Search

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., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 The Authors. Published by Elsevier B.V. All rights reserved.)

Purpose: To investigate whether minimum intensity projection (MinIP) reconstructions enable more accurate depiction of pulmonary ground-glass opacity (GGO) compared to standard transverse sections and multiplanar reformat (MPR) series in patients with suspected coronavirus disease 2019 (COVID-19)., Method: In this multinational study, chest CT scans of 185 patients were retrospectively analyzed. Diagnostic accuracy, diagnostic confidence, image quality regarding the assessment of GGO, as well as subjective time-efficiency of MinIP and standard MPR series were analyzed based on the assessment of six radiologists. In addition, the suitability for COVID-19 evaluation, image quality regarding GGO and subjective time-efficiency in clinical routine was assessed by five clinicians., Results: The reference standard revealed a total of 149 CT scans with pulmonary GGO. MinIP reconstructions yielded significantly higher sensitivity (99.9 % vs 95.6 %), specificity (95.8 % vs 86.1 %) and accuracy (99.1 % vs 93.8 %) for assessing of GGO compared with standard MPR series. MinIP reconstructions achieved significantly higher ratings by radiologists concerning diagnostic confidence (medians, 5.00 vs 4.00), image quality (medians, 4.00 vs 4.00), contrast between GGO and unaffected lung parenchyma (medians, 5.00 vs 4.00) as well as subjective time-efficiency (medians, 5.00 vs 4.00) compared with MPR-series (all P < .001). Clinicians preferred MinIP reconstructions for COVID-19 assessment (medians, 5.00 vs 3.00), image quality regarding GGO (medians, 5.00 vs 3.00) and subjective time-efficiency in clinical routine (medians, 5.00 vs 3.00)., Conclusions: MinIP reconstructions improve the assessment of COVID-19 in chest CT compared to standard images and may be suitable for routine application., (Copyright © 2020 Elsevier B.V. All rights reserved.)

Purpose: Dedicated post-processing of dual-energy computed tomography angiography (DE-CTA) datasets has been shown to allow for increased vascular contrast. The goal of our study was to define optimal window settings for displaying virtual monoenergetic images (VMI) reconstructed from dual-energy carotid and cerebrovascular DE-CTA., Methods: Fifty-seven patients who underwent clinically-indicated carotid and cerebrovascular third-generation dual-source DE-CTA were retrospectively evaluated. Standard linearly-blended (M&#95;0.6), 70-keV traditional VMI (M70), and 40-keV noise-optimized VMI (M40+) reconstructions were analyzed. For M70 and M40+ datasets, the subjectively best window setting (width and level, B-W/L) was independently determined by two observers and subsequently related with aortic arch attenuation to calculate optimized values (O-W/L) using linear regression. Subjective evaluation of image quality (IQ) between W/L settings were assessed by two additional readers. Repeated measures analysis of variance were performed to compare W/L settings and IQ indices between M&#95;0.6, M70, and M40 + ., Results: B-W/L and O-W/L for M70 were 580/210 and 560/200, and for M40+ were 1630/570 and 1560/550, respectively, higher than standard DE-CTA W/L settings (450/100). Highest subjective scores were observed for M40+ regarding overall IQ (all p < 0.001)., Conclusion: Application of O-W/L settings is mandatory to optimize subjective IQ of VMI reconstructions of DE-CTA., (Copyright © 2020 Elsevier B.V. All rights reserved.)

Purpose: To evaluate the diagnostic accuracy of color-coded dual-energy CT virtual noncalcium (VNCa) reconstructions for the assessment of traumatic bone marrow edema in sacral insufficiency fracture (SIF)., Method: Data from 52 consecutive patients (28 women, 24 men; mean age, 61 ± 13 years; range, 49-94 years) who had undergone third-generation dual-source CT and 3-Tesla (T) MRI due to low back pain without adequate trauma were retrospectively evaluated. Five radiologists, blinded to MRI and clinical information, independently analyzed conventional grayscale dual-energy CT series for sacral fractures according to the Denis classification. Eight weeks later, readers re-assessed all scans using color-coded VNCa reconstructions for sacral bone marrow edema. CT numbers on VNCa reconstructions were measured by a sixth radiologist. One experienced radiologist (33 years of experience in musculoskeletal [MSK] imaging), blinded to CT and clinical information, defined the reference standard by analyzing the MRI scans. The primary indices for diagnostic accuracy were sensitivity, specificity, and the area under the curve (AUC)., Results: MRI revealed a total of 39 zones with SIF-associated bone marrow edema in 27 patients. In the qualitative analysis, VNCa showed high overall sensitivity (93 %) and specificity (95 %) for assessing SIF-associated bone marrow edema. The quantitative analysis of color-coded VNCa reconstructions revealed an overall AUC of 0.976. A cut-off value of -43 Hounsfield units provided a sensitivity of 85 % and a specificity of 95 % for differentiating bone marrow edema., Conclusions: Color-coded dual-energy CT VNCa reconstructions yield excellent diagnostic accuracy in the analysis of SIF-associated bone marrow edema compared to MRI., Competing Interests: Declaration of Competing Interest The other authors have no potential conflict of interest to disclose., (Copyright © 2020 Elsevier B.V. All rights reserved.)

Purpose: To evaluate the impact of controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA) volume interpolated breath-hold examination (VIBE) magnetic resonance imaging (MRI) technique on image quality, reader confidence, and inter-observer agreement for the assessment of focal liver lesions in comparison with the standard VIBE approach., Material and Methods: In this IRB-approved intra-individual comparison study, abdominal arterial and portal-venous contrast-enhanced MRI studies were retrospectively analyzed in 38 patients with malignant liver lesions. Each patient underwent both CAIPIRINHA and conventional VIBE 3T MRI within 3 months, showing stable disease. Images were evaluated using 5-point rating scales by two blinded radiologists with more than 20 and 5 years of experience in MRI, respectively. Readers scored dignity of liver lesions and assessed which liver segments were affected by malignancy (ranging from 1=definitely benign/not affected to 5=definitely malignant/affected by malignancy). Readers also rated overall image quality, sharpness of intrahepatic veins, and diagnostic confidence (ranging from 1=poor to 5=excellent)., Results: Reviewers achieved a higher inter-observer reliability using CAIPIRINHA when they reported which liver segments were affected by malignancy compared to traditional VIBE series (κ=0.62 and 0.54, respectively, p<0.05). Similarly, CAIPIRINHA showed a slightly higher inter-rater agreement for the dignity of focal liver lesions versus the standard VIBE images (κ=0.50 and 0.49, respectively, p<0.05). CAIPIRINHA series also scored higher in comparison to standard VIBE sequences (mean scores: image quality, 4.2 and 3.5; sharpness of intrahepatic vessels, 3.8 and 3.2, respectively, p<0.05) for both reviewers and allowed for higher subjective diagnostic confidence (ratings, 3.8 and 3.2, respectively, p<0.05)., Conclusion: Compared to the standard VIBE approach, CAIPIRINHA VIBE technique provides improved image quality and sharpness of intrahepatic veins, as well as higher diagnostic confidence. Additionally, this technique allows for higher inter-observer agreement when reporting focal liver lesions for both dignity and allocation., (Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.)

Objectives: To systematically analyze the influence of technical parameters on quantification of epicardial fat volume (EATV) at cardiac CT., Methods: 153 routine cardiac CT data sets were analyzed using three-dimensional pericardial border delineation. Three image series were reconstructed per patient: (a) CTAD: coronary CT angiography (CTA), diastolic phase; (b) CTAS: coronary CTA, systolic phase; (c) CaScD: non-contrast CT, diastolic phase. EATV was calculated using three different upper thresholds (-15HU, -30 HU, -45HU). Repeated measures ANOVA, Spearman's rho, and Bland Altman plots were used., Results: Mean EATV differed between all three image series at a -30HU threshold (CTAD 87.2 ± 38.5 ml, CTAS 90.9 ± 37.7 ml, CaScD 130.7 ± 49.5 ml, P<0.001). EATV of diastolic and systolic CTA reconstructions did not differ significantly (P=0.225). Mean EATV for contrast enhanced CTA at a -15HU threshold (CTAD15 102.4 ± 43.6 ml, CTAS15 105.3 ± 42.3 ml) could be approximated most closely by non-contrast CT at -45HU threshold (CaScD45 105.3 ± 40.8 ml). The correlation was excellent: CTAS15-CTAD15, rho=0.943; CTAD15-CaScD45, rho=0.905; CTAS15-CaScD45, rho=0.924; each P<0.001). Bias values from Bland Altman Analysis were: CTAS15-CTAD15, 4.9%; CTAD15-CaScD45, -4.3%; CTAS15-CaScD45, 0.6%., Conclusions: Measured EATV can differ substantially between contrast enhanced and non-contrast CT studies, which can be reconciled by threshold modification. Heart cycle phase does not significantly influence EATV measurements., (Copyright © 2015 Elsevier Ireland Ltd. All rights reserved.)