scan Angles
*Conclusions: For equal patient dose, a physician can lower their in-room scatter dose by 33.4% relative to using a full scan. This is under the same conditions of patient absorbed dose.
*Clinical Relevance/Application: Physicians performing interventional MDCT procedures with fluoroscopy receive scatter dose. Shutting off x-rays when the tube is closest to the physician can reduce physician dose by 33.4% without changing the patient dose.
CANON: CT General
S2-SSPH02-2 - Silver (Ag) X-ray Spectrum Modulation Filter and Deep Learning Reconstruction: Comparison of Radiation Dose Reduction Capability for CT Lung Nodule Detection with Copper Filter and Clinically Applied Reconstruction Methods at Chest Phantom Study
*Conclusions: Ag filter can significantly improve image quality and nodule detection capability than Cu filter at each radiation dose. With keeping noduledetection rate as no significant difference with standard protocol, combination of Ag filter and DLR can achieve further radiation dose reduction as compared with other protocols.
*Clinical Relevance/Application: Ag filter can significantly improve image quality and nodule detection capability than Cu filter at each radiation dose. With keeping noduledetection rate as no significant difference with standard protocol, combination of Ag filter and DLR can achieve further radiation dose reduction as compared with other protocols.
S3A-SPCH-7 - Dynamic Contrast-Enhanced Whole-Lung Perfusion Area-Detector CT: Capability for Lung Structure Change and Pulmonary Functional Loss Evaluations in Operated Stage I Lung Cancer Patients
*Conclusions: Whole-lung dynamic first-pass CE-perfusion ADCT is useful for pathological change and pulmonary functional loss evaluations in operated stage I lung cancer patients.
*Clinical Relevance/Application: Whole-lung dynamic first-pass CE-perfusion ADCT is useful for pathological change and pulmonary functional loss evaluations in operated stage I lung cancer patients.
W2-SPCH-6 - Imaging the Thymus and Anterior Mediastinum with Subtraction Imaging in CT: An Initial Clinical Experience
*Conclusions: Hence we demonstrate no difference in iodine concentration between benign and malignant anterior mediastinal lesions. However further work is required to explore if iodine concentration from image subtraction CT could contribute to diagnosis when in concert with other imaging features seen on conventional CT imaging.
*Clinical Relevance/Application: The addition of subtraction imaging to CT assessment of anterior mediastinal mass is of questionable value as we show iodine concentration cannot discriminate between benign and malignant lesions.
R5A-SPPH-5 - Patient-Specific Analysis of Dose Metrics and Organ Doses Based on Monte Carlo Simulation in Multiphase Abdominal CT Examinations
*Conclusions: Organ doses for organs within or partially included in scan range in multiphase abdominal CT examinations will widely vary depending on the types of CT scanners and patient size. Organ-specific SSDE can be useful for total organ dose estimations for organs within scan range in the CT examinations.
*Clinical Relevance/Application: The reduction of organ doses for individual patients in multiphase abdominal CT examinations will require the optimization of scanning parameters including tube current modulation for each CT scanner.
M5A-SPIR-2 - Feasibility of EKG gated CT-Fluoroscopy to Guide Percutaneous Biopsy of Cardiac and Pericardial Masses
*Conclusions: Percutaneous cardiac and pericardial mass biopsies can be safely performed with an acceptable clinical risk and radiation dose profile. CT-fluoroscopy with real-time non-gated and gated image visualization, a well-trained interventional care team, and the backup availability of cardiac surgery may be prerequisites to ensure the best possible patient care.
*Clinical Relevance/Application: Percutaneous cardiac and pericardial biopsies represent a minimally invasive alternative to mediastinoscopy and open cardiac surgery in a specific patient population.
T7-SSCH05-1 - Keynote Speaker: Advancing MRI and CT for Thoracic Oncology and Airway Diseases
MRI has been tested mainly oncologic patients and suggested as useful for nodule evaluation, tumor staging or therapeutic effect prediction, etc. since 2000. Moreover, it has also been able to evaluate functional information with perfusion, ventilation or structure-based information in the last a few decades. Therefore, MRI is considered as useful in not only oncologic, but also other chest disease in academic and clinical purposes.
At the same time periods, CT techniques have also been advanced for improving temporal or spatial resolution, reducing radiation dose, enhancing contrast with dual-energy or subtraction techniques as well as newly introduced real functional examinations. In addition, artificial intelligence and radiomics have been also tested for quantitative assessment in chest diseases.
In this keynote lecture, I will summarize current status of chest MRI in thoracic oncology and currently applicable CT and MRI examinations for airway disease evaluations for academic and clinical purpose.
