Search

Purpose: To investigate the potential of combining Compressed Sensing (CS) and a newly developed AI-based super resolution reconstruction prototype consisting of a series of convolutional neural networks (CNN) for a complete five-minute 2D knee MRI protocol., Methods: In this prospective study, 20 volunteers were examined using a 3T-MRI-scanner (Ingenia Elition X, Philips). Similar to clinical practice, the protocol consists of a fat-saturated 2D-proton-density-sequence in coronal, sagittal and transversal orientation as well as a sagittal T1-weighted sequence. The sequences were acquired with two different resolutions (standard and low resolution) and the raw data reconstructed with two different reconstruction algorithms: a conventional Compressed SENSE (CS) and a new CNN-based algorithm for denoising and subsequently to interpolate and therewith increase the sharpness of the image (CS-SuperRes). Subjective image quality was evaluated by two blinded radiologists reviewing 8 criteria on a 5-point Likert scale and signal-to-noise ratio calculated as an objective parameter., Results: The protocol reconstructed with CS-SuperRes received higher ratings than the time-equivalent CS reconstructions, statistically significant especially for low resolution acquisitions (e.g., overall image impression: 4.3 ± 0.4 vs. 3.4 ± 0.4, p < 0.05). CS-SuperRes reconstructions for the low resolution acquisition were comparable to traditional CS reconstructions with standard resolution for all parameters, achieving a scan time reduction from 11:01 min to 4:46 min (57 %) for the complete protocol (e.g. overall image impression: 4.3 ± 0.4 vs. 4.0 ± 0.5, p < 0.05)., Conclusion: The newly-developed AI-based reconstruction algorithm CS-SuperRes allows to reduce scan time by 57% while maintaining unchanged image quality compared to the conventional CS reconstruction., Competing Interests: Declaration of competing interest David Maintz is part of the speakers bureau at Philips Healthcare; all other 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 Author(s). Published by Elsevier B.V. All rights reserved.)

Purpose: To develop a deep-learning (DL)-based approach for thoracic lymph node (LN) mapping based on their anatomical location., Method: The training-and validation-dataset included 89 contrast-enhanced computed tomography (CT) scans of the chest. 4201 LNs were semi-automatically segmented and then assigned to LN levels according to their anatomical location. The LN level classification task was addressed by a multi-class segmentation procedure using a fully convolutional neural network. Mapping was performed by firstly determining potential level affiliation for each voxel and then performing majority voting over all voxels belonging to each LN. Mean classification accuracies on the validation data were calculated separately for each level and overall Top-1, Top-2 and Top-3 scores were determined, where a Top-X score describes how often the annotated class was within the top-X predictions. To demonstrate the clinical applicability of our model, we tested its N-staging capabilities in a simulated clinical use case scenario assuming a patient diseased with lung cancer., Results: The artificial intelligence(AI)-based assignment revealed mean classification accuracies of 86.36 % (Top-1), 94.48 % (Top-2) and 96.10 % (Top-3). Best accuracies were achieved for LNs in the subcarinal level 7 (98.31 %) and axillary region (98.74 %). The highest misclassification rates were observed among LNs in adjacent levels. The proof-of-principle application in a simulated clinical use case scenario for automated tumor N-staging showed a mean classification accuracy of up to 96.14 % (Top-1)., Conclusions: The proposed AI approach for automatic classification of LN levels in chest CT as well as the proof-of-principle-experiment for automatic N-staging, revealed promising results, warranting large-scale validation for clinical application., (Copyright © 2021 Elsevier B.V. All rights reserved.)

Purpose: To evaluate potential clinical acceleration factors of Compressed SENSE (CS) 1 in direct comparison with SENSE for fat saturated (fs) 2 , proton density-weighted (PD) 3 2D and 3D sequences of the knee., Method: Twenty healthy volunteers were scanned with a 3 T scanner, all receiving a standard, fs 2D PD, three CS (CS 2, CS 3, CS 5) as well as time-equivalent SENSE accelerations (S 2, S 3, S 5). The fs 3D PD sequence was acquired with four CS (CS 6, CS 8, CS 10, CS 15) and equivalent SENSE (S 5.72, S 7.69, S 9.57, S 14) factors. Three independent readers rated the images. Signal-to-noise, contrast-to-noise, root-mean-square error and structural similarity index were analyzed for objective evaluation., Results: Scan time decreased with increasing CS factor (2D CS 2: 145 s, 2D CS 3: 95 s, 2D CS 5: 57 s, 3D CS 6: 293 s, 3D CS 8: 220 s, 3D CS 10: 176 s, 3D CS 15: 119 s). The 2D standard sequence was rated best for diagnostic certainty and overall image impression with an average of 4.97 ± 0.10 and 4.80 ± 0.24 (all p < 0.05), except for 2D CS 2 and 2D S 2. For the 3D sequences, the standard sequence performed better for both parameters for CS 15, S 9.57 and S 4, as well as S 7.69 for overall image impression while CS 8 was non-inferior for all tested criteria and CS 10 only inferior for delineation of the anterior cruciate ligament, both outperforming the time-equivalent SENSE accelerations., Conclusion: Compressed SENSE can significantly decrease (34.39 % for 2D CS 2 and 54.17 % for 3D CS 10) scan time in knee imaging with unchanged diagnostic certainty and overall image impression compared to the clinical reference., (Copyright © 2020 Elsevier B.V. All rights reserved.)

Objectives: To investigate the utilization of virtual mono-energetic images (MonoE) at low kiloelectron volt (keV) levels derived from a dual-layer spectral detector CT (SDCT) for the assessment of abdominal arteries in venous contrast phase scans using arterial phase imaging as an internal reference standard., Materials and Methods: A total of 50 patients who received arterial and venous phase imaging of the abdomen on a SDCT system were included in this study. Absolute attenuation, noise, signal- and contrast to noise ratios (SNR; CNR) as well as arterial diameters in defined landmarks were assessed. In arterial phase, conventional reconstructions (CR ART ) as well as MonoE ART at 40keV and in venous phase, conventional reconstructions (CR VEN ) as well as MonoE VEN at 70 and 40keV were investigated and intra-individual comparisons were performed. If an artery stenosis (10 patients) was present, the degree of stenosis was assessed according to the system of the North American Symptomatic Carotid Endarterectomy Trial (NASCET)., Results: MonoE 40keV yielded significantly higher attenuation values (in arterial as well as in venous phase) compared to CR ART (p<0.001) while noise levels were substantially low. This resulted in markedly superior SNR and CNR in large vessel compared to CR ART . Luminal diameters were significantly smaller in MonoE 40keV in both contrast phases compared to CR ART (p<0.001), whereas no significant differences were found between both MonoE reconstructions (p≥0.92). The degree of vessel stenosis was significantly higher in MonoE 40keV of both contrast phases compared to CR ART (p≥0.02)., Conclusion: MonoE at low keV of venous contrast phase scans derived from a novel SDCT are suitable for the assessment of arteries in the abdomen and subsequent stenosis assessment. However, MonoE at 40keV constantly showed significant smaller luminal diameters than the corresponding conventional reconstructions (including the reference standard). This is possibly due to an improved differentiation of the vessel lumen from the wall and raises the question, which imaging technique should be used as an appropriate reference standard for vascular SDCT imaging studies., (Copyright © 2017 Elsevier B.V. All rights reserved.)