Frank Zijlstra

Frank Zijlstra

Auto-generated profile, based on publicly available information

Frank Zijlstra

Department of Radiology and Nuclear Medicine St Olav's University Hospital Trondheim Norway

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 October 2019 in Magnetic Resonance in Medicine
Verified
Journal Article
Open access
PDF file
MFMateusz C. Florkow
Mateusz C. Florkow
FZFrank Zijlstra
Frank Zijlstra
KWKoen Willemsen
Koen Willemsen
MMMatteo Maspero
Matteo Maspero
CBCornelis A. T. van Den Berg
Cornelis A. T. van Den Berg
PSPeter R. Seevinck
Peter R. Seevinck
Purpose: To study the influence of gradient echo–based contrasts as input channels to a 3D patch‐based neural network trained for synthetic CT (sCT) generation in canine and human populations.Methods: Magnetic resonance images and CT scans of human and canine pelvic regions were acquired and paired using nonrigid registration. Magnitude MR images and Dixon reconstructed water, fat, in‐phase and opposed‐phase images were obtained from a single T1‐weighted multi‐echo gradient‐echo acquisition. From this set, 6 input configurations were defined, each containing 1 to 4 MR images regarded as input channels. For each configuration, a UNet‐derived deep learning model was trained for synthetic CT generation. Reconstructed Hounsfield unit maps were evaluated with peak SNR, mean absolute error, and mean error. Dice similarity coefficient and surface distance maps assessed the geometric fidelity of bones. Repeatability was estimated by replicating the training up to 10 times.Results: Seventeen canines and 23 human subjects were included in the study. Performance and repeatability of single‐channel models were dependent on the TE‐related water–fat interference with variations of up to 17% in mean absolute error, and variations of up to 28% specifically in bones. Repeatability, Dice similarity coefficient, and mean absolute error were statistically significantly better in multichannel models with mean absolute error ranging from 33 to 40 Hounsfield units in humans and from 35 to 47 Hounsfield units in canines.Conclusion: Significant differences in performance and robustness of deep learning models for synthetic CT generation were observed depending on the input. In‐phase images outperformed opposed‐phase images, and Dixon reconstructed multichannel inputs outperformed single‐channel inputs.

1 June 2016 in Investigative Radiology
Verified
Journal Article
Closed access
FZFrank Msc Zijlstra
Frank Msc Zijlstra
MVMax A. Viergever
Max A. Viergever
PSPeter R. Seevinck
Peter R. Seevinck
Objectives The aim of this study was to investigate the influence of variable density and data-driven k-space undersampling patterns on reconstruction quality for compressed sensing (CS) magnetic resonance imaging to provide recommendations on how to avoid suboptimal CS reconstructions. Materials and Methods First, we investigated the influence of randomness and sampling density on the reconstruction quality when using random variable density and variable density Poisson disk undersampling. Compressed sensing reconstructions on 1 knee and 2 brain data sets were compared with fully sampled data sets and reconstruction errors were measured. Sampling coherence was evaluated on the undersampling patterns to investigate whether there was a relation between this coherence measure and reconstruction error. Second, we investigated whether data-driven undersampling methods could improve reconstruction quality when 1 or more fully sampled scans are available as a training set. We implemented 3 different data-driven undersampling methods: (1) Monte Carlo optimization of variable density and variable density Poisson disk undersampling, (2) calculating sampling probabilities directly from the k-space power spectra of the training data, and (3) iterative design of undersampling patterns based on CS reconstruction errors in k-space. Two cross-validation experiments were set up using retrospective undersampling to evaluate the 3 data-driven methods and the influence of the size of the training set. Furthermore, in an experiment that included prospective under sampling, we show the practical applicability of 2 of the data-driven methods. Compressed sensing reconstruction quality was measured with both the normalized root-mean-square error metric and the mean structural similarity index measure. Results Different optimal variable sampling densities were found for each of the data sets, showing that the optimal sampling density is data dependent. Choosing a sampling density other than the optimal density decreased reconstruction quality. These results suggest that choosing a sampling density without having any reference scans is likely suboptimal. Furthermore, no meaningful correlation was found between sampling coherence and reconstruction error. For the data-driven methods, the iterative method yielded statistically significantly higher reconstruction quality in both retrospective and prospective experiments. In retrospective experiments, the power spectrum method yielded a reconstruction quality that was comparable with the data-driven variable density method. The size of the training set had only a minor influence on the reconstruction quality. Conclusions Data-driven undersampling methods can be used to avoid suboptimal reconstruction quality in CS magnetic resonance imaging, provided that at least 1 fully sampled scan is available to train the data-driven method. The iterative design method resulted in the highest reconstruction quality.

13 March 2023 in Magnetic Resonance in Medicine
Verified
Journal Article
Closed access
MKMisha Kaandorp
Misha Kaandorp
FZFrank Zijlstra
Frank Zijlstra
CFChristian Federau
Christian Federau
PWPeter T. While
Peter T. While

3 October 2017 in Physics in Medicine & Biology
Verified
Journal Article
Closed access
PDF file
MMMatteo Maspero
Matteo Maspero
Cássio van den BergCássio van den Berg
Cássio van den Berg
FZFrank Zijlstra
Frank Zijlstra
GSGonda G Sikkes
Gonda G Sikkes
HBHans C J de Boer
Hans C J de Boer
PSPeter R Seevinck
Peter R Seevinck
An MR-only radiotherapy planning (RTP) workflow would reduce the cost, radiation exposure and uncertainties introduced by CT-MRI registrations. In the case of prostate treatment, one of the remaining challenges currently holding back the implementation of an RTP workflow is the MR-based localisation of intraprostatic gold fiducial markers (FMs), which is crucial for accurate patient positioning. Currently, MR-based FM localisation is clinically performed manually. This is sub-optimal, as manual interaction increases the workload. Attempts to perform automatic FM detection often rely on being able to detect signal voids induced by the FMs in magnitude images. However, signal voids may not always be sufficiently specific, hampering accurate and robust automatic FM localisation. Here, we present an approach that aims at automatic MR-based FM localisation. This method is based on template matching using a library of simulated complex-valued templates, and exploiting the behaviour of the complex MR signal in the vicinity of the FM. Clinical evaluation was performed on seventeen prostate cancer patients undergoing external beam radiotherapy treatment. Automatic MR-based FM localisation was compared to manual MR-based and semi-automatic CT-based localisation (the current gold standard) in terms of detection rate and the spatial accuracy and precision of localisation. The proposed method correctly detected all three FMs in 15/17 patients. The spatial accuracy (mean) and precision (STD) were 0.9 mm and 0.5 mm respectively, which is below the voxel size of 1.1 × 1.1 × 1.2 mm 3 and comparable to MR-based manual localisation. FM localisation failed (3/51 FMs) in the presence of bleeding or calcifications in the direct vicinity of the FM. The method was found to be spatially accurate and precise, which is essential for clinical use. To overcome any missed detection, we envision the use of the proposed method along with verification by an observer. This will result in a semi-automatic workflow facilitating the introduction of an MR-only workflow.

7 December 2016 in Magnetic Resonance in Medicine
Verified
Journal Article
Open access
PDF file
FZFrank Zijlstra
Frank Zijlstra
JBJob G. Bouwman
Job G. Bouwman
IBIeva Braškutė
Ieva Braškutė
MVMax A. Viergever
Max A. Viergever
PSPeter R. Seevinck
Peter R. Seevinck
Purpose: To accelerate simulation of off‐resonance artifacts in steady‐state gradient echo MRI by using fast Fourier transforms and demonstrate its applicability to metal object localization.Theory and Methods: By exploiting the repetitive nature of steady‐state pulse sequences it is possible to use fast Fourier transforms to calculate the MR signal. Based on this principle, a method for fast simulation of off‐resonance artifacts was designed. The method was validated against Bloch simulations and MRI scans. Its clinical relevance was demonstrated by employing it for template matching‐based metal object localization, as applied to a titanium cylinder, an oxidized zirconium knee implant, and gold fiducials.Results: The fast simulations were accurate compared with actual MRI scans of the objects. The differences between the fast simulations and Bloch simulations were minor, while the acceleration scaled linearly with the number of phase‐encoding lines. The object localization method accurately localized the various metal objects.Conclusion: The proposed simulation methodology provided accurate 3D simulations of off‐resonance artifacts with a lower computational complexity than Bloch simulations. The speed of the simulations opens up possibilities in image reconstructions involving off‐resonance phenomena that were previously infeasible due to computational limitations, as demonstrated for metal object localization. Magn Reson Med 78:2035–2041, 2017. © 2016 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.

15 March 2019
Verified
Conference Paper
Closed access
FZFrank Zijlstra
Frank Zijlstra
KWKoen Willemsen
Koen Willemsen
MFMateusz C. Florkow
Mateusz C. Florkow
RSRalph J. B. Sakkers
Ralph J. B. Sakkers
HWHarrie H. Weinans
Harrie H. Weinans
PSPeter R. Seevinck
Peter R. Seevinck
Purpose: To assess the feasibility of deep learning-based high resolution synthetic CT generation from MRI scans of the lower arm for orthopedic applications. Methods: A conditional Generative Adversarial Network was trained to synthesize CT images from multi-echo MR images. A training set of MRI and CT scans of 9 ex vivo lower arms was acquired and the CT images were registered to the MRI images. Three-fold cross-validation was applied to generate independent results for the entire dataset. The synthetic CT images were quantitatively evaluated with the mean absolute error metric, and Dice similarity and surface to surface distance on cortical bone segmentations. Results: The mean absolute error was 63.5 HU on the overall tissue volume and 144.2 HU on the cortical bone. The mean Dice similarity of the cortical bone segmentations was 0.86. The average surface to surface distance between bone on real and synthetic CT was 0.48 mm. Qualitatively, the synthetic CT images corresponded well with the real CT scans and partially maintained high resolution structures in the trabecular bone. The bone segmentations on synthetic CT images showed some false positives on tendons, but the general shape of the bone was accurately reconstructed. Conclusions: This study demonstrates that high quality synthetic CT can be generated from MRI scans of the lower arm. The good correspondence of the bone segmentations demonstrates that synthetic CT could be competitive with real CT in applications that depend on such segmentations, such as planning of orthopedic surgery and 3D printing.