Potential of GPT-4 for Detecting Errors in Radiology Reports: Implications for Reporting Accuracy.

Background Errors in radiology reports may occur because of resident-to-attending discrepancies, speech recognition inaccuracies, and large workload. Large language models, such as GPT-4 (ChatGPT; OpenAI), may assist in generating reports. Purpose To assess effectiveness of GPT-4 in identifying common errors in radiology reports, focusing on performance, time, and cost-efficiency. Materials and Methods In this retrospective study, 200 radiology reports (radiography and cross-sectional imaging [CT and MRI]) were compiled between June 2023 and December 2023 at one institution. There were 150 errors from five common error categories (omission, insertion, spelling, side confusion, and other) intentionally inserted into 100 of the reports and used as the reference standard. Six radiologists (two senior radiologists, two attending physicians, and two residents) and GPT-4 were tasked with detecting these errors. Overall error detection performance, error detection in the five error categories, and reading time were assessed using Wald χ2 tests and paired-sample t tests. Results GPT-4 (detection rate, 82.7%;124 of 150; 95% CI: 75.8, 87.9) matched the average detection performance of radiologists independent of their experience (senior radiologists, 89.3% [134 of 150; 95% CI: 83.4, 93.3]; attending physicians, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; residents, 80.0% [120 of 150; 95% CI: 72.9, 85.6]; P value range, .522-.99). One senior radiologist outperformed GPT-4 (detection rate, 94.7%; 142 of 150; 95% CI: 89.8, 97.3; P = .006). GPT-4 required less processing time per radiology report than the fastest human reader in the study (mean reading time, 3.5 seconds ± 0.5 [SD] vs 25.1 seconds ± 20.1, respectively; P < .001; Cohen d = -1.08). The use of GPT-4 resulted in lower mean correction cost per report than the most cost-efficient radiologist ($0.03 ± 0.01 vs $0.42 ± 0.41; P < .001; Cohen d = -1.12). Conclusion The radiology report error detection rate of GPT-4 was comparable with that of radiologists, potentially reducing work hours and cost. © RSNA, 2024 See also the editorial by Forman in this issue.

AI support for accurate and fast radiological diagnosis of COVID-19: an international multicenter, multivendor CT study.

OBJECTIVES: Differentiation between COVID-19 and community-acquired pneumonia (CAP) in computed tomography (CT) is a task that can be performed by human radiologists and artificial intelligence (AI). The present study aims to (1) develop an AI algorithm for differentiating COVID-19 from CAP and (2) evaluate its performance. (3) Evaluate the benefit of using the AI result as assistance for radiological diagnosis and the impact on relevant parameters such as accuracy of the diagnosis, diagnostic time, and confidence. METHODS: We included n = 1591 multicenter, multivendor chest CT scans and divided them into AI training and validation datasets to develop an AI algorithm (n = 991 CT scans; n = 462 COVID-19, and n = 529 CAP) from three centers in China. An independent Chinese and German test dataset of n = 600 CT scans from six centers (COVID-19 / CAP; n = 300 each) was used to test the performance of eight blinded radiologists and the AI algorithm. A subtest dataset (180 CT scans; n = 90 each) was used to evaluate the radiologists' performance without and with AI assistance to quantify changes in diagnostic accuracy, reporting time, and diagnostic confidence. RESULTS: The diagnostic accuracy of the AI algorithm in the Chinese-German test dataset was 76.5%. Without AI assistance, the eight radiologists' diagnostic accuracy was 79.1% and increased with AI assistance to 81.5%, going along with significantly shorter decision times and higher confidence scores. CONCLUSION: This large multicenter study demonstrates that AI assistance in CT-based differentiation of COVID-19 and CAP increases radiological performance with higher accuracy and specificity, faster diagnostic time, and improved diagnostic confidence. KEY POINTS: • AI can help radiologists to get higher diagnostic accuracy, make faster decisions, and improve diagnostic confidence. • The China-German multicenter study demonstrates the advantages of a human-machine interaction using AI in clinical radiology for diagnostic differentiation between COVID-19 and CAP in CT scans.

Correlation of head screw lengths in proximal humerus nailing: a CT-based study on 289 cases.

INTRODUCTION: Nailing of the proximal humerus is an established method for the treatment of proximal humerus fractures. Choice of the correct length for potentially four proximal locking screws is essential for postoperative outcome. Due to positioning of the patient, intraoperative determination of the correct length of the anteroposterior (AP) screw with the x-ray beam is particularly challenging even for experienced surgeons. We hypothesized that there would be a correlation between the projected lengths of the different proximal locking screws and therefore the length of the AP-screw could be determined based on the three lateromedial (LM) screws. MATERIALS AND METHODS: In this retrospective study (level of evidence: III) CT-scans of shoulders of 289 patients were 3D reconstructed with the program Horos. Using the manufacturer Stryker's instructions, the four proximal locking screws of the T2 Humeral Nail system were reproduced in the 3D reconstructed shoulders. The length of the AP-screw was correlated with the lengths of the LM-screws by Linear Regression and Multiple Linear Regression. RESULTS: The results of this study showed that the lengths of proximal locking screws in proximal humeral nailing correlated significantly with each other. Based on the given data, a formula could be established to calculate the length of the AP-screw based on the lengths of the LM-screws with a probability of 76.5%. CONCLUSIONS: This study was able to show that the length of the AP-screw could be determined from the intraoperatively measured lengths of the LM-screws. As our findings base on measurements performed in CT scans, clinical studies are needed to support our data.

Feasibility of artificial intelligence-supported assessment of bone marrow infiltration using dual-energy computed tomography in patients with evidence of monoclonal protein - a retrospective observational study.

OBJECTIVES: To demonstrate the feasibility of an automated, non-invasive approach to estimate bone marrow (BM) infiltration of multiple myeloma (MM) by dual-energy computed tomography (DECT) after virtual non-calcium (VNCa) post-processing. METHODS: Individuals with MM and monoclonal gammopathy of unknown significance (MGUS) with concurrent DECT and BM biopsy between May 2018 and July 2020 were included in this retrospective observational study. Two pathologists and three radiologists reported BM infiltration and presence of osteolytic bone lesions, respectively. Bone mineral density (BMD) was quantified CT-based by a CE-certified software. Automated spine segmentation was implemented by a pre-trained convolutional neural network. The non-fatty portion of BM was defined as voxels > 0 HU in VNCa. For statistical assessment, multivariate regression and receiver operating characteristic (ROC) were conducted. RESULTS: Thirty-five patients (mean age 65 ± 12 years; 18 female) were evaluated. The non-fatty portion of BM significantly predicted BM infiltration after adjusting for the covariable BMD (p = 0.007, r = 0.46). A non-fatty portion of BM > 0.93% could anticipate osteolytic lesions and the clinical diagnosis of MM with an area under the ROC curve of 0.70 [0.49-0.90] and 0.71 [0.54-0.89], respectively. Our approach identified MM-patients without osteolytic lesions on conventional CT with a sensitivity and specificity of 0.63 and 0.71, respectively. CONCLUSIONS: Automated, AI-supported attenuation assessment of the spine in DECT VNCa is feasible to predict BM infiltration in MM. Further, the proposed method might allow for pre-selecting patients with higher pre-test probability of osteolytic bone lesions and support the clinical diagnosis of MM without pathognomonic lesions on conventional CT. KEY POINTS: • The retrospective study provides an automated approach for quantification of the non-fatty portion of bone marrow, based on AI-supported spine segmentation and virtual non-calcium dual-energy CT data. • An increasing non-fatty portion of bone marrow is associated with a higher infiltration determined by invasive biopsy after adjusting for bone mineral density as a control variable (p = 0.007, r = 0.46). • The non-fatty portion of bone marrow might support the clinical diagnosis of multiple myeloma when conventional CT images are negative (sensitivity 0.63, specificity 0.71).

The worldwide COVID-19 pandemic caused a decline in sonographic examinations - is this a continuing trend?

PURPOSE: Due to the increasing number of COVID-19 infections since spring 2020 the patient care workflow underwent changes in Germany. To minimize face-to-face exposure and reduce infection risk, non-time-critical elective medical procedures were postponed. Since ultrasound examinations include non-time-critical elective examinations and often can be substituted by other imaging modalities not requiring direct patient contact, the number of examinations has declined significantly. The aim of this study is to quantify the baseline number of ultrasound examinations in the years before, during, and in the early post-pandemic period of the COVID-19 pandemic (since January 2015 to September 2023), and to measure the number of examinations at different German university hospitals. MATERIALS AND METHODS: The number of examinations was assessed based on a web-based database at all participating clinics at the indicated time points. RESULTS: N = 288 562 sonographic examinations from four sites were included in the present investigation. From January 2020 to June 2020, a significantly lower number of examinations of n = 591.21 vs. 698.43 (p = 0.01) per month and included center was performed. Also, excluding the initial pandemic period until June 2020, significantly fewer ultrasound examinations were performed compared to pre-pandemic years 648.1 vs. 698.4 (p < 0.05), per month and included center, while here differences between the individual centers were observed. In the late phase of the pandemic (n = 681.96) and in the post-pandemic phase (as defined by the WHO criteria from May 2023; n = 739.95), the number of sonographic examinations returned to pre-pandemic levels. CONCLUSION: The decline in the number of sonographic examinations caused by the COVID-19 pandemic was initially largely intentional and can be illustrated quantitatively. After an initial abrupt decline in sonographic examinations, the pre-pandemic levels could not be reached for a long time, which could be due to restructuring of patient care and follow-up treatment. In the post-pandemic phase, the pre-pandemic level has been achieved again. The reasons for a prolonged reduction in ultrasound examinations are discussed in this article. KEY POINTS: · During the pandemic, significantly fewer ultrasound examinations were performed in the included centers.. · The number of examinations could not be reach the pre-pandemic level for a long time, which could be due to restructuring of patient care and follow-up treatment.. · Identifying causes for sonographic exam reduction is crucial in pandemic preparedness to uphold healthcare quality and continuity for all patients.. · The prolonged decline in sonographic examinations during the pandemic does not represent a lasting trend, as evidenced by the return to pre-pandemic levels..

Cooperative AI training for cardiothoracic segmentation in computed tomography: An iterative multi-center annotation approach.

PURPOSE: Radiological reporting is transitioning to quantitative analysis, requiring large-scale multi-center validation of biomarkers. A major prerequisite and bottleneck for this task is the voxelwise annotation of image data, which is time-consuming for large cohorts. In this study, we propose an iterative training workflow to support and facilitate such segmentation tasks, specifically for high-resolution thoracic CT data. METHODS: Our study included 132 thoracic CT scans from clinical practice, annotated by 13 radiologists. In three iterative training experiments, we aimed to improve and accelerate segmentation of the heart and mediastinum. Each experiment started with manual segmentation of 5-25 CT scans, which served as training data for a nnU-Net. Further iterations incorporated AI pre-segmentation and human correction to improve accuracy, accelerate the annotation process, and reduce human involvement over time. RESULTS: Results showed consistent improvement in AI model quality with each iteration. Resampled datasets improved the Dice similarity coefficients for both the heart (DCS 0.91 [0.88; 0.92]) and the mediastinum (DCS 0.95 [0.94; 0.95]). Our AI models reduced human interaction time by 50 % for heart and 70 % for mediastinum segmentation in the most potent iteration. A model trained on only five datasets achieved satisfactory results (DCS > 0.90). CONCLUSIONS: The iterative training workflow provides an efficient method for training AI-based segmentation models in multi-center studies, improving accuracy over time and simultaneously reducing human intervention. Future work will explore the use of fewer initial datasets and additional pre-processing methods to enhance model quality.

Conventional and Deep-Learning-Based Image Reconstructions of Undersampled K-Space Data of the Lumbar Spine Using Compressed Sensing in MRI: A Comparative Study on 20 Subjects.

Compressed sensing accelerates magnetic resonance imaging (MRI) acquisition by undersampling of the k-space. Yet, excessive undersampling impairs image quality when using conventional reconstruction techniques. Deep-learning-based reconstruction methods might allow for stronger undersampling and thus faster MRI scans without loss of crucial image quality. We compared imaging approaches using parallel imaging (SENSE), a combination of parallel imaging and compressed sensing (COMPRESSED SENSE, CS), and a combination of CS and a deep-learning-based reconstruction (CS AI) on raw k-space data acquired at different undersampling factors. 3D T2-weighted images of the lumbar spine were obtained from 20 volunteers, including a 3D sequence (standard SENSE), as provided by the manufacturer, as well as accelerated 3D sequences (undersampling factors 4.5, 8, and 11) reconstructed with CS and CS AI. Subjective rating was performed using a 5-point Likert scale to evaluate anatomical structures and overall image impression. Objective rating was performed using apparent signal-to-noise and contrast-to-noise ratio (aSNR and aCNR) as well as root mean square error (RMSE) and structural-similarity index (SSIM). The CS AI 4.5 sequence was subjectively rated better than the standard in several categories and deep-learning-based reconstructions were subjectively rated better than conventional reconstructions in several categories for acceleration factors 8 and 11. In the objective rating, only aSNR of the bone showed a significant tendency towards better results of the deep-learning-based reconstructions. We conclude that CS in combination with deep-learning-based image reconstruction allows for stronger undersampling of k-space data without loss of image quality, and thus has potential for further scan time reduction.

Association between pouch morphology, cardiovascular risk factors and ischemic brain lesions in patients with left-sided septal pouches.

BACKGROUND: Left atrial outpouching structures such as left atrial diverticula (LADs) and left-sided septal pouches (LSSPs) might be a source of cryptogenic stroke. This imaging study evaluates the association between pouch morphology, patient comorbidities and ischemic brain lesions (IBLs). METHODS: This is a retrospective single-center analysis of 195 patients who received both a cardiac CT and a cerebral MRI. LADs, LSSPs, and IBLs were retrospectively identified. Size measurements included pouch width, length and volume for LADs and circumference, area and volume for LSSPs. The association between LADs/LSSPs, IBLs and cardiovascular comorbidities was determined by univariate and bivariate regression analyses. RESULTS: The prevalence and mean volume were 36.4% and 372 ± 569 mm3 for LSSPs, and 40.5% and 415 ± 541 mm3 for LADs. The IBL prevalence was 67.6% in the LSSP group and 48.1% in the LAD group. LSSPs had 2.9-fold increased hazards of IBLs (95%CI: 1.2-7.4, p = 0.024), and LADs showed no significant correlation with IBLs. Size measurements had no impact on IBLs. A co-existing LSSP was associated with an increased prevalence of IBLs in patients with coronary artery disease (HR: 1.5, 95%CI: 1.1-1.9, p = 0.048), heart failure (HR: 3.7, 95%CI: 1.1-14.6, p = 0.032), arterial hypertension (HR: 1.9, 95%CI: 1.1-3.3, p = 0.017), and hyperlipidemia (HR: 2.2, 95%CI: 1.1-4.4, p = 0.018). CONCLUSION: Co-existing LSSPs were associated with IBLs in patients with cardiovascular risk factors, however, pouch morphology did not correlate with the IBL rate. Upon confirmation by further studies, these findings might be considered in the treatment, risk stratification, and stroke prophylaxis of these patients.

Physiological iodine uptake of the spine's bone marrow in dual-energy computed tomography - using artificial intelligence to define reference values based on 678 CT examinations of 189 individuals.

Purpose: The bone marrow's iodine uptake in dual-energy CT (DECT) is elevated in malignant disease. We aimed to investigate the physiological range of bone marrow iodine uptake after intravenous contrast application, and examine its dependence on vBMD, iodine blood pool, patient age, and sex. Method: Retrospective analysis of oncological patients without evidence of metastatic disease. DECT examinations were performed on a spectral detector CT scanner in portal venous contrast phase. The thoracic and lumbar spine were segmented by a pre-trained neural network, obtaining volumetric iodine concentration data [mg/ml]. vBMD was assessed using a phantomless, CE-certified software [mg/cm3]. The iodine blood pool was estimated by ROI-based measurements in the great abdominal vessels. A multivariate regression model was fit with the dependent variable "median bone marrow iodine uptake". Standardized regression coefficients (ß) were calculated to assess the impact of each covariate. Results: 678 consecutive DECT exams of 189 individuals (93 female, age 61.4 ± 16.0 years) were evaluated. AI-based segmentation provided volumetric data of 97.9% of the included vertebrae (n=11,286). The 95th percentile of bone marrow iodine uptake, as a surrogate for the upper margin of the physiological distribution, ranged between 4.7-6.4 mg/ml. vBMD (p <0.001, mean ß=0.50) and portal vein iodine blood pool (p <0.001, mean ß=0.43) mediated the strongest impact. Based thereon, adjusted reference values were calculated. Conclusion: The bone marrow iodine uptake demonstrates a distinct profile depending on vBMD, iodine blood pool, patient age, and sex. This study is the first to provide the adjusted reference values.

Morphological appearance of the B.1.1.7 mutation of the novel coronavirus 2 (SARS-CoV-2) in chest CT.

Background: Diagnosing a coronavirus disease 2019 (COVID-19) infection with high specificity in chest computed tomography (CT) imaging is considered possible due to distinctive imaging features of COVID-19 pneumonia. Since other viral non-COVID pneumonia show mostly a different distribution pattern, it is reasonable to assume that the patterns observed caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are a consequence of its genetically encoded molecular properties when interacting with the respiratory tissue. As more mutations of the initial SARS-CoV-2 wild-type with varying aggressiveness have been detected in the course of 2021, it became obvious that its genome is in a state of transformation and therefore a potential modification of the specific morphological appearance in CT may occur. The aim of this study was to quantitatively analyze the morphological differences of the SARS-CoV-2-B.1.1.7 mutation and wildtype variant in CT scans of the thorax. Methods: We analyzed a dataset of 140 patients, which was divided into pneumonias caused by n=40 wildtype variants, n=40 B.1.1.7 variants, n=20 bacterial pneumonias, n=20 viral (non-COVID) pneumonias, and a test group of n=20 unremarkable CT examinations of the thorax. Semiautomated 3D segmentation of the lung tissue was performed for quantification of lung pathologies. The extent, ratio, and specific distribution of inflammatory affected lung tissue in each group were compared in a multivariate group analysis. Results: Lung segmentation revealed significant difference between the extent of ground glass opacities (GGO) or consolidation comparing SARS-CoV-2 wild-type and B.1.1.7 variant. Wildtype and B.1.1.7 variant showed both a symmetric distribution pattern of stage-dependent GGO and consolidation within matched COVID-19 stages. Viral non-COVID pneumonias had significantly fewer consolidations than the bacterial, but also than the COVID-19 B.1.1.7 variant groups. Conclusions: CT based segmentation showed no significant difference between the morphological appearance of the COVID-19 wild-type variant and the SARS-CoV-2 B.1.1.7 mutation. However, our approach allowed a semiautomatic quantification of bacterial and viral lung pathologies. Quantitative CT image analyses, such as the one presented, appear to be an important component of pandemic preparedness considering an organism with ongoing genetic change, to describe a potential arising change in CT morphological appearance of possible new upcoming COVID-19 variants of concern.

Quantitative determination of pulmonary emphysema in follow-up LD-CTs of patients with COVID-19 infection.

PURPOSE: To evaluate the association between the coronavirus disease 2019 (COVID-19) and post-inflammatory emphysematous lung alterations on follow-up low-dose CT scans. METHODS: Consecutive patients with proven COVID-19 infection and a follow-up CT were retrospectively reviewed. The severity of pulmonary involvement was classified as mild, moderate and severe. Total lung volume, emphysema volume and the ratio of emphysema/-to-lung volume were quantified semi-automatically and compared inter-individually between initial and follow-up CT and to a control group of healthy, age- and sex-matched patients. Lung density was further assessed by drawing circular regions of interest (ROIs) into non-affected regions of the upper lobes. RESULTS: A total of 32 individuals (mean age: 64 ± 13 years, 12 females) with at least one follow-up CT (mean: 52 ± 66 days, range: 5-259) were included. In the overall cohort, total lung volume, emphysema volume and the ratio of lung-to-emphysema volume did not differ significantly between the initial and follow-up scans. In the subgroup of COVID-19 patients with > 30 days of follow-up, the emphysema volume was significantly larger as compared to the subgroup with a follow-up < 30 days (p = 0.045). Manually measured single ROIs generally yielded lower attenuation values prior to COVID-19 pneumonia, but the difference was not significant between groups (all p > 0.05). CONCLUSION: COVID-19 patients with a follow-up CT >30 days showed significant emphysematous lung alterations. These findings may help to explain the long-term effect of COVID-19 on pulmonary function and warrant validation by further studies.

Meaningful use of imaging resources to rule out cerebral venous sinus thrombosis after ChAdOx1 COVID-19 vaccination: Evaluation of the AHA diagnostic algorithm with a clinical cohort and a systematic data review.

Vaccine-induced immune thrombotic thrombocytopenia (VITT) with cerebral venous thrombosis (CVST) is an improbable (0.0005%), however potentially lethal complication after ChAdOx1 vaccination. On the other hand, headache is among the most frequent side effects of ChAdOx1 (29.3%). In September 2021, the American Heart Association (AHA) suggested a diagnostic workflow to facilitate risk-adapted use of imaging resources for patients with neurological symptoms after ChAdOx1. We aimed to evaluate the AHA workflow in a retrospective patient cohort presenting at four primary care hospitals in Germany for neurological complaints after ChAdOx1. Scientific literature was screened for case reports of VITT with CVST after ChAdOx1, published until September 1st, 2021. One-hundred-thirteen consecutive patients (77 female, mean age 38.7 +/- 11.9 years) were evaluated at our institutes, including one case of VITT with CVST. Further 228 case reports of VITT with CVST are published in recent literature, which share thrombocytopenia (225/227 reported) and elevated d-dimer levels (100/101 reported). The AHA workflow would have recognized all VITT cases with CVST (100% sensitivity), the number needed to diagnose (NND) was 1:113. Initial evaluation of thrombocytopenia or elevated d-dimer levels would have lowered the NND to 1:68, without cost of sensitivity. Hence, we suggest that in case of normal thrombocyte and d-dimer levels, the access to further diagnostics should be limited by the established clinical considerations regardless of vaccination history.

Dual-Energy CT, Virtual Non-Calcium Bone Marrow Imaging of the Spine: An AI-Assisted, Volumetric Evaluation of a Reference Cohort with 500 CT Scans.

Virtual non-calcium (VNCa) images from dual-energy computed tomography (DECT) have shown high potential to diagnose bone marrow disease of the spine, which is frequently disguised by dense trabecular bone on conventional CT. In this study, we aimed to define reference values for VNCa bone marrow images of the spine in a large-scale cohort of healthy individuals. DECT was performed after resection of a malignant skin tumor without evidence of metastatic disease. Image analysis was fully automated and did not require specific user interaction. The thoracolumbar spine was segmented by a pretrained convolutional neuronal network. Volumetric VNCa data of the spine's bone marrow space were processed using the maximum, medium, and low calcium suppression indices. Histograms of VNCa attenuation were created for each exam and suppression setting. We included 500 exams of 168 individuals (88 female, patient age 61.0 ± 15.9). A total of 8298 vertebrae were segmented. The attenuation histograms' overlap of two consecutive exams, as a measure for intraindividual consistency, yielded a median of 0.93 (IQR: 0.88-0.96). As our main result, we provide the age- and sex-specific bone marrow attenuation profiles of a large-scale cohort of individuals with healthy trabecular bone structure as a reference for future studies. We conclude that artificial-intelligence-supported, fully automated volumetric assessment is an intraindividually robust method to image the spine's bone marrow using VNCa data from DECT.

Assessment of COVID-19 lung involvement on computed tomography by deep-learning-, threshold-, and human reader-based approaches-an international, multi-center comparative study.

Background: The extent of lung involvement in coronavirus disease 2019 (COVID-19) pneumonia, quantified on computed tomography (CT), is an established biomarker for prognosis and guides clinical decision-making. The clinical standard is semi-quantitative scoring of lung involvement by an experienced reader. We aim to compare the performance of automated deep-learning- and threshold-based methods to the manual semi-quantitative lung scoring. Further, we aim to investigate an optimal threshold for quantification of involved lung in COVID pneumonia chest CT, using a multi-center dataset. Methods: In total 250 patients were included, 50 consecutive patients with RT-PCR confirmed COVID-19 from our local institutional database, and another 200 patients from four international datasets (n=50 each). Lung involvement was scored semi-quantitatively by three experienced radiologists according to the established chest CT score (CCS) ranging from 0-25. Inter-rater reliability was reported by the intraclass correlation coefficient (ICC). Deep-learning-based segmentation of ground-glass and consolidation was obtained by CT Pulmo Auto Results prototype plugin on IntelliSpace Discovery (Philips Healthcare, The Netherlands). Threshold-based segmentation of involved lung was implemented using an open-source tool for whole-lung segmentation under the presence of severe pathologies (R231CovidWeb, Hofmanninger et al., 2020) and consecutive quantitative assessment of lung attenuation. The best threshold was investigated by training and testing a linear regression of deep-learning and threshold-based results in a five-fold cross validation strategy. Results: Median CCS among 250 evaluated patients was 10 [6-15]. Inter-rater reliability of the CCS was excellent [ICC 0.97 (0.97-0.98)]. Best attenuation threshold for identification of involved lung was -522 HU. While the relationship of deep-learning- and threshold-based quantification was linear and strong (r2 deep-learning vs. threshold=0.84), both automated quantification methods translated to the semi-quantitative CCS in a non-linear fashion, with an increasing slope towards higher CCS (r2 deep-learning vs. CCS= 0.80, r2 threshold vs. CCS=0.63). Conclusions: The manual semi-quantitative CCS underestimates the extent of COVID pneumonia in higher score ranges, which limits its clinical usefulness in cases of severe disease. Clinical implementation of fully automated methods, such as deep-learning or threshold-based approaches (best threshold in our multi-center dataset: -522 HU), might save time of trained personnel, abolish inter-reader variability, and allow for truly quantitative, linear assessment of COVID lung involvement.

Virtual calcium-suppression in dual energy computed tomography predicts metabolic activity of focal MM lesions as determined by fluorodeoxyglucose positron-emission-tomography.

PURPOSE: Recent studies showed that dual energy CT (DECT) allows for detection of bone marrow infiltration in multiple myeloma (MM) by obtaining virtual non-calcium (VNCa) images. This feasibility study investigated, if VNCa imaging might discriminate metabolically active, focal lesions in MM against avital lesions in MM patients, considering fluorodeoxyglucose positron-emission-tomography CT (FDG PET/CT) as the standard of reference. METHOD: The study included 60 osteolytic lesions in 10 consecutive low-dose whole body CT scans of patients with MM, who underwent both FDG PET/CT and DECT at a tertiary care university hospital. Circular ROI measurements were performed in predefined lesions on the monoenergetic CT (MECT) and VNCa images by three blinded radiologists. Each lesion was rated vital or avital by a blinded specialist of nuclear medicine, based on their FDG metabolism. RESULTS: Each of the three readers could separate FDG PET/CT negative and positive MM lesions when analyzing the VNCa images, while MECT did not show a significant difference. Best results were yielded by high calcium suppression with excellent inter-rater reliability (average sensitivity 0.91, specificity 0.88, cutoff -46.9 HU), followed by medium and low calcium suppression. CONCLUSIONS: In contrast to MECT imaging, VNCa imaging in DECT appears to be feasible to assess metabolic activity of focal MM lesions as defined by the standard of reference, FDG PET/CT. Considering the higher cost and radiation exposure of FDG PET/CT, DECT VNCa imaging might develop to be the modality of choice to assess metabolic activity of focal MM lesions.

Radiotherapy Response Assessment of Multiple Myeloma: A Dual-Energy CT Approach With Virtual Non-Calcium Images.

BACKGROUND: Life expectancy of patients with multiple myeloma (MM) has increased over the past decades, underlining the importance of local tumor control and avoidance of dose-dependent side effects of palliative radiotherapy (RT). Virtual noncalcium (VNCa) imaging from dual-energy computed tomography (DECT) has been suggested to estimate cellularity and metabolic activity of lytic bone lesions (LBLs) in MM. OBJECTIVE: To explore the feasibility of RT response monitoring with DECT-derived VNCa attenuation measurements in MM. METHODS: Thirty-three patients with 85 LBLs that had been irradiated and 85 paired non-irradiated LBLs from the same patients were included in this retrospective study. Irradiated and non-irradiated LBLs were measured by circular regions of interest (ROIs) on conventional and VNCa images in a total of 216 follow-up measurements (48 before and 168 after RT). Follow-ups were rated as therapy response, stable disease, or local progression according to the MD Anderson criteria. Receiver operating characteristic (ROC) analysis was performed to discriminate irradiated vs. non-irradiated and locally progressive vs. stable/responsive LBLs using absolute attenuation post-irradiation and percentage attenuation change for patients with pre-irradiation DECT, if available. RESULTS: Attenuation of LBLs decreased after RT depending on the time that had passed after irradiation [absolute thresholds for identification of irradiated LBLs 30.5-70.0 HU [best area under the curve [AUC] 0.75 (0.59-0.91)] and -77.0 to -22.5 HU [best AUC 0.85 (0.65-1.00)]/-50% and -117% to -167% proportional change of attenuation on conventional and VNCa images, respectively]. VNCa CT was significantly superior for identification of RT effects in LBLs with higher calcium content [best VNCa AUC 0.96 (0.91-1.00), best conventional CT AUC 0.64 (0.45-0.83)]. Thresholds for early identification of local irradiation failure were >20.5 HU on conventional CT [AUC 0.78 (0.68-0.88)] and >-27 HU on VNCa CT [AUC 0.83 (0.70-0.96)]. CONCLUSION: Therapy response of LBLs after RT can be monitored by VNCa imaging based on regular myeloma scans, which yields potential for optimizing the lesion-specific radiation dose for local tumor control. Decreasing attenuation indicates RT response, while above threshold attenuation of LBLs precedes local irradiation failure.

Early extrapulmonary prognostic features in chest computed tomography in COVID-19 pneumonia: Bone mineral density is a relevant predictor for the clinical outcome - A multicenter feasibility study.

BACKGROUND: Besides throat-nose swab polymerase chain reaction (PCR), unenhanced chest computed tomography (CT) is a recommended diagnostic tool for early detection and quantification of pulmonary changes in COVID-19 pneumonia caused by the novel corona virus. Demographic factors, especially age and comorbidities, are major determinants of the outcome in COVID-19 infection. This study examines the extra pulmonary parameter of bone mineral density (BMD) from an initial chest computed tomography as an associated variable of pre-existing comorbidities like chronic lung disease or demographic factors to determine the later patient's outcome, in particular whether treatment on an intensive care unit (ICU) was necessary in infected patients. METHODS: We analyzed 58 PCR-confirmed COVID-19 infections that received an unenhanced CT at admission at one of the included centers. In addition to the extent of pulmonary involvement, we performed a phantomless assessment of bone mineral density of thoracic vertebra 9-12. RESULTS: In a univariate regression analysis BMD was found to be a significant predictor of the necessity for intensive care unit treatment of COVID-19 patients. In the subgroup requiring intensive care treatment within the follow-up period a significantly lower BMD was found. In a multivariate logistic regression model considering gender, age and CT measurements of bone mineral density, BMD was eliminated from the regression analysis as a significant predictor. CONCLUSION: Phantomless assessed BMD provides prognostic information on the necessity for ICU treatment in course of COVID-19 pneumonia. We recommend using the measurement of BMD in an initial CT image to facilitate a potentially better prediction of severe patient outcomes within the 22 days after an initial CT scan. Consequently, in the present sample, additional bone density analysis did not result in a prognostic advantage over simply considering age. Significantly larger patient cohorts with a more homogenous patient age should be performed in the future to illustrate potential effects. CLINICAL RELEVANCE: While clinical capacities such as ICU beds and ventilators are more crucial than ever to help manage the current global corona pandemic, this work introduces an approach that can be used in a cost-effective way to help determine the amount of these rare clinical resources required in the near future.

Coronary artery calcification on low-dose chest CT is an early predictor of severe progression of COVID-19-A multi-center, multi-vendor study.

PURPOSE: Cardiovascular comorbidity anticipates severe progression of COVID-19 and becomes evident by coronary artery calcification (CAC) on low-dose chest computed tomography (LDCT). The purpose of this study was to predict a patient's obligation of intensive care treatment by evaluating the coronary calcium burden on the initial diagnostic LDCT. METHODS: Eighty-nine consecutive patients with parallel LDCT and positive RT-PCR for SARS-CoV-2 were included from three centers. The primary endpoint was admission to ICU, tracheal intubation, or death in the 22-day follow-up period. CAC burden was represented by the Agatston score. Multivariate logistic regression was modeled for prediction of the primary endpoint by the independent variables "Agatston score > 0", as well as the CT lung involvement score, patient sex, age, clinical predictors of severe COVID-19 progression (history of hypertension, diabetes, prior cardiovascular event, active smoking, or hyperlipidemia), and laboratory parameters (creatinine, C-reactive protein, leucocyte, as well as thrombocyte counts, relative lymphocyte count, d-dimer, and lactate dehydrogenase levels). RESULTS: After excluding multicollinearity, "Agatston score >0" was an independent regressor within multivariate analysis for prediction of the primary endpoint (p<0.01). Further independent regressors were creatinine (p = 0.02) and leucocyte count (p = 0.04). The Agatston score was significantly higher for COVID-19 cases which completed the primary endpoint (64.2 [interquartile range 1.7-409.4] vs. 0 [interquartile range 0-0]). CONCLUSION: CAC scoring on LDCT might help to predict future obligation of intensive care treatment at the day of patient admission to the hospital.

Contrast-Enhanced Black Blood MRI Sequence Is Superior to Conventional T1 Sequence in Automated Detection of Brain Metastases by Convolutional Neural Networks.

BACKGROUND: in magnetic resonance imaging (MRI), automated detection of brain metastases with convolutional neural networks (CNN) represents an extraordinary challenge due to small lesions sometimes posing as brain vessels as well as other confounders. Literature reporting high false positive rates when using conventional contrast enhanced (CE) T1 sequences questions their usefulness in clinical routine. CE black blood (BB) sequences may overcome these limitations by suppressing contrast-enhanced structures, thus facilitating lesion detection. This study compared CNN performance in conventional CE T1 and BB sequences and tested for objective improvement of brain lesion detection. METHODS: we included a subgroup of 127 consecutive patients, receiving both CE T1 and BB sequences, referred for MRI concerning metastatic spread to the brain. A pretrained CNN was retrained with a customized monolayer classifier using either T1 or BB scans of brain lesions. RESULTS: CE T1 imaging-based training resulted in an internal validation accuracy of 85.5% vs. 92.3% in BB imaging (p < 0.01). In holdout validation analysis, T1 image-based prediction presented poor specificity and sensitivity with an AUC of 0.53 compared to 0.87 in BB-imaging-based prediction. CONCLUSIONS: detection of brain lesions with CNN, BB-MRI imaging represents a highly effective input type when compared to conventional CE T1-MRI imaging. Use of BB-MRI can overcome the current limitations for automated brain lesion detection and the objectively excellent performance of our CNN suggests routine usage of BB sequences for radiological analysis.

A reporting and analysis framework for structured evaluation of COVID-19 clinical and imaging data.

The COVID-19 pandemic has worldwide individual and socioeconomic consequences. Chest computed tomography has been found to support diagnostics and disease monitoring. A standardized approach to generate, collect, analyze, and share clinical and imaging information in the highest quality possible is urgently needed. We developed systematic, computer-assisted and context-guided electronic data capture on the FDA-approved mint LesionTM software platform to enable cloud-based data collection and real-time analysis. The acquisition and annotation include radiological findings and radiomics performed directly on primary imaging data together with information from the patient history and clinical data. As proof of concept, anonymized data of 283 patients with either suspected or confirmed SARS-CoV-2 infection from eight European medical centers were aggregated in data analysis dashboards. Aggregated data were compared to key findings of landmark research literature. This concept has been chosen for use in the national COVID-19 response of the radiological departments of all university hospitals in Germany.