Super-resolution Deep-learning Reconstruction: Image Quality and Myocardial Homogeneity in Coronary Computed Tomography Angiography | W2-SPCA-7

Sung Min Ko, PhD - Seoul National University

Conclusions: SR-DLR improved image quality, demonstrated clearer delineation of distal segments of coronary arteries, and was seemingly accurate for quantifying CT attenuation in the myocardium.

Clinical Relevance/Application: The results of this study demonstrate that SR-DLR achieves higher image quality, better visualization of small distal segments of coronary arteries, and minimum inhomogeneous myocardium attenuation on coronary CT angiography than those associated with FBP, hybrid IR, and DLR.

Wednesday, Nov 29 9:00 AM - 9:30 AM CST @ Rm S404

Impact of Super-Resolution Deep Learning Reconstruction Technique on Dynamic Myocardial Computed Tomography Perfusion Imaging | W5A-SPCA-4

Tomoro Morikawa, MD – Ehime University

Conclusions: SR-DLR was superior to HIR and DLR with respect to the image noise and the sharpness of myocardial margins without altering CT-MBF quantification in dynamic myocardial CTP imaging.

Clinical Relevance/Application: SR-DLR allows for improving the image noise and sharpness more effectively than conventional reconstruction techniques in dynamic myocardial CTP imaging without requiring hardware changes. SR-DLR has a potential to improve the detectability of myocardial perfusion abnormality by taking advantage of the high spatial resolution and noise reduction capability.

Wednesday, Nov 29 12:15 PM - 12:45 PM CST @ Learning Center

Coronary Artery Calcium Quantification with Super Resolution Deep-learning Reconstruction Algorithm: An Anthropomorphic Phantom Study | W5B-SPCA-6

Shinsuke Shigematsu - Kumamoto University

Conclusions: SR-DLR may improve image quality and facilitate accurate CAC quantification even at lower radiation doses, but its clinical application requires caution due to the deviation of the Agatston score from the conventional high-dose FBP setting.

Clinical Relevance/Application: The novel SR-DLR algorithm may yield accurate detection and quantification of CAC due to its excellent noise and spatial resolution characteristics, but the clinical evaluation for potential impact on risk classification and patient management is warranted before routine clinical use.

Wednesday, Nov 29 12:15 PM - 12:45 PM CST @ Learning Center

Coronary Stent Evaluation by CCTA using Super-Resolution Deep Learning Reconstruction: Compared with Invasive Coronary Angiography | W2-SPCA-4

Cheng Xu - Peking Union Medical College Hospital

Conclusions: Compared with HIR and DLR algorithm, SR-DLR improves the visible and diagnostic performance of CCTA in stents evaluation.

Clinical Relevance/Application: SR-DLR algorithm improves the diagnostic accuracy and confidence of coronary stent evaluation.

Thursday, Nov 30 12:45 PM - 1:15 PM CST @ Learning Center

Super-resolution Deep-learning Reconstruction: Image Quality and Myocardial Homogeneity in Coronary Computed Tomography Angiography | W2-SPCA-7

Patrik Rogalla MD – University Health Network

Conclusions: The SR-DLR using a 1024 matrix provides superior clinical image quality in chest CT; SR-DLR thin slices are overall preferred for lung parenchyma. SR-DLR holds promise to eliminate the need for separate reconstructions for the lung parenchyma and soft tissues.

Clinical Relevance/Application: The SR-DLR outperforms current clinical standard-of-care reconstructions and may contribute to improving the diagnostic value of chest CT

Thursday, Nov 30 12:45 PM - 1:15 PM CST @ Learning Center

Deep-learning CT Reconstruction in the Abdomen: A Systematic Review and Meta-Analysis | S2-SSGI02-4

Mostafa A Shehata, MD, Mayo Clinic

Conclusions: Deep learning reconstructions significantly improve image quality in CT of the abdomen. Assessment of other dose levels and clinical indications is necessary. Careful choice of radiation dose levels is necessary, particularly for small liver lesion assessment.

Clinical Relevance: The use of DLR reconstruction techniques could lead to more accurate diagnosis and treatment planning, while also reducing the risks associated with radiation exposure during CT imaging. However, careful selection of appropriate radiation doses is necessary to avoid signal loss and blurring at high DLR reconstruction strengths.

Sunday, Nov 26, 10:30 AM - 11:30 AM CST Rm S404

Dark Blood Computed Tomography Angiography Combined with Deep Learning Reconstruction for Thickened Carotid Artery Wall Imaging in Takayasu Arteritis Patients | S3A-SPVA-2

Tong Su, MD - Peking Union Medical College Hospital

Conclusions: Compared to HIR, dark blood method combined with DLR reconstruction improved CTA image quality, and enhanced the visualization of thickened vessel wall of TAK patients.

Clinical Relevance/Application: Dark blood method combined with DLR reconstruction is conducive to rapid and accurate measurement of vessel wall thickness.

Sunday, Nov 26 11:45 AM - 12:15 PM CST @ Learning Center

Comparison of Imaging Methods for Ultra-high-resolution CT to Delineate Very Small Vessels: Low Tube Voltage Imaging vs. Small Focal Spot Imaging | M5B-SPPH-5

Masatoshi Hori, MD, PhD - Kobe University

Conclusions: In microvascular imaging, low-tube-voltage imaging is effective when a sufficient contrast cannot be obtained, whereas small-focus imaging is effective when a large imaging dose is required.

Clinical Relevance/Application: In ultra-high-resolution CT, if the patient is large, rapid injection of a high-density contrast agent and small-focus imaging at 120 kVp can be used to ensure depiction of very small vessels, although the dose will increase. If the patient is small, Low-tube-voltage imaging can be used to ensure the depiction of very small vessels without increased invasiveness of rapid injection of high-density contrast medium.

Monday, Nov 27 12:45 PM - 1:15 PM CST @ Learning Center

Application of Deep Learning Reconstruction Algorithm (AiCE) in Low-dose CT Scanning of Sacroiliac Joints | T5B-SPMK-3

Likun Cao, MD - Peking Union Medical College Hospital

Conclusions: The use of DLR can reduce the radiation dose and improve the image quality of sacroiliac joint CT, with comparable efficacy of diagnosis to HIR. Thus, DLR algorithm is helpful to improve the safety od CT scanning of sacroiliac joint.

Clinical Relevance/Application: The DLR algorithm, Deep Learning Reconstruction algorithm(AiCE), can help improve image quality and detect sacroiliac joint diseases with extremely low radiation dose for pelvis area of patients.

Tuesday, Nov 28 12:45 PM - 1:15 PM CST @ Learning Center

Value of Deep Learning Reconstruction of Chest Low-dose CT for Image Quality Improvement and Lung Parenchyma Assessment on Lung Window | W2-SPCH-8

Jinhua Wang, MD - Peking Union Medical College Hospital

Conclusions: LDCT-DLR can effectively reduce image noise and improve image quality. LDCT-DLR provides good performance for evaluating pulmonary lesions, except for subsolid nodules and decreased lung attenuation, compared to RDCT-HIR.

Clinical Relevance/Application: DLR enables LDCT maintaining image quality even with very low radiation doses

Wednesday, Nov 29 9:00 AM - 9:30 AM CST @ Learning Center

Impact of Deep Learning Reconstruction on Radiation Dose Reduction and Cancer Risk in CT Examinations: A Real-World Clinical Analysis | W5B-SPPH-10

Naoki Kobayashi, MD - Kumamoto University

Conclusions: Our single-center study suggests that the implementation of DLR has the possibility to reduce radiation dose and radiation-induced cancers by nearly half in the clinical routine CT compared with the iterative reconstruction.

Clinical Relevance/Application: This study shows the degree to which the introduction of DLR contributes to a reduction in radiation dose and thus in the incidence of radiation-induced cancer.

Wednesday, Nov 29 12:15 PM - 12:45 PM CST @ Learning Center

Deep-learning Reconstruction for Reducing Slice Thickness and Radiation Dose in Pediatric CT: Matched Pair Comparisons with Standard Dose Thick Slice Iterative Reconstruction | W5B-SPPD-1

Ryota Harai, MD – Kumamoto University

Conclusions: In pediatric CT, the application of DLR to low-dose thin-slice (0.5 mm) images can yield equivalent or even better subjective and objective quality compared to standard-dose thick-slice (3 mm) HIR images.

Clinical Relevance/Application: DLR allows reduction of both radiation dose and slice thickness in pediatric CT without increasing image noise and degrading the subjective image quality compared to the HIR algorithm.

Wednesday, Nov 29 12:45 PM - 1:15 PM CST @ Learning Center

Silver (Ag) X-ray Spectrum Modulation Filter vs. Copper (Cu) X-ray Spectrum Modulation Filter: Capability of Low-Dose CT for Lung Cancer Screening at In Vitro and In Vivo Studies | R2-SPPH-8

Yoshiharu Ohno, MD, PhD - Fujita Health University

Conclusions: Ag filter can significantly improve image quality and nodule detection capability than Cu filter on low-dose CT screening at in vitro and in vivo studies.

Clinical Relevance/Application: Ag filter can significantly improve image quality and nodule detection capability than Cu filter on low-dose CT screening at in vitro and in vivo studies.

Thursday, Nov 30 9:00 AM - 9:30 AM CST @ Learning Center

Evaluation and Comparison of CT Angiography Image Quality Acquired with Single-Energy Metal Artifact Reduction Algorithm and Deep Learning Reconstruction in Patients after Endovascular Aortic Repair | R2-SPVA-1

Zhiman Lai - The First Affilated Hospital, Sun Yat-sen University

Conclusions: Compared to HIR, DLR, and HIR+SEMAR, DLR+SEMAR provides superior image quality in terms of quantitative and qualitative parameters, not only markedly reducing metal artifacts but also improving reliably attenuation of soft tissues.

Clinical Relevance/Application: The combination of DLR and SEMAR algorithm technique significantly improving CTA image quality not only in the images with metal artifacts but also in those without metal artifacts in patients after EVAR, beneficial for postoperative surveillance.

Thursday, Nov 30 9:00 AM - 9:30 AM CST @ Learning Center

Brain Image Reconstruction using Deep Learning CT Image Quality | R5B-SPNR-8

Michaela Cellina, Asst Fatebenefratelli Sacco

Conclusion: In brain CT, deep learning based reconstructions allow significant reduction of noise and artifacts and better subjective image quality compared with iterative reconstructions.

Clinical Relevance: Deep learning based reconstructions improve image quality when compared to iterative reconstruction: this fact may allow the reduction of CT acquisition parameters and radiation exposure in CT protocols.

Thursday, Nov 30, 12:45 PM - 1:15 PM CST Learning Center

Deep Learning Reconstruction vs. Hybrid-Type Iterative Reconstruction: Capability for Image Quality Improvement and Lung Finding Evaluation on Standard-, Reduced- and Ultra-Low-Dose High-Definition CTs | S1-SSCH01-6

Yoshiharu Ohno, MD, PhD - Fujita Health University

Conclusions: DLR is potentially more effective for image quality improvement and lung finding evaluation than hybrid-type IR on all radiation dose CTs obtained at HDCT in patients with a variety of pulmonary diseases.

Clinical Relevance/Application: DLR is potentially more effective for image quality improvement and lung finding evaluation than hybrid-type IR on all radiation dose CTs obtained at HDCT in patients with a variety of pulmonary diseases.

Sunday, Nov 26 9:00 AM - 10:00 AM CST @ Rm S401

Ultra-high-resolution Drip-infusion CT Cholangiography with Deep Learning-based Reconstruction: Comparison with Normal-resolution Simulation | S3B-SPGI-4

Nakamoto Atsushi, MD, PhD – Osaka University

Conclusions: UHR CT combined with a deep learning-based reconstruction provided drip-infusion CT cholangiography images with higher spatial resolution and improved the delineation of the peripheral bile duct, despite increased image noise and decreased CNR.

Clinical Relevance/Application: Ultra-high-resolution CT combined with a deep learning-based reconstruction provides high-quality drip-infusion CT cholangiography and improves peripheral bile duct delineation.

Sunday, Nov 26 12:15 PM - 12:45 PM CST @ Learning Center

Phantom Study on Radiomic Features in Ultra-High-Resolution CT Imaging: Matrix Size, Radiation Dose, and Reconstruction Algorithms | M5A-SPPH-13

Tomoki Maebayashi, RT - Kobe University

Conclusions: Significant discrepancies in feature measurements can be discerned when utilizing UHR CT. Approximately 30% of feature measurements using UHR CT images were notably influenced by dose but not by the reconstruction algorithm.

Clinical Relevance/Application: Radiomic features derived from CT images using a 1024 matrix deviate from those originating from conventional CT, with measurements dependent on radiation dose while remaining largely unaffected by the reconstruction.

Monday, Nov 2712:15 PM - 12:45 PM CST @ Learning Center

High Spatial Resolution Data Improves Diagnostic Performance of Machine-Learning Radiomics Model: Prediction for Invasive Adenocarcinoma of the Lung | T5B-SPCH-3

Masahiro Yanagawa, MD, PhD - Osaka University

Conclusions: High spatial resolution significantly improved diagnostic performance of IVA by the machine-learning radiomics model. When used by radiologists, the present model tended to increase the accuracy and sensitivity of IVA diagnosis at the expense of specificity.

Clinical Relevance/Application: Machine-learning radiomics model trained by high spatial resolution data can greatly enhance diagnostic performance of invasive adenocarcinoma, providing support to radiologist, especially in improving accuracy and sensitivity.

Tuesday, Nov 28 12:45 PM - 1:15 PM CST@ Learning Center

Radiological-Histological Correlation on Ultra-High-Resolution CT Using Cadaveric Human Lungs: Nodule and Airway Analysis | T5B-SPCH-4

Akinori Hata, MD, PhD - Osaka University

Conclusions: UHR-CT images with high resolution reconstruction setting outperformed C-CT in the assessment of nodules and airways. Hybrid IR with 2048 matrix showed a better score than DLR with 1024 matrix for nodule assessment.

Clinical Relevance/Application: UHR-CT is suitable for the assessment of nodules and airways in detail. Hybrid IR with 2048 matrix may be the most advantageous for the assessment of the fine morphology of nodules.

Tuesday, Nov 28 12:45 PM - 1:15 PM CST @ Learning Center

High Resolution CT Imaging with a 1024 Matrix: Impact of Matrix Size, Slice Thickness, Reconstruction Algorithm, and Reslicing on Radiomic Feature Quantification in Hepatocellular Carcinoma | T5B-SPGI-1

Masatoshi Hori, MD, PhD – Kobe University

Conclusions: Slice thickness emerged as the most influential factor affecting the measurement values of features. Subsequently, when employing UHR images (1024 matrix) compared to conventional-resolution images (512 matrix), significant differences were observed in 15-16% of feature measurements in both arterial and portal venous phases. The effects of reslicing were discernible in approximately 10% of features, while the impact of the reconstruction algorithm was minimal.

Clinical Relevance/Application: Radiomic features procured from UHR CT employing a 1024 matrix diverge from those garnered through conventional CT. Slice thickness prevails as the primary determinant, with the 1024-matrix as the subsequent influencer.

Tuesday, Nov 28 12:45 PM - 1:15 PM CST @ Learning Center

Comparison of Ultra-High-Resolution and Normal-Resolution CT-Angiography for Intracranial Aneurysm Detection in Patients with Subarachnoid Hemorrhage | W2-SPNR-11

Marius Frenzel, MD – Mainz University

Conclusions: We found Ultra-High-Resolution CT-Angiography to provide higher sensitivity than Normal-Resolution CT-Angiography for the detection of intracranial aneurysms in patients with aneurysmal subarachnoid hemorrhage while improving image quality and reducing patient radiation exposure.

Clinical Relevance/Application: In addition to achieving the universally sought-after reduction of radiation dose, the increased detection of intracranial aneurysms in UHR-CT imaging may facilitate therapeutic decisions and hence improve patient outcome.

Wednesday, Nov 29 9:00 AM - 9:30 AM CST@ Learning Center

Improved Neurovascular Imaging using Advanced Intelligent Clear-IQ Engine (AiCE) | R2-SPNR-8

M. Abello Mercado, MD – Mainz University

Conclusions: Deep-Learning image reconstruction significantly improves image quality of ultra-high resolution neurovascular CT-angiography allowing for higher diagnostic confidence, potentially improving the detection of subtle but oftentimes-significant pathologies.

Clinical Relevance/Application: Deep-learning image reconstruction improves the quality of UHR-CTA images, leading to higher diagnostic confidence and potentially aiding in the detection of subtle but clinically significant pathologies.

Thursday, Nov 30 9:00 AM - 9:30 AM CST @ Learning Center

A Game-Changing Approach to Non-Invasive Vascular Imaging: Ultra High-Resolution CT with Novel Deep Learning Reconstruction | NREE-121

Yuta Shimizu, MD – Kyorin University

Teaching Points:

1. To learn the basic principles of ultra high-resolution CT (UHRCT).
2. To learn the basic principles of the novel reconstruction technology “deep learning reconstruction (DLR)”.
3. To know the differences in image quality between DLR and conventional
4. Clinical applications of brain CTA with DLR using UHRCT and comparison with digital subtraction angiography (DSA).
5. To know the pitfalls of CTA with DLR and DSA on cerebral diseases.
6. To present the outlook of brain CTA with DLR using UHRCT in the near future.

Neuroradiology Education Exhibits

Analysis of Fine Anatomical Structures Involved in Pulmonary Nodules by Ultra-High-Resolution CT | CHEE-104

Hiroshi Moriya, MD, PhD – Ohara General Hospital

Teaching Points:

1. Spatial resolution of ultra-high-resolution CT (UHRCT)
2. Depiction of intralobular structure by UHRCT
3. MPR display of bronchus long-axis cross-section and short-axis cross-section (parallel pleural cross-section)
4. Typical cases: Analysis of fine anatomical structures involved in pulmonary nodules by UHRCT.

Chest Imaging Education Exhibits

Analysis of Solitary Lung Nodules Using Summation Images Created from Ultra-High-Resolution CT Data | CHEE-91

Hiroshi Moriya, MD, PhD – Ohara General Hospital

Teaching Points:

1. Advantages of ultrahigh-resolution CT: Ultra-high resolution CT has excellent spatial resolution in lung field analysis.
2. 3D image analysis method using ultra-high resolution CT data: Coronal MPR and VR image are useful for chest image analysis. Summation image (SUM) can observe a X-ray projection image of the chest from any viewpoint.
3. There are two factors that make lung lesions difficult to detect: (1) lesions with little density difference from the surrounding lung, and (2) presence in areas where other structures are projected in the anterior-posterior direction.
4. Points to note when interpreting chest radiographs in routine clinical practice.

Chest Imaging Education Exhibits

Distinguishing Common Renal Cell Carcinomas from Benign Renal Tumors Based on Machine Learning: Comparing Various CT Imaging Phases, Slices, Tumor Sizes and ROI Segmentation Strategies | S1-SSGU01-1

Tao Zhou – Jinan People's Hospital

Conclusions: HU-augmentation improved DL algorism to detect rib metastases on CT.

Clinical Relevance/Application: This study reveals that HU-augmentation using the appropriate CT values for the three types of bone metastases (osteoblastic, osteolytic, and mixed) improves the detection accuracy of rib metastases.

Sunday, Nov 26 9:00 AM - 10:00 AM CST @ Rm S404

Detectability of Noise Texture Changes in CT | M3-SSPH03-1

Luuk J Oostveen, PhD - University Medical Center, Nijmegen

Conclusions: Fpeak and Fav alone are insufficient to describe noise texture since textures with the same Fpeak or Fav can be discriminable. Discrimination of noise texture changes depends on its frequency content. Radiologists do not discriminate noise texture changes better than laymen.

Clinical Relevance/Application: For the development of non-linear reconstruction techniques and reconstruction kernels it is important to understand how human observers react to various noise textures.

Monday, Nov 27 9:30 AM - 10:30 AM CST @ Rm S405

Non-invasive Liver Fibrosis Assessment with CT-based Iodine-uptake Parameters and Hepatosplenic Volumetric Indices | M5A-SPGI-8

Kenichiro Yoshida, MD – Kumamoto University

Conclusions: IWR and SV/BSA can allow better prediction of liver fibrosis severity compared with ECV and LV/BSA in multiphase hepatic CT.

Clinical Relevance/Application: IWR and SV/BSA may be served as a non-invasive imaging biomarker for predicting liver fibrosis severity on routine multiphase CT.

Monday, Nov 27 12:15 PM - 12:45 PM CST @ Learning Center

A New Approach to Detect Rib Metastases Using Deep Learning with Hounsfield Unit-augmentation | R5A-SPMK-2

Misato Sone, MD – Iwate Medical University

Conclusions: HU-augmentation improved DL algorism to detect rib metastases on CT.

Clinical Relevance/Application: This study reveals that HU-augmentation using the appropriate CT values for the three types of bone metastases (osteoblastic, osteolytic, and mixed) improves the detection accuracy of rib metastases.

Monday, Nov 27 12:15 PM - 12:45 PM CST @ Learning Center

Contrast-enhancement Boost Technique on Abdominal-enhanced CT Improves Image Quality of Adamkiewicz Artery | M5B-SPPH-4

Huiqing Zhou – Jinan People's Hospital

Conclusions: CE-Boost technique can improve the image quality of adamkiewicz artery.

Clinical Relevance/Application: CE-Boost technology can improve the visualization of adamkiewicz artery, which is beneficial for understanding spinal cord blood supply and improving the therapeutic effect of diseases of the thoracic aorta and abdominal aorta.

Monday, Nov 27 12:45 PM - 1:15 PM CST @ Learning Center

Improvement of Image Quality of Portal Vein with Contrast Enhancement Boost Technique in Low Dose and Low Contrast Media Usage Abdominal-enhanced Computed Tomography | M5B-SPPH-9

Dandan Niu – Jinan People's Hospital

Conclusions: CE-Boost technique can significantly improve the image quality of portal vein imaging in low dose and low contrast media usage abdominal-enhanced CT.

Clinical Relevance/Application: Portal vein imaging has poor image quality in patients with blood circulation disorders. If these patients are complicated with portal thrombosis and cancer embolus, CE-Boost technology can improve image quality and reduce the rate of misdiagnosis of portal thrombosis and cancer embolus.

Monday, Nov 27 12:45 PM - 1:15 PM CST @ Learning Center

Imaging Features in Female Patients with Wild-type Transthyretin Amyloidosis Cardiomyopathy | M5A-SPCA-7

Seitaro Oda, MD – Kumamoto University

Conclusions: Female patients with ATTRwt-CM were predominantly older and had preserved left ventricular ejection fraction, weaker cardiac uptake of the 99mTc-PYP, and lower ECV value compared with male patients. These imaging features may contribute to the underdiagnosis of ATTRwt-CM in female patients.

Clinical Relevance/Application: Female patients with ATTRwt-CM have different imaging features than the male counterparts, and we need to recognize these imaging features to avoid underdiagnosing it.

Monday, Nov 27 12:45 PM - 1:15 PM CST @ Learning Center

Assessing the Efficacy of Multiple Additive Processing in Contrast Enhancement Boost CT Technique for the Diagnosis of Hypervascular Hepatocellular Carcinoma | M5B-SPPH-11

Shinji Yabe, MD, PhD - Kobe University

Conclusions: The CE-boost technique with multiple additive processing can increase the conspicuity of arterial phase hyperenhancement of hypervascular HCC by improving the tumor-to-liver CNR. In contrast, improvements in tumor-to-liver CNR were not observed in the PVP for washout assessment of hypervascular HCC.

Clinical Relevance/Application: CE-boost with multiple additive processing can improve the degree of contrast effect of CE-CT effectively. This technique is especially beneficial in the detection of focal liver lesions that show arterial phase hyperenhancement.

Monday, Nov 27 12:45 PM - 1:15 PM CST @ Learning Center

Four-Dimensional Esophageal Computed Tomography Imaging: Assessment of Treatment Effect for Esophagogastric Junction Outflow Disorders after Per-oral Endoscopic Myotomy | T5A-SPGI-6

Hikaru Nishiyama – Ehime University

Conclusions: 4D-ECT allowed for the quantitative assessment of improving both impaired relaxation of LES and esophageal clearance after POEM.

Clinical Relevance/Application: 4D-ECT is a novel imaging technique for the assessment of esophageal dynamics, and provides quantitative parameters for evaluating esophageal morphology, motility, and clearance. 4D-ECT enables quantitative and objective assessment of treatment effect for EGJO disorders after POEM.

Tuesday, Nov 28 12:15 PM - 12:45 PM CST @ Learning Center

Investigation of Buffer Thickness for Reducing Artifacts from the Table in Computed Tomography Examinations During the Extremities Three-Dimensional Imaging | S3B-SPMK-8

Yoshiki Kamiho Riuchi

Conclusions: By inserting a buffer material with a thickness of at least 1mm between the hand and the CT table mat at creating 3D imaging, the separation of the hand and the CT table mat can be achieved easier for regardless of the conditions.

Clinical Relevance: It was confirmed that using a buffer material of 1 mm or more can significantly reduce the effects of artifacts from the CT table mat, which can improve the quality of 3D imaging of the hand and increase flexibility in positioning.

Tuesday, Nov 28, 1:30 PM - 2:30 PM CST E450B

The Utilization of Double Contrast-Enhancement Ce-Boost for the Lower Extremity Vessel on CT Angiography | W1-SSVA03-2

Chuluunbaatar Otgonbaatar, MD – Seoul National University

Conclusions: Implementation of double CE-boost technique improves the image quality by lower image noise, higher CT attenuation, SNR, CNR, and sharpness compared with CE-boost and conventional image.

Clinical Relevance/Application: CE-boost technique and double CE-boost technique improve the vascular contrast attenuation, SNR, CNR, subjective analysis and reduces the image noise in the lower extremity vessels especially below the knee arteries without increasing the injection rate and concentration of contrast media.

Wednesday, Nov 29 8:00 AM - 9:00 AM CST @ N226

Improvement of Diagnosis of Adjacent Organ Invasion of Extrahepatic Bile Duct Cancer by Adding Arterial and Delayed Phases | W3-SSGI12-6

Eisuke Mukaida, MD – Iwate Medical University

Conclusions: Addition of the arterial and delayed phases increased the AUC of pancreatic, duodenal, and arterial invasion.

Clinical Relevance/Application: Histological evaluation of patients with EHD cancer often reveals invasion of adjacent organs, which may lead to curative resection failure or early postoperative recurrence. Because the adjacent organs of EHD cancer vary according to the site of origin, the diagnostic performance of adjacent organ invasion may be influenced by contrast effects specific to that organ; however, no reports have been retrieved on this effect. Our results provide an important preoperative imaging strategy to achieve R0 resection of EHD cancer.

Wednesday, Nov 29 9:30 AM - 10:30 AM CST @ Rm N228

Patient-specific Analysis of Organ Doses and Image Quality in Abdominal Single Energy and Dual Energy CT Examinations | R5A-SPPH-10

Keisuke Fujii, PhD - NCCE

Conclusions: DECT examinations allows for the reduction of organ doses for organs within scan range by approximately 30% compared to conventional SECT examinations, and VMI images obtained from DECT scans also provides higher signal detectability than SECT image.

Clinical Relevance/Application: DECT examinations can be performed with lower radiation doses than conventional SECT examinations, and VMI images obtained from DECT scans can provide higher signal detectability than SECT images.

Thursday, Nov 30 12:15 PM - 12:45 PM CST @ Learning Center

Measurement of Spleen Volume Using CT Images with Deep Learning Segmentation: Definition of Normal Spleen and Splenomegaly | R5B-SPGI-1

Kotarou Fujita, MD – Iwate Medical University

Conclusions: Weight-based thresholds for objective normal spleen and splenomegaly were derived from a spleen volume database using CT images, based on deep learning.

Clinical Relevance/Application: Splenomegaly is defined as an abnormal enlargement of the spleen resulting from various clinical conditions such as infections, hematologic diseases, congestion, autoimmune disease or tumors. Splenomegaly is often a diagnostic entry point for various background diseases, and accurate detection of splenomegaly is clinically important. The purpose of our study is to establish the normal distribution of splenic volume in healthy adults using automated organ segmentation techniques on abdominal CT images. Furthermore, this data will be used to establish a potential threshold for splenomegaly.

Thursday, Nov 30 12:45 PM - 1:15 PM CST @ Learning Center

Effects of Model-based Iterative Reconstruction Algorithm and Contrast Enhancement Boost Postprocessing Technique on the Pulmonary CT Angiography: A Evaluation of Image Quality for Obese Patients | R5A-SPCH-5

Mei Ye – The First Affiliated Hospital of Xinjiang Medical University

Conclusions: The postprocessing technique CE-Boost and FIRST reconstruction algorithm can both improve overall image quality and diagnostic confidence of pulmonary CTA in obese patients.

Clinical Relevance/Application: High-quality pulmonary CTA image provided by CE-Boost or FIRST enables a better visualization of the main pulmonary artery and its branch vessels. This may enable early diagnosis of PE and reduce exposure dose in obese patients.

Thursday, Nov 30 12:15 PM - 12:45 PM CST @ Learning Center

The Application Value of Black Blood Imaging Based on CT Flexible Subtraction Technology in Carotid Artery | R5A-SPVA-1

Tao Zhou – Jinan People's Hospital

Conclusions: HU-augmentation improved DL algorism to detect rib metastases on CT.

Clinical Relevance/Application: This study reveals that HU-augmentation using the appropriate CT values for the three types of bone metastases (osteoblastic, osteolytic, and mixed) improves the detection accuracy of rib metastases.

Thursday, Nov 30 12:15 PM - 12:45 PM CST @ Learning Center

Purchasing Decisions for an Interventional CT Program | IREE-6

Jonathan Troville, PhD - UW Madison

Teaching Points:

1. Interventional CT fluoroscopy is becoming widely used and scanner features that meet both diagnostic and interventional needs are becoming increasingly complex. In this paper, the most important scanner features are discussed and ranked for both diagnostic and interventional purposes.
2. Vendors provide different offerings for the features, so hypothetical comparisons for three vendors are included to facilitate explaining various tradeoffs in purchasing decisions.

Interventional Radiology Education Exhibits | IREE

The Emerging Role of Cardiac CT in Clinical Practice of Cardiac Amyloidosis | CAEE-85

Seitaro Oda, MD - Kumamoto University

Teaching Points:

1. Recent research has revealed a considerable number of hidden cardiac amyloidosis (Ca) especially in patients with heart failure with left ventricular hypertrophy. Ca is, however, frequently misdiagnosed as the other hypertrophic heart disease, and delayed diagnosis results in significant consequences for patients.
2. Current noninvasive diagnostic imaging strategies for Ca include echocardiography, cardiac MRI, and bone scintigraphy.
3. Myocardial late iodine enhancement (LIE) and extracellular volume (ECV) quantification by cardiac CT has become clinically available and clinical evidence is accumulating.
4. Since CT is more practical than MRI and scintigraphy, it plays a clinically important role in the diagnosis and management of Ca.

Cardiac Imaging Education Exhibits CAEE

Emerging Applications of Non-Contrast Cardiac CT: Beyond Coronary Score | CAEE-41

Masafumi Kidoh, MD - Kumamoto University

Teaching Points:

1. Non-contrast cardiac CT is widely used to detect and quantify coronary and heart valve calcium. On the other hand, detection of fat, myocardial edema, thrombus/hematoma, and subtle myocardial calcification beyond coronary and heart valve calcification using non-contrast cardiac CT has not been widely performed, possibly due to poor image quality.
2. Recent advances in CT technology (such as model-based iterative reconstruction and dual-layer spectral/photon-counting CT) have drastically improved the low-contrast resolution of CT images, allowing the detection of small differences in CT values within soft tissue on non-contrast and delayed phase cardiac CT images. Side-by-side viewing of non-contrast cardiac CT images and delayed (or arterial) phase cardiac CT images may be useful in differentiating various pathologies.
3. In this exhibit, we report that non-contrast cardiac CT is useful for the detection and characterization of cardiac low-contrast lesions (soft tissue lesions) beyond coronary and heart valve calcification.

Cardiac Imaging Education Exhibits - CAEE

Motion Artifact Correction using a New Deep Learning Reconstruction (MC-DLR) in the Chest Computed Tomography | T5B-SPCH-6

Yoichiro Ota, RT / Tae Iwasawa, MD - Kanagawa Cardiovascuiar and Respiratory Center

Conclusions: MC-DLR, a new motion correction method, decreases motion artifacts in chest CT images.

Clinical Relevance/Application: Obtaining high-quality CT images is often challenging in emergency cases. MC-DLR effectively reduces motion artifacts in post-processed images, improving the image quality.

Tuesday, Nov 28 12:45 PM - 1:15 PM CST @ Learning Center

Evaluation of Male Voiding Function using Upright CT in the Standing Posture: A Pilot Study | T6-SSGU04-2

Yoichi Yokoyama, MD Keio University

Conclusions: The 4D CT imaging technique in a standing position has the potential to illuminate the standard urination mechanism in males. Identifying variations from this norm in dysuria patients can guide toward more appropriate treatment strategies for such conditions.

Clinical Relevance/Application: Upright 4D CT imaging could improve LUTD diagnostics, offering a non-invasive method to understand urination mechanisms and guide appropriate treatment for conditions like BPH.

Tuesday, Nov 28 1:30 PM - 2:30 PM CST @ N227B

Fourier Diffusion Models for CT Denoising and Restoration for Known Image Properties | W7-SSPH11-4

Web Stayman, PhD, JHU

Conclusions: Fourier diffusion models allows for explicit modeling of the image properties in CT data. Integration of this information into the model permits both improved restoration of image data, but also a smaller number of time steps in the iterative update equation. This addresses a commonly cited disadvantage of diffusion models - namely, the increased computation time required by fine discretization of the reverse stochastic process.

Clinical Relevance/Application: Statistically valid uncertainty quantification techniques build radiologists’ confidence in diffusion models used for inverse problems in medical imaging.

Wednesday, Nov 29 3:00 PM - 4:00 PM CST @ Rm 227B

K-RCPS: Uncertainty Quantification for Diffusion Models via Conformal Prediction and Conformal Risk Control in CT Denoising | R5B-SPPH-5

Jacopo Teneggi, PhD, JHU

Conclusions: Conformal prediction and conformal risk control can be deployed for diffusion models in CT denoising to construct uncertainty intervals that provide finite-sample and distribution-free guarantees on the generated noiseless images. K-RCPS provably minimizes the mean interval length, and it consistently outperforms existing methods.

Clinical Relevance/Application: Statistically valid uncertainty quantification techniques build radiologists’ confidence in diffusion models used for inverse problems in medical imaging.

Thursday, Nov 30 12:45 PM - 1:15 PM CST

Current and Novel Techniques for Artifact Reduction on Chest CT | CHEE-117

Yoichiro Ota, RT / Tae Iwasawa, MD - Kanagawa Cardiovascuiar and Respiratory Center

Teaching Points: Various artifacts can be observed on chest CT. Motion artifacts are an important issue, especially in emergency cases in which patients are unable to hold their breath. Motion correction-deep learning reconstruction (MC-DLR) is a motion correction method developed using a deep learning framework to estimate patient motion in the lung region. This method comprises three major steps. First, a full-image reconstruction is decomposed into several partial-angle reconstructions. Second, a pre-trained deep convolutional neural network (DCNN) is applied to estimate the motion vector field between these partial angle reconstruction images. Finally, a motion-corrected image is reconstructed by incorporating the DCNN-predicted motion vector field.

Chest Imaging Education Exhibits

Upright Chest CT Imaging | CHEE-103

Yoshitake Yamada, MD, Keio University

Teaching Points: The purposes of our exhibit are as follows:

1. To explain the development, background, and performance of upright CT.
2. To describe the anatomical changes in each structure of the chest due to differences in body position, that is, the direction of gravity.
3. To illustrate the clinical applications of upright CT for various chest diseases with clinical case presentations, and the potential applications in particle therapy.

Chest Imaging Education Exhibits