Phenotypes and outcome of diffuse pulmonary non-amyloid light chain deposition disease.

BACKGROUND: Light chain deposition disease (LCDD) is a very rare entity. Clinical manifestations of LCDD vary according to the organs involved. Data on pulmonary LCDD are scarce and limited to small series or case reports. This study aimed to describe the characteristics and outcome of diffuse pulmonary non-amyloid LCDD localized to the lungs. STUDY DESIGN AND METHODS: A multicenter retrospective cohort study was conducted. Clinical characteristics were collected, and chest CTs were centrally reviewed. The diagnosis of pulmonary non-amyloid LCDD was confirmed by immunohistochemistry. RESULTS: Thirty-one cases were identified (68% female), with a median age at diagnosis of 50 years (IQR 20). Baseline FEV1/FVC was < 0.70 in 45% of patients. Mean (± SD) FEV1 and DLCO were 86% ± 26.2 and 52% ± 23.9, respectively. CT revealed peculiar patterns of thin-walled cysts (58%) and thin-walled cystic bronchiectases (27%). Increased serum kappa light chain was found in 87% of patients. Histological analysis showed kappa light chain deposits in all patients, except one with lambda chain deposits. Median annual FEV1 decline was 127 ml (IQR 178) and median DLCO decline was 4.3% (IQR 4.3). Sixteen patients received immunomodulatory treatment or chemotherapy; serum light chain levels decreased in 9 cases (75%), without significant improvement in FEV1 (p = 0.173). Overall, 48% of patients underwent bilateral lung transplantation. Transplant-free survival at 5 and 10 years were 70% and 30%, respectively. An annual FEV1 decline greater than 127 ml/year was associated with increased risk of death or transplantation (p = 0.005). CONCLUSIONS: Diffuse pulmonary LCDD is characterised by female predominance, a peculiar imaging pattern with bronchiectasis and/or cysts, progressive airway obstruction and severe DLCO impairment, and poor outcome. Lung transplantation is a treatment of choice.

Intra-bin correction and inter-bin compensation of respiratory motion in free-running five-dimensional whole-heart magnetic resonance imaging.

BACKGROUND: Free-running cardiac and respiratory motion-resolved whole-heart five-dimensional (5D) cardiovascular magnetic resonance (CMR) can reduce scan planning and provide a means of evaluating respiratory-driven changes in clinical parameters of interest. However, respiratory-resolved imaging can be limited by user-defined parameters which create trade-offs between residual artifact and motion blur. In this work, we develop and validate strategies for both correction of intra-bin and compensation of inter-bin respiratory motion to improve the quality of 5D CMR. METHODS: Each component of the reconstruction framework was systematically validated and compared to the previously established 5D approach using simulated free-running data (N = 50) and a cohort of 32 patients with congenital heart disease. The impact of intra-bin respiratory motion correction was evaluated in terms of image sharpness while inter-bin respiratory motion compensation was evaluated in terms of reconstruction error, compression of respiratory motion, and image sharpness. The full reconstruction framework (intra-acquisition correction and inter-acquisition compensation of respiratory motion [IIMC] 5D) was evaluated in terms of image sharpness and scoring of image quality by expert reviewers. RESULTS: Intra-bin motion correction provides significantly (p < 0.001) sharper images for both simulated and patient data. Inter-bin motion compensation results in significant (p < 0.001) lower reconstruction error, lower motion compression, and higher sharpness in both simulated (10/11) and patient (9/11) data. The combined framework resulted in significantly (p < 0.001) sharper IIMC 5D reconstructions (End-expiration (End-Exp): 0.45 ± 0.09, End-inspiration (End-Ins): 0.46 ± 0.10) relative to the previously established 5D implementation (End-Exp: 0.43 ± 0.08, End-Ins: 0.39 ± 0.09). Similarly, image scoring by three expert reviewers was significantly (p < 0.001) higher using IIMC 5D (End-Exp: 3.39 ± 0.44, End-Ins: 3.32 ± 0.45) relative to 5D images (End-Exp: 3.02 ± 0.54, End-Ins: 2.45 ± 0.52). CONCLUSION: The proposed IIMC reconstruction significantly improves the quality of 5D whole-heart MRI. This may be exploited for higher resolution or abbreviated scanning. Further investigation of the diagnostic impact of this framework and comparison to gold standards is needed to understand its full clinical utility, including exploration of respiratory-driven changes in physiological measurements of interest.

Interstitial Lung Disease Associated with Systemic Sclerosis.

Systemic sclerosis (SSc) is a rare autoimmune disease characterized by a tripod combining vasculopathy, fibrosis, and immune-mediated inflammatory processes. The prevalence of interstitial lung disease (ILD) in SSc varies according to the methods used to detect it, ranging from 25 to 95%. The fibrotic and vascular pulmonary manifestations of SSc, particularly ILD, are the main causes of morbidity and mortality, contributing to 35% of deaths. Although early trials were conducted with cyclophosphamide, more recent randomized controlled trials have been performed to assess the efficacy and tolerability of several medications, mostly mycophenolate, rituximab, tocilizumab, and nintedanib. Although many uncertainties remain, expert consensus is emerging to optimize the therapeutic management and to provide clinicians with evidence-based clinical practice guidelines for patients with SSc-ILD. This article provides an overview, in the light of the latest advances, of the available evidence for the diagnosis and management of SSc-ILD.

Similarities and differences of interstitial lung disease associated with pathogenic variants in SFTPC and ABCA3 in adults.

BACKGROUND AND OBJECTIVE: Variants in surfactant genes SFTPC or ABCA3 are responsible for interstitial lung disease (ILD) in children and adults, with few studies in adults. METHODS: We conducted a multicentre retrospective study of all consecutive adult patients diagnosed with ILD associated with variants in SFTPC or ABCA3 in the French rare pulmonary diseases network, OrphaLung. Variants and chest computed tomography (CT) features were centrally reviewed. RESULTS: We included 36 patients (median age: 34 years, 20 males), 22 in the SFTPC group and 14 in the ABCA3 group. Clinical characteristics were similar between groups. Baseline median FVC was 59% ([52-72]) and DLco was 44% ([35-50]). An unclassifiable pattern of fibrosing ILD was the most frequent on chest CT, found in 85% of patients, however with a distinct phenotype with ground-glass opacities and/or cysts. Nonspecific interstitial pneumonia and usual interstitial pneumonia were the most common histological patterns in the ABCA3 group and in the SFTPC group, respectively. Annually, FVC and DLCO declined by 1.87% and 2.43% in the SFTPC group, respectively, and by 0.72% and 0.95% in the ABCA3 group, respectively (FVC, p = 0.014 and DLCO , p = 0.004 for comparison between groups). Median time to death or lung transplantation was 10 years in the SFTPC group and was not reached at the end of follow-up in the ABCA3 group. CONCLUSION: SFTPC and ABCA3-associated ILD present with a distinct phenotype and prognosis. A radiologic pattern of fibrosing ILD with ground-glass opacities and/or cysts is frequently found in these rare conditions.

Device Design and Diagnostic Imaging of Radiopaque 3D Printed Tissue Engineering Scaffolds.

3D printed biomaterial implants are revolutionizing personalized medicine for tissue repair, especially in orthopedics. In this study, a radiopaque Bi 2 O 3 doped polycaprolactone ( PCL ) composite is developed and implemented to enable the use of diagnostic X-ray technologies, especially photon counting X-ray computed tomography ( PCCT ), for comprehensive in vivo device monitoring. PCL filament with homogeneous Bi 2 O 3 nanoparticle ( NP ) dispersion (0.8 to 11.7 wt%) are first fabricated. Tissue engineered scaffolds ( TES ) are then 3D printed with the composite filament, optimizing printing parameters for small feature size and severely overhung geometries. These composite TES are characterized via micro-computed tomography ( µ CT ), tensile testing, and a cytocompatibility study, with Bi 2 O 3 mass fractions as low as 2 wt% providing excellent radiographic distinguishability, improved tensile properties, and equivalent cytocompatibility of neat PCL. The excellent radiographic distinguishability is validated in situ by imaging 4 and 7 wt% TES in a mouse model with µCT, showing excellent agreement with in vitro measurements. Subsequently, CT image-derived swine menisci are 3D printed with composite filament and re-implanted in their corresponding swine legs ex vivo . Re-imaging the swine legs via clinical CT allows facile identification of device location and alignment. Finally, the emergent technology of PCCT unambiguously distinguishes implanted menisci in situ.

Ultra-High-Resolution and K-Edge Imaging of Prosthetic Heart Valves With Spectral Photon-Counting CT: A Phantom Study.

BACKGROUND AND PURPOSE: The contribution of cardiac computed tomography (CT) for the detection and characterization of prosthetic heart valve (PHV) complications is still limited due mainly to artifacts. Computed tomography systems equipped with photon-counting detectors (PCDs) have the potential to overcome these limitations. Therefore, the aim of the study was to compare image quality of PHV with PCD-CT and dual-energy dual-layer CT (DEDL-CT). MATERIALS AND METHODS: Two metallic and 3 biological PHVs were placed in a tube containing diluted iodinated contrast inside a thoracic phantom and scanned repeatedly at different angles on a DEDL-CT and PCD-CT. Two small lesions (~2 mm thickness; containing muscle and fat, respectively) were attached to the structure of 4 valves, placed inside the thoracic phantom, with and without an extension ring, and scanned again. Acquisition parameters were matched for the 2 CT systems and used for all scans. Metallic valves were scanned again with parameters adapted for tungsten K-edge imaging. For all valves, different metallic parts were measured on conventional images to assess their thickness and blooming artifacts. In addition, 6 parallelepipeds per metallic valve were drawn, and all voxels with density <3 times the standard deviation of the contrast media were recorded as an estimate of streak artifacts. For subjective analysis, 3 expert readers assessed conventional images of the valves, with and without lesions, and tungsten K-edge images. Conspicuity and sharpness of the different parts of the valve, the lesions, metallic, and blooming artifacts were scored on a 4-point scale. Measurements and scores were compared with the paired t test or Wilcoxon test. RESULTS: The objective analysis showed that, with PCD-CT, valvular metallic structures were thinner and presented less blooming artifacts. Metallic artifacts were also reduced with PCD-CT (11 [interquartile (IQ) = 6] vs 40 [IQ = 13] % of voxels). Subjective analysis allowed noticing that some structures were visible or clearly visible only with PCD-CT. In addition, PCD-CT yielded better scores for the conspicuity and for the sharpness of all structures (all P s < 0.006), except for the conspicuity of the leaflets of the mechanical valves, which were well visible with either technique (4 [IQ = 3] for both). Both blooming and streak artifacts were reduced with PCD-CT ( P ≤ 0.01). Overall, the use of PCD-CT resulted in better conspicuity and sharpness of the lesions compared with DEDL-CT (both P s < 0.02). In addition, only with PCD-CT some differences between the 2 lesions were detectable. Adding the extension ring resulted in reduced conspicuity and sharpness with DEDL-CT ( P = 0.04 and P = 0.02, respectively) and only in reduced sharpness with PCD-CT ( P = 0.04). Tungsten K-edge imaging allowed for the visualization of the only dense structure containing it, the leaflets, and it resulted in images judged having less blooming and metallic artifacts as compared with conventional PCD-CT images ( P < 0.01). CONCLUSIONS: With PCD-CT, objective and subjective image quality of metallic and biological PHVs is improved compared with DEDL-CT. Notwithstanding the improvements in image quality, millimetric lesions attached to the structure of the valves remain a challenge for PCD-CT. Tungsten K-edge imaging allows for even further reduction of artifacts.

Quantification of Coronary Artery Stenosis in Very-High-Risk Patients Using Ultra-High Resolution Spectral Photon-Counting CT.

OBJECTIVE: Development of spectral photon-counting computed tomography (SPCCT) for ultra-high-resolution coronary CT angiography (CCTA) has the potential to accurately evaluate the coronary arteries of very-high-risk patients. The aim of this study was to compare the diagnostic performances of SPCCT against conventional CT for quantifying coronary stenosis in very-high-risk patients, with invasive coronary angiography (ICA) as the reference method. MATERIALS AND METHODS: In this prospective institutional review board-approved study, very-high-risk patients addressed for ICA following an acute coronary syndrome were consecutively included. CCTA was performed for each patient with both SPCCT and conventional CT before ICA within 3 days. Stenoses were assessed using the minimal diameter over proximal and distal diameters method for CCTA and the quantitative coronary angiography method for ICA. Intraclass correlation coefficients and mean errors were assessed. Sensitivity and specificity were calculated for a >50% diameter stenosis threshold. Reclassification rates for conventional CT and SPCCT were assessed according to CAD-RADS 2.0, using ICA as the gold standard. RESULTS: Twenty-six coronary stenoses were identified in 26 patients (4 women [15%]; age 64 ± 8 years) with 19 (73%) above 50% and 9 (35%) equal or above 70%. The median stenosis value was 64% (interquartile range, 48%-73%). SPCCT showed a lower mean error (6% [5%, 8%]) than conventional CT (12% [9%, 16%]). SPCCT demonstrated greater sensitivity (100%) and specificity (90%) than conventional CT (75% and 50%, respectively). Ten (38%) stenoses were reclassified with SPCCT and one (4%) with conventional CT. CONCLUSIONS: In very-high-risk patients, ultra-high-resolution SPCCT coronary angiography showed greater accuracy, sensitivity, and specificity, and led to more stenosis reclassifications than conventional CT.

Weakly-supervised learning-based pathology detection and localization in 3D chest CT scans.

BACKGROUND: Recent advancements in anomaly detection have paved the way for novel radiological reading assistance tools that support the identification of findings, aimed at saving time. The clinical adoption of such applications requires a low rate of false positives while maintaining high sensitivity. PURPOSE: In light of recent interest and development in multi pathology identification, we present a novel method, based on a recent contrastive self-supervised approach, for multiple chest-related abnormality identification including low lung density area ("LLDA"), consolidation ("CONS"), nodules ("NOD") and interstitial pattern ("IP"). Our approach alerts radiologists about abnormal regions within a computed tomography (CT) scan by providing 3D localization. METHODS: We introduce a new method for the classification and localization of multiple chest pathologies in 3D Chest CT scans. Our goal is to distinguish four common chest-related abnormalities: "LLDA", "CONS", "NOD", "IP" and "NORMAL". This method is based on a 3D patch-based classifier with a Resnet backbone encoder pretrained leveraging recent contrastive self supervised approach and a fine-tuned classification head. We leverage the SimCLR contrastive framework for pretraining on an unannotated dataset of randomly selected patches and we then fine-tune it on a labeled dataset. During inference, this classifier generates probability maps for each abnormality across the CT volume, which are aggregated to produce a multi-label patient-level prediction. We compare different training strategies, including random initialization, ImageNet weight initialization, frozen SimCLR pretrained weights and fine-tuned SimCLR pretrained weights. Each training strategy is evaluated on a validation set for hyperparameter selection and tested on a test set. Additionally, we explore the fine-tuned SimCLR pretrained classifier for 3D pathology localization and conduct qualitative evaluation. RESULTS: Validated on 111 chest scans for hyperparameter selection and subsequently tested on 251 chest scans with multi-abnormalities, our method achieves an AUROC of 0.931 (95% confidence interval [CI]: [0.9034, 0.9557], p $ p$ -value < 0.001) and 0.963 (95% CI: [0.952, 0.976], p $ p$ -value < 0.001) in the multi-label and binary (i.e., normal versus abnormal) settings, respectively. Notably, our method surpasses the area under the receiver operating characteristic (AUROC) threshold of 0.9 for two abnormalities: IP (0.974) and LLDA (0.952), while achieving values of 0.853 and 0.791 for NOD and CONS, respectively. Furthermore, our results highlight the superiority of incorporating contrastive pretraining within the patch classifier, outperforming Imagenet pretraining weights and non-pretrained counterparts with uninitialized weights (F1 score = 0.943, 0.792, and 0.677 respectively). Qualitatively, the method achieved a satisfactory 88.8% completeness rate in localization and maintained an 88.3% accuracy rate against false positives. CONCLUSIONS: The proposed method integrates self-supervised learning algorithms for pretraining, utilizes a patch-based approach for 3D pathology localization and develops an aggregation method for multi-label prediction at patient-level. It shows promise in efficiently detecting and localizing multiple anomalies within a single scan.

Quiescent frame, contrast-enhanced coronary magnetic resonance angiography reconstructed using limited number of physiologic frames from 5D free-running acquisitions.

BACKGROUND: 5D, free-running imaging resolves sets of 3D whole-heart images in both cardiac and respiratory dimensions. In an application such as coronary imaging when a single, static image is of interest, computationally expensive offline iterative reconstruction is still needed to compute the multiple 3D datasets. PURPOSE: Evaluate how the number of physiologic bins included in the reconstruction affects the computational cost and resulting image quality of a single, static volume reconstruction. STUDY TYPE: Retrospective. SUBJECTS: 15 pediatric patients following Ferumoxytol infusion (4 mg/kg). FIELD STRENGTH/SEQUENCE: 1.5 T/Ungated 5D free-running GRE sequence. ASSESSMENT: The raw data of each subject were binned and reconstructed into a 5D (x-y-z-cardiac-respiratory) images. 1, 3, 5, 7, and 9 bins adjacent to both sides of the retrospectively determined cardiac resting phase and 1, 3 bins adjacent to the end-expiration phase are used for limited frame reconstructions. The static volume within each limited reconstruction was compared with the corresponding full 5D reconstruction using the structural similarity index measure (SSIM). A non-linear regression model was used to fit SSIM with the percentage of data used compared to full reconstruction (% data). A linear regression model was used to fit computation time with % raw data used. Coronary artery sharpness is measured on each limited reconstructed images to determine the minimal number of cardiac and respiratory bins needed to preserve image quality. STATISTICAL TESTS: The coefficient of determination (R2) is computed for each regression model. RESULTS: The % of data used in the reconstruction was linearly related to the computational time (R2 = 0.99). The SSIM of the static image from the limited reconstructions is non-linearly related with the % of data used (R2 = 0.80). Over the 15 patients, the model showed SSIM of 0.9 with 18% of data, and SSIM of 0.96 with 30% of data. The coronary artery sharpness of images reconstructed using no less than 5 cardiac and all respiratory phases is not significantly different from the full reconstructed images using all cardiac and respiratory bins. DATA CONCLUSION: Reconstruction using only a limited number of acquired physiological states can linearly reduce the computational cost while preserving similarity to the full reconstruction image. It is suggested to use no less than 5 cardiac and all respiratory phases in the limited reconstruction to best preserve the original quality seen on the full reconstructed images.

Similarity-driven motion-resolved reconstruction for ferumoxytol-enhanced whole-heart MRI in congenital heart disease.

A similarity-driven multi-dimensional binning algorithm (SIMBA) reconstruction of free-running cardiac magnetic resonance imaging data was previously proposed. While very efficient and fast, the original SIMBA focused only on the reconstruction of a single motion-consistent cluster, discarding the remaining data acquired. However, the redundant data clustered by similarity may be exploited to further improve image quality. In this work, we propose a novel compressed sensing (CS) reconstruction that performs an effective regularization over the clustering dimension, thanks to the integration of inter-cluster motion compensation (XD-MC-SIMBA). This reconstruction was applied to free-running ferumoxytol-enhanced datasets from 24 patients with congenital heart disease, and compared to the original SIMBA, the same XD-MC-SIMBA reconstruction but without motion compensation (XD-SIMBA), and a 5D motion-resolved CS reconstruction using the free-running framework (FRF). The resulting images were compared in terms of lung-liver and blood-myocardium sharpness, blood-myocardium contrast ratio, and visible length and sharpness of the coronary arteries. Moreover, an automated image quality score (IQS) was assigned using a pretrained deep neural network. The lung-liver sharpness and blood-myocardium sharpness were significantly higher in XD-MC-SIMBA and FRF. Consistent with these findings, the IQS analysis revealed that image quality for XD-MC-SIMBA was improved in 18 of 24 cases, compared to SIMBA. We successfully tested the hypothesis that multiple motion-consistent SIMBA clusters can be exploited to improve the quality of ferumoxytol-enhanced cardiac MRI when inter-cluster motion-compensation is integrated as part of a CS reconstruction.

Ultra-low dose chest CT for the diagnosis of pulmonary arteriovenous malformation in patients with hereditary hemorrhagic telangiectasia.

PURPOSE: The purpose of this study was to compare ultra-low dose (ULD) and standard low-dose (SLD) chest computed tomography (CT) in terms of radiation exposure, image quality and diagnostic value for diagnosing pulmonary arteriovenous malformation (AVM) in patients with hereditary hemorrhagic telangiectasia (HHT). MATERIALS AND METHODS: In this prospective board-approved study consecutive patients with HHT referred to a reference center for screening and/or follow-up chest CT examination were prospectively included from December 2020 to January 2022. Patients underwent two consecutive non-contrast chest CTs without dose modulation (i.e., one ULD protocol [80 kVp or 100 kVp, CTDIvol of 0.3 mGy or 0.6 mGy] and one SLD protocol [140 kVp, CTDIvol of 1.3 mGy]). Objective image noises measured at the level of tracheal carina were compared between the two protocols. Overall image quality and diagnostic confidence were scored on a 4-point Likert scale (1 = insufficient to 4 = excellent). Sensitivity, specificity, positive predictive value and negative predictive value of ULD CT for diagnosing pulmonary AVM with a feeding artery of over 2 mm in diameter were calculated along with their 95% confidence intervals (CI) using SLD images as the standard of reference. RESULTS: A total of 44 consecutive patients with HHT (31 women; mean age, 42 ± 16 [standard deviation (SD)] years; body mass index, 23.2 ± 4.5 [SD] kg/m2) were included. Thirty-four pulmonary AVMs with a feeding artery of over 2 mm in diameter were found with SLD images versus 35 with ULD images. Sensitivity, specificity, predictive positive value, and predictive negative value of ULD CT for the diagnosis of PAVM were 100% (34/34; 95% CI: 90-100), 96% (18/19; 95% CI: 74-100), 97% (34/35; 95% CI: 85-100) and 100% (18/18; 95% CI: 81-100), respectively. A significant difference in diagnostic confidence scores was found between ULD (3.8 ± 0.4 [SD]) and SLD (3.9 ± 0.1 [SD]) CT images (P = 0.03). No differences in overall image quality scores were found between ULD CT examinations (3.9 ± 0.2 [SD]) and SLD (4 ± 0 [SD]) CT examinations (P = 0.77). Effective radiation dose decreased significantly by 78.8% with ULD protocol, with no significant differences in noise values between ULD CT images (16.7 ± 5.0 [SD] HU) and SLD images (17.7 ± 6.6 [SD] HU) (P = 0.07). CONCLUSION: ULD chest CT provides 100% sensitivity and 96% specificity for the diagnosis of treatable pulmonary AVM with a feeding artery of over 2 mm in diameter, leading to a 78.8% dose-saving compared with a standard low-dose protocol.

Surface modification effect on contrast agent efficiency for X-ray based spectral photon-counting scanner/luminescence imaging: from fundamental study to in vivo proof of concept.

X-Ray imaging techniques are among the most widely used modalities in medical imaging and their constant evolution has led to the emergence of new technologies. The new generation of computed tomography (CT) systems - spectral photonic counting CT (SPCCT) and X-ray luminescence optical imaging - are examples of such powerful techniques. With these new technologies the rising demand for new contrast agents has led to extensive research in the field of nanoparticles and the possibility to merge the modalities appears to be highly attractive. In this work, we propose the design of lanthanide-based nanocrystals as a multimodal contrast agent with the two aforementioned technologies, allowing SPCCT and optical imaging at the same time. We present a systematic study on the effect of the Tb3+ doping level and surface modification on the generation of contrast with SPCCT and the luminescence properties of GdF3:Tb3+ nanocrystals (NCs), comparing different surface grafting with organic ligands and coatings with silica to make these NCs bio-compatible. A comparison of the luminescence properties of these NCs with UV revealed that the best results were obtained for the Gd0.9Tb0.1F3 composition. This property was confirmed under X-ray excitation in microCT and with SPCCT. Moreover, we could demonstrate that the intensity of the luminescence and the excited state lifetime are strongly affected by the surface modification. Furthermore, whatever the chemical nature of the ligand, the contrast with SPCCT did not change. Finally, the successful proof of concept of multimodal imaging was performed in vivo with nude mice in the SPCCT taking advantage of the so-called color K-edge imaging method.

Detection and severity quantification of pulmonary embolism with 3D CT data using an automated deep learning-based artificial solution.

PURPOSE: The purpose of this study was to propose a deep learning-based approach to detect pulmonary embolism and quantify its severity using the Qanadli score and the right-to-left ventricle diameter (RV/LV) ratio on three-dimensional (3D) computed tomography pulmonary angiography (CTPA) examinations with limited annotations. MATERIALS AND METHODS: Using a database of 3D CTPA examinations of 1268 patients with image-level annotations, and two other public datasets of CTPA examinations from 91 (CAD-PE) and 35 (FUME-PE) patients with pixel-level annotations, a pipeline consisting of: (i), detecting blood clots; (ii), performing PE-positive versus negative classification; (iii), estimating the Qanadli score; and (iv), predicting RV/LV diameter ratio was followed. The method was evaluated on a test set including 378 patients. The performance of PE classification and severity quantification was quantitatively assessed using an area under the curve (AUC) analysis for PE classification and a coefficient of determination (R²) for the Qanadli score and the RV/LV diameter ratio. RESULTS: Quantitative evaluation led to an overall AUC of 0.870 (95% confidence interval [CI]: 0.850-0.900) for PE classification task on the training set and an AUC of 0.852 (95% CI: 0.810-0.890) on the test set. Regression analysis yielded R² value of 0.717 (95% CI: 0.668-0.760) and of 0.723 (95% CI: 0.668-0.766) for the Qanadli score and the RV/LV diameter ratio estimation, respectively on the test set. CONCLUSION: This study shows the feasibility of utilizing AI-based assistance tools in detecting blood clots and estimating PE severity scores with 3D CTPA examinations. This is achieved by leveraging blood clots and cardiac segmentations. Further studies are needed to assess the effectiveness of these tools in clinical practice.

BERT-based natural language processing analysis of French CT reports: Application to the measurement of the positivity rate for pulmonary embolism.

Rationale and objectives: To develop a Natural Language Processing (NLP) method based on Bidirectional Encoder Representations from Transformers (BERT) adapted to French CT reports and to evaluate its performance to calculate the diagnostic yield of CT in patients with clinical suspicion of pulmonary embolism (PE). Materials and methods: All the CT reports performed in our institution in 2019 (99,510 reports, training and validation dataset) and 2018 (94,559 reports, testing dataset) were included after anonymization. Two BERT-based NLP sentence classifiers were trained on 27.700, manually labeled, sentences from the training dataset. The first one aimed to classify the reports' sentences into three classes ("Non chest", "Healthy chest", and "Pathological chest" related sentences), the second one to classify the last class into eleven sub classes pathologies including "pulmonary embolism". F1-score was reported on the validation dataset. These NLP classifiers were then applied to requested CT reports for pulmonary embolism from the testing dataset. Sensitivity, specificity, and accuracy for detection of the presence of a pulmonary embolism were reported in comparison to human analysis of the reports. Results: The F1-score for the 3-Classes and 11-SubClasses classifiers was 0.984 and 0.985, respectively. 4,042 examinations from the testing dataset were requested for pulmonary embolism of which 641 (15.8%) were positively evaluated by radiologists. The sensitivity, specificity, and accuracy of the NLP network for identifying pulmonary embolism in these reports were 98.2%, 99.3% and 99.1%, respectively. Conclusion: BERT-based NLP sentences classifier enables the analysis of large databases of radiological reports to accurately determine the diagnostic yield of CT screening.