Search

Background Spatial resolution, soft-tissue contrast, and dose-efficient capabilities of photon-counting CT (PCCT) potentially allow a better quality and diagnostic confidence of coronary CT angiography (CCTA) in comparison to conventional CT. Purpose To compare the quality of CCTA scans obtained with a clinical prototype PCCT system and an energy-integrating detector (EID) dual-layer CT (DLCT) system. Materials and Methods In this prospective board-approved study with informed consent, participants with coronary artery disease underwent retrospective electrocardiographically gated CCTA with both systems after injection of 65-75 mL of 400 mg/mL iodinated contrast agent at 5 mL/sec. A prior phantom task-based quality assessment of the detectability index of coronary lesions was performed. Ultra-high-resolution parameters were used for PCCT (1024 matrix, 0.25-mm section thickness) and EID DLCT (512 matrix, 0.67-mm section thickness). Three cardiac radiologists independently performed a blinded analysis using a five-point quality score (1 = insufficient, 5 = excellent) for overall image quality, diagnostic confidence, and diagnostic quality of calcifications, stents, and noncalcified plaques. A logistic regression model, adjusted for radiologists, was used to evaluate the proportion of improvement in scores with the best method. Results Fourteen consecutive participants (12 men; mean age, 61 years ± 17) were enrolled. Scores of overall quality and diagnostic confidence were higher with PCCT images with a median of 5 (interquartile range [IQR], 2) and 5 (IQR, 1) versus 4 (IQR, 1) and 4 (IQR, 3) with EID DLCT images, using a mean tube current of 255 mAs ± 0 versus 349 mAs ± 111 for EID DLCT images (

Resume Introduction Dans le cadre de la reforme du deuxieme cycle des etudes medicales en France, l’accent est particulierement mis sur l’evaluation des competences des etudiants en utilisant un outil standardise dit « ECOS » pour examen clinique objectif et structure. Donnees recentes Cet outil, largement utilise dans de nombreux pays comme le Canada, la Suisse et l’Inde, se compose d’une succession d’ateliers de simulation d’une competence de courte duree permettant une evaluation standardisee, contrairement aux methodes d’evaluation dites traditionnelles. Pour ce qui concerne le deuxieme cycle des etudes medicales en France, l’etudiant, pendant sa formation approfondie en sciences medicales devra acquerir des connaissances de radio-anatomie, devra connaitre les principales indications des examens d’imagerie et enfin avoir des connaissances sur la semeiologie en imagerie des principales urgences et pathologies qu’un medecin de premier recours pourra rencontrer. Conclusion L’ECOS va permettre ainsi d’evaluer l’ensemble des competences requises a la pratique de l’imagerie medicale pour un etudiant de deuxieme cycle d’etudes.

Objectives: To evaluate the image quality (IQ) of a spectral photon-counting CT (SPCCT) using filtered back projection (FBP) and hybrid iterative reconstruction (IR) algorithms (iDose4), in comparison with a dual-layer CT (DLCT) system, and to choose the best image quality according to the IR level for SPCCT. Methods: Two phantoms were scanned using a standard lung protocol (120 kVp, 40 mAs) with SPCCT and DLCT systems. Raw data were reconstructed using FBP and 9 iDose4 levels (i1/i2/i3/i4/i5/i6/i7/i9/i11) for SPCCT and 7 for DLCT (i1/i2/i3/i4/i5/i6/i7). Noise power spectrum and task-based transfer function (TTF) were computed. Detectability index (d′) was computed for detection of 4 mm ground-glass nodule (GGN) and solid nodule. Two chest radiologists performed an IQ evaluation (noise/nodule sharpness/nodule conspicuity/overall IQ) in consensus, and chose the best image for SPCCT. Results: Noise magnitude was −47% ± 2% lower on average with SPCCT than with DLCT for iDose4 range from i1 to i6. Average NPS spatial frequencies increased for SPCCT in comparison with DLCT. TTF also increased, except for the air insert with FBP, and i1/i2/i3. Higher detectability was found for SPCCT for both GGN and solid nodules. IQ for both types of nodule was rated consistently higher with SPCCT than with DLCT for the same iDose4 level. For SPCCT and both nodules, the scores for noise and conspicuity improved with increasing iDose4 level. iDose4 level 6 provided the best subjective IQ for both types of nodule. Conclusions: Higher IQ for GGN and solid nodules was demonstrated with SPCCT compared with DLCT with better detectability using iDose4. Key Points: Using spectral photon-counting CT compared with dual-layer CT, noise magnitude was reduced with improvements in spatial resolution and detectability of ground-glass nodules and solid lung nodules.As the iDose4level increased, noise magnitude was reduced and detectability of ground-glass and solid lung nodules was better for both CT systems.For spectral photon-counting CT imaging, two chest radiologists determined iDose4level 6 as the best image quality for detecting ground-glass nodules and solid lung nodules.

Objectives To evaluate the quality of virtual monochromatic images (VMIs) from spectral photon-counting CT (SPCCT) and two energy-integrating detector dual-energy CT (EID-DECT) scanners from the same manufacturer, for the coronary lumen. Methods A 21-cm section of the Mercury v4.0 phantom was scanned using a cardiac CT protocol. VMIs from 40 to 90 keV were reconstructed using high-resolution (HR) parameters for EID-DECT and SPCCT (CB and HRB kernels at 0.67 mm slice thickness, respectively). Ultra-high-resolution (UHR) parameters were used in addition to SPCCT (detailed-2 kernel, 0.43 mm slice thickness). Noise-power-spectrum (NPS), task-based transfer function (TTF), and detectability index (d′) were computed for 2-mm-diameter lumen detection. In consensus, two radiologists analyzed the quality of the images from 8 patients who underwent coronary CTA on both CT systems. Results For all keV images, fpeak, f50, and d′ were higher with SPCCT. The fpeak and f50 were higher with UHR-SPCCT with greater noise and lower d′ compared to those of the HR-SPCCT images. Noise magnitude was constant for all energy levels (keV) with both systems, and lower with HR images, and d′ decreased as keV decreased. Subjective analysis showed greater lumen sharpness and overall quality for HR and UHR-SPCCT images using all keV, with a greater difference at low keV compared to HR-EID-DECT images. Conclusion HR and UHR-SPCCT images gave greater detectability of the coronary lumen for 40 to 90 keV VMIs compared to two EID-DECT systems, with benefits of higher lumen sharpness and overall quality. Key Points • Compared with 2 dual-energy CT systems, spectral photon-counting CT (SPCCT) improved spatial resolution, noise texture, noise magnitude, and detectability of the coronary lumen. • Use of ultra-high-resolution parameters with SPCCT improved spatial resolution and noise texture and provided high detectability of the coronary lumen, despite an increase in noise magnitude. • In eight patients, radiologists found greater overall image quality with SPCCT for all virtual monochromatic images with a greater difference at low keV, compared with dual-energy CT systems.

Recently, machine learning has outperformed established tools for automated segmentation in medical imaging. However, segmentation of cardiac chambers still proves challenging due to the variety of contrast agent injection protocols used in clinical practice, inducing disparities of contrast between cavities. Hence, training a generalist network requires large training datasets representative of these protocols. Furthermore, segmentation on unenhanced CT scans is further hindered by the challenge of obtaining ground truths from these images. Newly available spectral CT scanners allow innovative image reconstructions such as virtual non-contrast (VNC) imaging, mimicking non-contrasted conventional CT studies from a contrasted scan. Recent publications have demonstrated that networks can be trained using VNC to segment contrasted and unenhanced conventional CT scans to reduce annotated data requirements and the need for annotations on unenhanced scans. We propose an extensive evaluation of this statement. We undertake multiple trainings of a 3D multi-label heart segmentation network with (HU-VNC) and without (HUonly) VNC as augmentation, using decreasing training dataset sizes (114, 76, 57, 38, 29, 19 patients). At each step, both networks are tested on a multi-vendor, multi-centric dataset of 122 patients, including different protocols: pulmonary embolism (PE), chest-abdomen-pelvis (CAP), heart CT angiography (CTA) and true non-contrast scans (TNC). An in-depth comparison of resulting Dice coefficients and distance metrics is performed for the networks trained on the largest dataset. HU-VNC-trained on 57 patients significantly outperforms HUonly trained on 114 regarding CAP and TNC scans (mean Dice coefficients of 0.881/0.835 and 0.882/0.416, respectively). When trained on the largest dataset, significant improvements in all labels are noted for TNC and CAP scans (mean Dice coefficient of 0.882/0.416 and 0.891/0.835, respectively). Adding VNC images as training augmentation allows the network to perform on unenhanced scans and improves segmentations on other imaging protocols, while using a reduced training dataset.

Objectives To quantitatively evaluate the impact of virtual monochromatic images (VMI) on reduced-iodine-dose dual-energy coronary computed tomography angiography (CCTA) in terms of coronary lumen segmentation in vitro, and secondly to assess the image quality in vivo, compared with conventional CT obtained with regular iodine dose. Materials and methods A phantom simulating regular and reduced iodine injection was used to determine the accuracy and precision of lumen area segmentation for various VMI energy levels. We retrospectively included 203 patients from December 2017 to August 2018 (mean age, 51.7 ± 16.8 years) who underwent CCTA using either standard (group A, n = 103) or reduced (group B, n = 100) iodine doses. Conventional images (group A) were qualitatively and quantitatively compared with 55-keV VMI (group B). We recorded the location of venous catheters. Results In vitro, VMI outperformed conventional CT, with a segmentation accuracy of 0.998 vs. 1.684 mm2, respectively (p < 0.001), and a precision of 0.982 vs. 1.229 mm2, respectively (p < 0.001), in simulated overweight adult subjects. In vivo, the rate of diagnostic CCTA in groups A and B was 88.4% (n = 91/103) vs. 89% (n = 89/100), respectively, and noninferiority of protocol B was inferred. Contrast-to-noise ratios (CNR) of lumen versus fat and muscle were higher in group B (p < 0.001) and comparable for lumen versus calcium (p = 0.423). Venous catheters were more often placed on the forearm or hand in group B (p < 0.001). Conclusion In vitro, low-keV VMI improve vessel area segmentation. In vivo, low-keV VMI allows for a 40% iodine dose and injection rate reduction while maintaining diagnostic image quality and improves the CNR between lumen versus fat and muscle. Key Points • Dual-energy coronary CT angiography is becoming increasingly available and might help improve patient management. • Compared with regular-iodine-dose coronary CT angiography, reduced-iodine-dose dual-energy CT with low-keV monochromatic image reconstructions performed better in phantom-based vessel cross-sectional segmentation and proved to be noninferior in vivo. • Patients receiving reduced-iodine-dose dual-energy coronary CT angiography often had the venous catheter placed on the forearm or wrist without compromising image quality.

Aims In low-gradient aortic stenosis (LGAS), the high valvulo-arterial impedance observed despite low valvular gradient suggests a high vascular load. Thoracic aortic calcifications (TACs) and valvular aortic calcifications (VACs) are, respectively, surrogates of aortic load and aortic valvular gradient. The aim of this study was to compare the respective contributions of TAC and VAC on 3-year cardiovascular (CV) mortality following TAVI in LGAS vs. high-gradient aortic stenosis (HGAS) patients. Methods and results A total of 1396 consecutive patients were included. TAC and VAC were measured on the pre-TAVI CT-scan. About 435 (31.2%) patients had LGAS and 961 (68.8%) HGAS. LGAS patients were more prone to have diabetes, coronary artery disease (CAD), atrial fibrillation (AF), and lower left ventricular ejection fraction (LVEF), P Conclusion TAC plays a major prognostic role in LGAS while VAC remains the key in HGAS patients. This confirms that LGAS is a complex vascular and valvular disease.

Computed tomography imaging plays a major role in the preoperative assessment of tumor burden by providing an accurate mapping of the distribution of peritoneal metastases (PM). Spectral Photon Counting Computed Tomography (SPCCT) is an innovative imaging modality that could overcome the current limitations of conventional CT, offering not only better spatial resolution but also better contrast resolution by allowing the discrimination of multiple contrast agents. Based on this capability, we tested the feasibility of SPCCT in the detection of PM at different time of tumor growth in 16 rats inoculated with CC531 cells using dual-contrast injection protocols in two compartments (i.e. intravenous iodine and intraperitoneal gadolinium or the reverse protocol), compared to surgery. For all peritoneal regions and for both protocols, sensitivity was 69%, specificity was 100% and accuracy was 80%, and the correlation with surgical exploration was strong (p = 0.97; p = 0.0001). No significant difference was found in terms of diagnostic performance, quality of peritoneal opacification or diagnostic quality between the 2 injection protocols. We also showed poor vascularization of peritoneal metastases by measuring low concentrations of contrast agent in the largest lesions using SPCCT, which was confirmed by immunohistochemical analyses. In conclusion, SPCCT using dual-contrast agent injection protocols in 2 compartments is a promising imaging modality to assess the extent of PM in a rat model.

To establish national reference levels (RLs) in interventional procedures under CT guidance as required by the 2013/59/Euratom European Directive. Seventeen categories of interventional procedures in thoracic, abdominopelvic, and osteoarticular specialties (percutaneous infiltration, vertebroplasty, biopsy, drainage, tumor destruction) were analyzed. Total dose length product (DLP), number of helical acquisitions (NH), and total DLP for helical, sequential, or fluoroscopic acquisitions were recorded for 10 to 20 patients per procedure at each center. RLs were calculated as the 3rd quartiles of the distributions and target values for optimization process (TVOs) as the median. RLs and TVOs were compared with previously published studies. Results on 5001 procedures from 49 centers confirmed the great variability in patient dose for the same category of procedures. RLs were proposed for the DLPs and NHs in the seventeen categories. RLs in terms of DLP and NH were 375 mGy.cm and 2 NH for spinal or peri-spinal infiltration, 1630 mGy.cm and 3 NH for vertebroplasty, 845 mGy.cm and 4 NH for biopsy, 1950 mGy.cm and 8 NH for destruction of tumors, and 1090 mGy.cm and 5 NH for drainage. DLP and NH increased with the complexity of procedures. This study was the first nationwide multicentric survey to propose RLs for interventional procedures under CT guidance. Heterogeneity of practice in centers were found with different levels of patient doses for the same procedure. The proposed RLs will allow imaging departments to benchmark their practice with others and optimize their protocols. • National reference levels are proposed for 17 categories of interventional procedures under CT guidance. • Reference levels are useful for benchmarking practices and optimizing protocols. • Reference levels are proposed for dose length product and the number of helical acquisitions.

To compare the quantitative and qualitative lung perfusion data acquired with dual energy CT (DECT) to that acquired with a large field-of-view cadmium-zinc-telluride camera single-photon emission CT coupled to a CT system (SPECT-CT).A total of 53 patients who underwent both dual-layer DECT angiography and perfusion SPECT-CT for pulmonary hypertension or pre-operative lobar resection surgery were retrospectively included. There were 30 men and 23 women with a mean age of 65.4±17.5 (SD)years (range: 18-88years). Relative lobar perfusion was calculated by dividing the amount (of radiotracer or iodinated contrast agent) per lobe by the total amount in both lungs. Linear regression, Bland-Altman analysis, and Pearson's correlation coefficient were also calculated. Kappa test was used to test agreements in morphology and severity of perfusion defects assessed on SPECT-CT and on DECT iodine maps with a one-month interval. Wilcoxon rank sum test was used to compare the sharpness of perfusion defects and radiation dose among modalities.Strong correlations for relative lobar perfusion using linear regression analysis and Pearson's correlation coefficient (r=0.93) were found. Bland-Altman analysis revealed a -0.10 bias, with limits of agreement between [-6.01; 5.81]. With respect to SPECT- CT as standard of reference, the sensitivity, specificity, PPV, NPV, accuracy for lobar perfusion defects were 89.4% (95% CI: 82.6-93.4%), 96.5% (95% CI: 92.1-98.5%), 95.6% (95% CI: 90.9-97.8%), 91.4% (95% CI: 85.6-94.9%) and 93.0% (95% CI: 87.6-96.1%) respectively. High level of agreement was found for morphology and severity of perfusion defects between modalities (Kappa=0.84 and 0.86 respectively) and on DECT images among readers (Kappa=0.94 and 0.89 respectively). A significantly sharper delineation of perfusion defects was found on DECT images (P0.0001) using a significantly lower equivalent dose of 4.1±2.3 (SD) mSv (range: 1.9-11.85mSv) compared to an equivalent dose of 5.3±1.1 (SD) mSv (range: 2.8-7.3mSv) for SPECT-CT, corresponding to a 21.2% dose reduction (P=0.0004).DECT imaging shows strong quantitative correlations and qualitative agreements with SPECT-CT for the evaluation of lung perfusion.

During the first wave of the COVID-19 pandemic, the French Society of Radiology and the French College of Radiology, in partnership with NEHS Digital, have set up a system to collect chest computed tomography (CT) examinations with clinical, virological and radiological metadata, from patients clinically suspected of COVID-19 pneumonia. This allowed the constitution of an anonymized multicenter database, named FIDAC (French Imaging Database Against Coronavirus). The aim of this report was to describe the content of this public database.Twenty-two French radiology centers participated to the data collection. The data collected were chest CT examinations in DICOM format associated with the following metadata: patient age and sex, originating facility identifier, originating facility region, time from symptom onset to CT examination, indication for CT examination, reverse transcription-polymerase chain reaction (RT-PCR) results and normalized CT report performed by a senior radiologist. All the data were anonymized and sent through a NEHS Digital system to a centralized data center.A total of 5944 patients were included from the 22 centers aggregated into 8 regions with a mean number of patients of 743 ± 603.3 [SD] per region (range: 102-1577 patients). Reasons for CT examination and normalized CT reports were provided for all patients. RT-PCR results were provided in 5574 patients (93.77%) with a positive result of RT-PCR in 44.6% of patients.The FIDAC project allowed the creation of a large database of chest CT images and metadata available, under conditions, in open access through the CERF-SFR website.

Background There is a paucity of data on the epidemiology, survival estimates and healthcare resource utilisation and associated costs of patients with progressive fibrosing interstitial lung disease (PF-ILD) in France. An algorithm for extracting claims data was developed to indirectly identify and describe patients with PF-ILD in the French national administrative healthcare database. Methods The French healthcare database, the Système National des Données de Santé (SNDS), includes data related to ambulatory care, hospitalisations and death for 98.8% of the population. In this study, algorithms based on age, diagnosis and healthcare consumption were created to identify adult patients with PF-ILD other than idiopathic pulmonary fibrosis between 2010 and 2017. Incidence, prevalence, survival estimates, clinical features and healthcare resource usage and costs were described among patients with PF-ILD. Results We identified a total of 14,413 patients with PF-ILD. Almost half of them (48.1%) were female and the mean (± standard deviation) age was 68.4 (± 15.0) years. Between 2010 and 2017, the estimated incidence of PF-ILD ranged from 4.0 to 4.7/100,000 person-years and the estimated prevalence from 6.6 to 19.4/100,000 persons. The main diagnostic categories represented were exposure-related ILD other than hypersensitivity pneumonitis (n = 3486; 24.2%), idiopathic interstitial pneumonia (n = 3113; 21.6%) and rheumatoid arthritis-associated ILD (n = 2521; 17.5%). Median overall survival using Kaplan–Meier estimation was 3.7 years from the start of progression. During the study, 95.2% of patients had ≥ 1 hospitalisation for respiratory care and 34.3% were hospitalised in an intensive care unit. The median (interquartile range) total specific cost per patient during the follow-up period was €25,613 (10,622–54,287) and the median annual cost per patient was €18,362 (6856–52,026), of which €11,784 (3003–42,097) was related to hospitalisations. Limitations included the retrospective design and identification of cases through an algorithm in the absence of chest high-resolution computed tomography scans and pulmonary function tests. Conclusions This large, real-world, longitudinal study provides important insights into the characteristics, epidemiology and healthcare resource utilisation and costs associated with PF-ILD in France using a comprehensive and exhaustive database, and provides vital evidence that PF-ILD represents a high burden on both patients and healthcare services. Trial registration ClinicalTrials.gov, NCT03858842. ISRCTN, ISRCTN12345678. Registered 3 January 2019—Retrospectively registered, https://clinicaltrials.gov/ct2/show/NCT03858842

The aim of this study is to compare the image quality of in vivo coronary stents between an energy integrating detectors dual-layer computed tomography (EID-DLCT) and a clinical prototype of spectral photon counting computed tomography (SPCCT).In January to June 2021, consecutive patients with coronary stents were prospectively enrolled to undergo a coronary computed tomography (CT) with an EID-DLCT (IQon, Philips) and an SPCCT (Philips). The study was approved by the local ethical committee and patients signed an informed consent. A retrospectively electrocardiogram-gated acquisition was performed with optimized matching parameters on the 2 scanners (EID-DLCT: collimation, 64 × 0.625 mm; kVp, 120, automatic exposure control with target current at 255 mAs; rotation time, 0.27 seconds; SPCCT: collimation, 64 × 0.275 mm; kVp, 120; mAs, 255; rotation time, 0.33 seconds). The injection protocol was the same on both scanners: 65 to 75 mL of Iomeron (Bracco) at 5 mL/s. Images were reconstructed with slice thickness of 0.67 mm, 512 matrix, XCB (Xres cardiac standard) and XCD (Xres cardiac detailed) kernel, iDose 3 for EID-DLCT and 0.25-mm slice thickness, 1024 matrix, Detailed 2 and Sharp kernel, and iDose 6 for SPCCT. Two experienced observers measured the proximal and distal external and internal diameters of the stents to quantify blooming artifacts. Regions of interest were drawn in the lumen of the stent and of the upstream coronary artery. The difference (Δ S-C) between the respective attenuation values was calculated as a quantification of stent-induced artifacts on intrastent image quality. For subjective image quality, 3 experienced observers graded with a 4-point scale the image quality of different parameters: coronary wall before the stent, stent lumen, stent structure, calcifications surrounding the stent, and beam-hardening artifacts.Eight patients (age, 68 years [interquartile range, 8]; all men; body mass index, 26.2 kg/m2 [interquartile range, 4.2]) with 16 stents were scanned. Five stents were not evaluable owing to motion artifacts on the SPCCT. Of the remaining, all were drug eluting stents, of which 6 were platinum-chromium, 3 were cobalt-platinum-iridium, and 1 was stainless steel. For 1 stent, no information could be retrieved. Radiation dose was lower with the SPCCT (fixed CT dose index of 25.7 mGy for SPCCT vs median CT dose index of 35.7 [IQ = 13.6] mGy; P = 0.02). For 1 stent, the internal diameter was not assessable on EID-DLCT. External diameters were smaller and internal diameters were larger with SPCCT (all P0.05). Consequently, blooming artifacts were reduced on SPCCT (P0.05). Whereas Hounsfield unit values within the coronary arteries on the 2 scanners were similar, the Δ S-C was lower for SPCCT-Sharp as compared with EID-DLCT-XCD and SPCCT-Detailed 2 (P0.05). The SPCCT received higher subjective scores than EID-DLCT for stent lumen, stent structure, surrounding calcifications and beam-hardening for both Detailed 2 and Sharp (all P ≤ 0.05). The SPCCT-Sharp was judged better for stent structure and beam-hardening assessment as compared with SPCCT-Detailed 2.Spectral photon counting CT demonstrated improved objective and subjective image quality as compared with EID-DLCT for the evaluation of coronary stents even with a reduced radiation dose.

PURPOSE In cardiovascular imaging, the numerous contrast injection protocols used to enhance structures make it difficult to gather training datasets for deep learning applications supporting diverse protocols. Moreover, creating annotations on non-contrast scans is extremely tedious. Recently, spectral CT's virtual-non-contrast images (VNC) have been used as data augmentation to train segmentation networks performing on enhanced and true-non-contrast (TNC) scans alike, while improving results on protocols absent of their training dataset. However, spectral data are not widely available, making it difficult to gather specific datasets for each task. As a solution, we present a data augmentation workflow based on a trained image translation network, to bring spectral-like augmentation to any conventional CT dataset. METHOD The HU-to-spectral image translation network (HUSpecNet) was first trained to generate VNC from HU images, using an unannotated spectral dataset of 1830 patients. It was then tested on a second dataset of 300 spectral CT scans, by comparing generated VNC (VNCDL ) to their true counterparts. To illustrate and compare our workflow's efficiency with true spectral augmentation, HUSpecNet was applied to a third dataset of 112 spectral scans to generate VNCDL along HU and VNC images. Three different 3D networks (U-Net, X-Net, U-Net++) were trained for multi-label heart segmentation, following four augmentation strategies. As baselines, trainings were performed on contrasted images without (HUonly) and with conventional gray-values augmentation (HUaug). Then, the same networks were trained using a proportion of contrasted and VNC/VNCDL images (TrueSpec/GenSpec). Each training strategy applied to each architecture was evaluated using Dice coefficients on a fourth multi-centric multi-vendor single-energy CT dataset of 121 patients, including different contrast injection protocols and unenhanced scans. The U-Net++ results were further explored with distance metrics on every label. RESULTS Tested on 300 full scans, our HUSpectNet translation network shows a mean absolute error of 6.70±2.83 HU between VNCDL and VNC, while peak-signal-to-noise-ratio reaches 43.89 dB. GenSpec and TrueSpec show very close results regardless of the protocol and used architecture: mean Dice coefficients (DSCmean ) are equal with a margin of 0.006, ranging from 0.879 to 0.938. Their performances significantly increase on TNC scans (p-values

The original version of this article, published on 02 May 2020, unfortunately contained a mistake.

To assess the capability of a newly developed material decomposition method from contrast-enhanced dual-energy CT images, aiming to better visualize the aortic wall and aortic intramural hematoma (IMH), compared with true non-contrast (TNC) CT. Twenty-two patients (11 women; mean age, 61 ± 20 years) with acute chest pain underwent 25 dual-layer non-contrast and contrast-enhanced CT. CT-angiography images were retrospectively processed using two-material decomposition analysis, where we defined the first material as the content of a region of interest placed in the ascending aorta for each patient, and the second material as water. Two independent radiologists assessed the images from the second material termed “dark-blood” images and the TNC images regarding contrast-to-noise ratio (CNR) between the wall and the lumen, diagnostic quality regarding the presence of aortic wall thickening, and the inner/outer vessel wall conspicuity. Diagnostic quality scores in normal aortic segments were 0.9 ± 0.3 and 2.7 ± 0.6 (p

To assess whether virtual non-contrast (VNC) images derived from contrast dual-layer dual-energy computed tomography (DL-DECT) images could replace true non-contrast (TNC) images for aortic intramural hematoma (IMH) diagnosis in acute aortic syndrome (AAS) imaging protocols by performing quantitative as well as qualitative phantom and clinical studies. Patients with confirmed IMH were included retrospectively in two centers. For in vitro imaging, a custom-made phantom of IMH was placed in a semi-anthropomorphic thorax phantom (QRM GmbH) and imaged on a DL-DECT at 120 kVp under various conditions of patient size, radiation exposure, and reconstruction modes. For in vivo imaging, 21 patients (70 ± 13 years) who underwent AAS imaging protocols at 120 kVp were included. In both studies, contrast-to-noise ratio (CNR) between hematoma and lumen was compared using a paired t test. Diagnostic confidence (1 = non-diagnostic, 4 = exemplary) for VNC and TNC images was rated by two radiologists and compared. Effective radiation doses for each acquisition were calculated. In both the phantom and clinical studies, we observed that the CNRs were similar between the VNC and TNC images. Moreover, both methods allowed differentiating the hyper-attenuation within the hematoma from the blood. Finally, we obtained equivalent high diagnostic confidence with both VNC and TNC images (VNC = 3.2 ± 0.7, TNC = 3.1 ± 0.7; p = 0.3). Finally, by suppressing TNC acquisition and using VNC, the mean effective dose reduction would be 40%. DL-DECT offers similar performances with VNC and TNC images for IMH diagnosis without compromise in diagnostic image quality. • Dual-layer dual-energy CT enables virtual non-contrast imaging from a contrast-enhanced acquisition. • Virtual non-contrast imaging with dual-layer dual-energy CT reduces the number of acquisitions and radiation exposure in acute aortic syndrome imaging protocol. • Dual-layer dual-energy CT has the potential to become a suitable imaging tool for acute aortic syndrome.

Purpose This work presents our contribution to a data challenge organized by the French Radiology Society during the Journees Francophones de Radiologie in October 2018. This challenge consisted in classifying MR images of the knee with respect to the presence of tears in the knee menisci, on meniscal tear location, and meniscal tear orientation. Materials and methods We trained a mask region-based convolutional neural network (R-CNN) to explicitly localize normal and torn menisci, made it more robust with ensemble aggregation, and cascaded it into a shallow ConvNet to classify the orientation of the tear. Results Our approach predicted accurately tears in the database provided for the challenge. This strategy yielded a weighted AUC score of 0.906 for all three tasks, ranking first in this challenge. Conclusion The extension of the database or the use of 3D data could contribute to further improve the performances especially for non-typical cases of extensively damaged menisci or multiple tears.

Background Dual-energy computed tomography has shown a great interest for musculoskeletal pathologies. Photon-counting spectral computed tomography (PCSCT) can acquire data in multiple energy bins with the potential to increase contrast, especially for soft tissues. Our objectives were to assess the value of PCSST to characterise cartilage and to extract quantitative measures of subchondral bone integrity. Methods Seven excised human knees (3 males and 4 females; 4 normal and 3 with osteoarthritis; age 80.6 ± 14 years, mean ± standard deviation) were scanned using a clinical PCSCT prototype scanner. Tomographic image reconstruction was performed after Compton/photoelectric decomposition. Virtual monoenergetic images were generated from 40 keV to 110 keV every 10 keV (cubic voxel size 250 × 250 × 250 μm3). After selecting an optimal virtual monoenergetic image, we analysed the grey level histograms of different tissues and extracted quantitative measurements on bone cysts. Results The optimal monoenergetic images were obtained for 60 keV and 70 keV. Visual inspection revealed that these images provide sufficient spatial resolution and soft-tissue contrast to characterise surfaces, disruption, calcification of cartilage, bone osteophytes, and bone cysts. Analysis of attenuation versus energy revealed different energy fingerprint according to tissues. The volumes and numbers of bone cyst were quantified. Conclusions Virtual monoenergetic images may provide direct visualisation of both cartilage and bone details. Thus, unenhanced PCSCT appears to be a new modality for characterising the knee joint with the potential to increase the diagnostic capability of computed tomography for joint diseases and osteoarthritis.

Objectives To compare the lung CT volume (CTvol) and pulmonary function tests in an interstitial lung disease (ILD) population. Then to evaluate the CTvol loss between idiopathic pulmonary fibrosis (IPF) and non-IPF and explore a prognostic value of annual CTvol loss in IPF. Methods We conducted in an expert center a retrospective study between 2005 and 2018 on consecutive patients with ILD. CTvol was measured automatically using commercial software based on a deep learning algorithm. In the first group, Spearman correlation coefficients (r) between forced vital capacity (FVC), total lung capacity (TLC), and CTvol were calculated. In a second group, annual CTvol loss was calculated using linear regression analysis and compared with the Mann–Whitney test. In a last group of IPF patients, annual CTvol loss was calculated between baseline and 1-year CTs for investigating with the Youden index a prognostic value of major adverse event at 3 years. Univariate and log-rank tests were calculated. Results In total, 560 patients (4610 CTs) were analyzed. For 1171 CTs, CTvol was correlated with FVC (r: 0.86) and TLC (r: 0.84) (p vol loss was 155.7 mL in IPF versus 50.7 mL in non-IPF (p vol loss of 7.9% was associated with major adverse events (log-rank, p p Conclusions Automated lung CT volume may be an alternative or a complementary biomarker to pulmonary function tests for the assessment of lung volume loss in ILD. Key Points • There is a good correlation between lung CT volume and forced vital capacity, as well as for with total lung capacity measurements (r of 0.86 and 0.84 respectively, p 0.0001). • Median annual CT volume loss is significantly higher in patients with idiopathic pulmonary fibrosis than in patients with other fibrotic interstitial lung diseases (155.7 versus 50.7 mL, p 0.0001). • In idiopathic pulmonary fibrosis, a relative annual CT volume loss higher than 9.4% is associated with a significantly reduced mean survival time at 2.0 years versus 2.8 years (log-rank, p 0.0001).

Background Macrophage burden is a major factor in the risk of atherosclerotic plaque rupture, and its evaluation remains challenging with molecular noninvasive imaging approaches. Photon-counting CT (PCCT) with k-edge imaging aims to allow for the specific detection of macrophages using gold nanoparticles. Purpose To perform k-edge imaging in combination with gold nanoparticles to detect and quantify the macrophage burden within the atherosclerotic aortas of rabbits. Materials and Methods Atherosclerotic and control New Zealand white rabbits were imaged before and at several time points up to 2 days after intravenous injection of gold nanoparticles (3.5 mL/kg, 65 mg gold per milliliter). Aortic CT angiography was performed at the end of the follow-up using an intravenous injection of an iodinated contrast material. Gold k-edge and conventional CT images were reconstructed for qualitative and quantitative assessment of the macrophage burden. PCCT imaging results were compared with findings at histologic examination, quantitative histomorphometry, transmission electron microscopy, and quantitative inductively coupled plasma optical emission spectrometry. Pearson correlations between the macrophage area measured in immunostained sections and the concentration of gold and attenuation measured in the corresponding PCCT sections were calculated. Results Seven rabbits with atherosclerosis and four control rabbits without atherosclerosis were analyzed. In atherosclerotic rabbits, calcifications were observed along the aortic wall before injection. At 2 days after injection of gold nanoparticles, only gold k-edge images allowed for the distinction of plaque enhancement within calcifications and for lumen enhancement during angiography. A good correlation was observed between the gold concentration measured within the wall and the macrophage area in 35 plaques (five per rabbit) (

To evaluate the image quality (IQ) of a spectral photon-counting CT (SPCCT) using filtered back projection (FBP) and hybrid iterative reconstruction (IR) algorithms (iDoseTwo phantoms were scanned using a standard lung protocol (120 kVp, 40 mAs) with SPCCT and DLCT systems. Raw data were reconstructed using FBP and 9 iDoseNoise magnitude was -47% ± 2% lower on average with SPCCT than with DLCT for iDoseHigher IQ for GGN and solid nodules was demonstrated with SPCCT compared with DLCT with better detectability using iDoseUsing spectral photon-counting CT compared with dual-layer CT, noise magnitude was reduced with improvements in spatial resolution and detectability of ground-glass nodules and solid lung nodules. As the iDose

Objective The aim of the current study was, first, to assess the coronary artery calcium (CAC) scoring potential of spectral photon-counting CT (SPCCT) in comparison with computed tomography (CT) for routine clinical protocols. Second, improved CAC detection and quantification at reduced slice thickness were assessed. Methods Raw data was acquired and reconstructed with several combinations of reduced slice thickness and increasing strengths of iterative reconstruction (IR) for both CT systems with routine clinical CAC protocols for CT. Two CAC-containing cylindrical inserts, consisting of CAC of different densities and sizes, were placed in an anthropomorphic phantom. A specific CAC was detectable when 3 or more connected voxels exceeded the CAC scoring threshold of 130 Hounsfield units (HU). For all reconstructions, total CAC detectability was compared between both CT systems. Significant differences in CAC quantification (Agatston and volume scores) were assessed with Mann-Whitney U tests. Furthermore, volume scores were compared with the known CAC physical. Results CAC scores for routine clinical protocols were comparable between SPCCT and CT. SPCCT showed 34% and 4% higher detectability of CAC for the small and large phantom, respectively. At reduced slice thickness, CAC detection increased by 142% and 169% for CT and SPCCT, respectively. In comparison with CT, volume scores from SPCCT were more comparable with the physical volume of the CAC. Conclusion CAC scores using routine clinical protocols are comparable between conventional CT and SPCCT. The increased spatial resolution of SPCCT allows for increased detectability and more accurate CAC volume estimation. Key Points • Coronary artery calcium scores using routine clinical protocols are comparable between conventional CT and spectral photon-counting CT. • In comparison with conventional CT, increased coronary artery calcium detectability was shown for spectral photon-counting CT due to increased spatial resolution. • Volumes scores were more accurately determined with spectral photon-counting CT.

Sarcoidosis is a multi-system disease of unknown etiology characterized by the formation of granulomas in various organs. It affects people of all ethnic backgrounds and occurs at any time of life but is more frequent in African Americans and Scandinavians and in adults between 30 and 50 years of age. Sarcoidosis can affect any organ with a frequency varying according to ethnicity, sex and age. Intrathoracic involvement occurs in 90% of patients with symmetrical bilateral hilar adenopathy and/or diffuse lung micronodules, mainly along the lymphatic structures which are the most affected system. Among extrapulmonary manifestations, skin lesions, uveitis, liver or splenic involvement, peripheral and abdominal lymphadenopathy and peripheral arthritis are the most frequent with a prevalence of 25–50%. Finally, cardiac and neurological manifestations which can be the initial manifestation of sarcoidosis, as can be bilateral parotitis, nasosinusal or laryngeal signs, hypercalcemia and renal dysfunction, affect less than 10% of patients. The diagnosis is not standardized but is based on three major criteria: a compatible clinical and/or radiological presentation, the histological evidence of non-necrotizing granulomatous inflammation in one or more tissues and the exclusion of alternative causes of granulomatous disease. Certain clinical features are considered to be highly specific of the disease (e.g., Löfgren’s syndrome, lupus pernio, Heerfordt’s syndrome) and do not require histological confirmation. New diagnostic guidelines were recently published. Specific clinical criteria have been developed for the diagnosis of cardiac, neurological and ocular sarcoidosis. This article focuses on the clinical presentation and the common differentials that need to be considered when appropriate.

PURPOSE: To develop and evaluate a complete deep learning pipeline that allows fully automated end-diastolic left ventricle (LV) cardiac MRI segmentation, including trabeculations and automatic quality control of the predicted segmentation. MATERIALS AND METHODS: This multicenter retrospective study includes training, validation, and testing datasets of 272, 27, and 150 cardiac MR images, respectively, collected between 2012 and 2018. The reference standard was the manual segmentation of four LV anatomic structures performed on end-diastolic short-axis cine cardiac MRI: LV trabeculations, LV myocardium, LV papillary muscles, and the LV blood cavity. The automatic pipeline was composed of five steps with a DenseNet architecture. Intraobserver agreement, interobserver agreement, and interaction time were recorded. The analysis includes the correlation between the manual and automated segmentation, a reproducibility comparison, and Bland-Altman plots. RESULTS: The automated method achieved mean Dice coefficients of 0.96 ± 0.01 (standard deviation) for LV blood cavity, 0.89 ± 0.03 for LV myocardium, and 0.62 ± 0.08 for LV trabeculation (mean absolute error, 3.63 g ± 3.4). Automatic quantification of LV end-diastolic volume, LV myocardium mass, LV trabeculation, and trabeculation mass–to–total myocardial mass (TMM) ratio showed a significant correlation with the manual measures (r = 0.99, 0.99, 0.90, and 0.83, respectively; all P < .01). On a subset of 48 patients, the mean Dice value for LV trabeculation was 0.63 ± 0.10 or higher compared with the human interobserver (0.44 ± 0.09; P < .01) and intraobserver measures (0.58 ± 0.09; P < .01). Automatic quantification of the trabeculation mass–to-TMM ratio had a higher correlation (0.92) compared with the intra- and interobserver measures (0.74 and 0.39, respectively; both P < .01). CONCLUSION: Automated deep learning framework can achieve reproducible and quality-controlled segmentation of cardiac trabeculations, outperforming inter- and intraobserver analyses. Supplemental material is available for this article. © RSNA, 2020

Additional file 1. Additional methods and results.

International audience; No abstract available

Purpose The purpose of this study was to characterize the technical capabilities and feasibility of a large field-of-view clinical spectral photon-counting computed tomography (SPCCT) prototype for high-resolution (HR) lung imaging. Materials and methods Measurement of modulation transfer function (MTF) and acquisition of a line pairs phantom were performed. An anthropomorphic lung nodule phantom was scanned with standard (120 kVp, 62 mAs), low (120 kVp, 11 mAs), and ultra-low (80 kVp, 3 mAs) radiation doses. A human volunteer underwent standard (120 kVp, 63 mAs) and low (120 kVp, 11 mAs) dose scans after approval by the ethics committee. HR images were reconstructed with 1024 matrix, 300 mm field of view and 0.25 mm slice thickness using a filtered-back projection (FBP) and two levels of iterative reconstruction (iDose 5 and 9). The conspicuity and sharpness of various lung structures (distal airways, vessels, fissures and proximal bronchial wall), image noise, and overall image quality were independently analyzed by three radiologists and compared to a previous HR lung CT examination of the same volunteer performed with a conventional CT equipped with energy integrating detectors (120 kVp, 10 mAs, FBP). Results Ten percent MTF was measured at 22.3 lp/cm with a cut-off at 31 lp/cm. Up to 28 lp/cm were depicted. While mixed and solid nodules were easily depicted on standard and low-dose phantom images, higher iDose levels and slice thicknesses (1 mm) were needed to visualize ground-glass components on ultra-low-dose images. Standard dose SPCCT images of in vivo lung structures were of greater conspicuity and sharpness, with greater overall image quality, and similar image noise (despite a flux reduction of 23%) to conventional CT images. Low-dose SPCCT images were of greater or similar conspicuity and sharpness, similar overall image quality, and lower but acceptable image noise (despite a flux reduction of 89%). Conclusions A large field-of-view SPCCT prototype demonstrates HR technical capabilities and high image quality for high resolution lung CT in human.

Spectral CT scanner is an emerging technology in the clinical field, which al-lows to generate from a single scan conventional HU images, virtual non-contrast images (VNC) corresponding to unenhanced CT images on conventional CT scanners, and virtual mono-energetic (monoE) images at different keV that mimics low (at high keV) to high (at low keV) iodine-based contrast-enhanced studies. It has been demonstrated that these spectral images could be used as data augmentation for cardiovascular structures segmentation, improving dice scores on contrasted images, and enabling to segment true non-contrast (TNC) conventional scans. However, as of now, spectral CT are not widely distributed, making this data augmentation process not widely available.To overcome this limitation, we propose in this study CASPER, a full data augmentation workflow to provide any conventional CT training dataset with spectral CT augmentation capabilities. We trained on an unannotated data-base of 2500 spectral scans an image translation network (HUSpecNet) to create monoEs and VNC from conventional HU images. Any conventional CT scan exam can then be translated to mimic different contrast agent injection protocols.We evaluated our CASPER methodology by training a 3D U-net to segment the aorta, both with and without data augmentation on 143 annotated conventional scans. The network performances were compared on a testing dataset of 70 patients, including pulmonary embolism (PE) early enhancement scans, full chest-abdominal-pelvic delayed enhancement scans and TNC scans. The network trained using CASPER outperformed significantly the network trained without it, on every imaging protocols, and especially on TNC.