Search

The vast majority of PET detectors in the field today are based on pixelated scintillators. Yet, the resolution of this type of detector is limited by the pixel size. To overcome this limitation, one can use monolithic detectors. However, this detector architecture demands specific and high-speed detector readout of the photodetector array. A commonly used approach is to integrate the current pulses generated by every pixel but such circuitry quickly becomes bulky, power consuming and expensive. The objective of this work is to investigate a novel readout and event positioning scheme for monolithic PET detectors, based on time-over-threshold (ToT). In this case, we measure the time that the pulse is above a certain threshold through a comparator. The pulse widths are used for event positioning using a mean nearest neighbour approach (mNN ToT ). For energy determination one integrating multiplexed channel is foreseen. We evaluate the positioning accuracy and uniformity of such a ToT detector by means of Monte Carlo simulations. The impact of the threshold value is investigated and the results are compared to a detector using mean nearest neighbour with pulse-integration (mNN int ), which has already proven to allow sub-mm resolution. We show minimal degradation in spatial resolution and bias performance compared to mNN int . The highest threshold results in the worst resolution performance but degradation remains below 0.1 mm. Bias is largely constant over different thresholds for mNN ToT and close to identical to mNN int . Furthermore we show that ToT performs well in terms of detector uniformity and that scattered photons can be positioned inside the crystal with high accuracy. We conclude from this work that ToT is a valuable alternative to pulse-integration for monolithic PET detectors. This novel approach has an impact on PET detector development since it has the advantage of lower power consumption, compactness and inherent amplitude-to-time conversion., (© 2021 Institute of Physics and Engineering in Medicine.)

The use of deep learning in medical imaging has increased rapidly over the past few years, finding applications throughout the entire radiology pipeline, from improved scanner performance to automatic disease detection and diagnosis. These advancements have resulted in a wide variety of deep learning approaches being developed, solving unique challenges for various imaging modalities. This paper provides a review on these developments from a technical point of view, categorizing the different methodologies and summarizing their implementation. We provide an introduction to the design of neural networks and their training procedure, after which we take an extended look at their uses in medical imaging. We cover the different sections of the radiology pipeline, highlighting some influential works and discussing the merits and limitations of deep learning approaches compared to other traditional methods. As such, this review is intended to provide a broad yet concise overview for the interested reader, facilitating adoption and interdisciplinary research of deep learning in the field of medical imaging., (© 2021. The Author(s).)

The system spatial resolution of whole-body positron emission tomography (PET) is limited to around 2 mm due to positron physics and the large diameter of the bore. To stay below this 'physics'-limit a scintillation detector with an intrinsic spatial resolution of around 1.3 mm is needed. Currently used detector technology consists of arrays of 2.6-5 mm segmented scintillator pixels which are the dominant factor contributing to the system resolution. Pixelated detectors using smaller pixels exist but face major drawbacks in sensitivity, timing, energy resolution and cost. Monolithic continuous detectors, where the spatial resolution is determined by the shape of the light distribution on the photodetector array, are a promising alternative. Without having the drawbacks of pixelated detectors, monolithic ones can also provide depth-of-interaction (DOI) information. In this work we present a monolithic detector design aiming to serve high-resolution clinical PET systems while maintaining high sensitivity. A 50 × 50 × 16 mm 3 Lutetium-Yttrium oxyorthosilicate scintillation crystal with silicon photomultiplier (SiPM) backside readout is calibrated in singles mode by a collimated beam obtaining a reference dataset for the event positioning. A mean nearest neighbour (MNN) algorithm and an artificial neural network for positioning are compared. The targeted intrinsic detector resolution of 1.3 mm needed to reach a 2 mm resolution on system level was accomplished with both algorithms. The neural network achieved a mean spatial resolution of 1.14 mm FWHM for the whole detector and 1.02 mm in the centre (30 × 30 mm 2 ). The MNN algorithm performed slightly worse with 1.17 mm for the whole detector and 1.13 mm in the centre. The intrinsic DOI information will also result in uniform system spatial resolution over the full field of view., (© 2021 Institute of Physics and Engineering in Medicine.)

To detect gamma rays with good spatial, timing and energy resolution while maintaining high sensitivity we need accurate and efficient algorithms to estimate the first gamma interaction position from the measured light distribution. Furthermore, monolithic detectors are investigated as an alternative to pixelated detectors due to increased sensitivity, resolution and intrinsic DOI encoding. Monolithic detectors, however, are challenging because of complicated calibration setups and edge effects. In this work, we evaluate the use of neural networks to estimate the 3D first (Compton or photoelectric) interaction position. Using optical simulation data of a 50 × 50 × 16 mm 3 LYSO crystal, performance is evaluated as a function of network complexity (two to five hidden layers with 64 to 1024 neurons) and amount of training data (1000-8000 training events per calibration position). We identify and address the potential pitfall of overfitting on the training grid through evaluation on intermediate positions that are not in the training set. Additionally, the performance of neural networks is directly compared with nearest neighbour positioning. Optimal performance was achieved with a network containing three hidden layers of 256 neurons trained on 1000 events/position. For more complex networks, the performance degrades at intermediate positions and overfitting starts to occur. A median 3D positioning error of 0.77 mm and a 2D FWHM of 0.46 mm is obtained. This is a 17% improvement in terms of FWHM compared to the nearest neighbour algorithm. Evaluation only on events that are not Compton scattered results in a 3D positioning error of 0.40 mm and 2D FWHM of 0.42 mm. This reveals that Compton scatter results in a considerable increase of 93% in positioning error. This study demonstrates that very good spatial resolutions can be achieved with neural networks, superior to nearest neighbour positioning. However, potential overfitting on the training grid should be carefully evaluated., (© 2021 Institute of Physics and Engineering in Medicine.)

In the WHO glioma classification guidelines grade (glioblastoma versus lower-grade glioma), IDH mutation and 1p/19q co-deletion status play a central role as they are important markers for prognosis and optimal therapy planning. Currently, diagnosis requires invasive surgical procedures. Therefore, we propose an automatic segmentation and classification pipeline based on routinely acquired pre-operative MRI (T1, T1 postcontrast, T2 and/or FLAIR). A 3D U-Net was designed for segmentation and trained on the BraTS 2019 training dataset. After segmentation, the 3D tumor region of interest is extracted from the MRI and fed into a CNN to simultaneously predict grade, IDH mutation and 1p19q co-deletion. Multi-task learning allowed to handle missing labels and train one network on a large dataset of 628 patients, collected from The Cancer Imaging Archive and BraTS databases. Additionally, the network was validated on an independent dataset of 110 patients retrospectively acquired at the Ghent University Hospital (GUH). Segmentation performance calculated on the BraTS validation set shows an average whole tumor dice score of 90% and increased robustness to missing image modalities by randomly excluding input MRI during training. Classification area under the curve scores are 93%, 94% and 82% on the TCIA test data and 94%, 86% and 87% on the GUH data for grade, IDH and 1p19q status respectively. We developed a fast, automatic pipeline to segment glioma and accurately predict important (molecular) markers based on pre-therapy MRI., (Copyright © 2020 Elsevier Ltd. All rights reserved.)

Objective: Excitation of myelinated nerve fibers is investigated by means of numerical simulations, for the application of percutaneous auricular vagus nerve stimulation (pVNS). High sensitivity to axon diameter is of interest regarding the goal of targeting thicker fibers., Methods: Excitation and blocking thresholds for different pulse types, phase durations, axon depths, axon-electrode distances, temperatures and axon diameters are investigated. The used model consists of a 50 mm long axon and a centrally located needle electrode in a layered medium representing the auricle. Neuronal excitation is simulated using the Frankenhaeuser-Huxley equations for all combinations of parameter values., Results and Conclusion: Multiple modes and locations of excitation along the axon were observed, depending on the pulse type and amplitude. When increasing the axon-electrode distance from 1 mm to 2 mm, sensitivity of thresholds to axon depth decreased with ca. 50%, while sensitivity to axon-electrode distance, axon diameter and phase duration each increased with ca. 15% to 20%, except from monophasic anodal pulses, showing a 45% decrease for axon-electrode distance. These trends for axon diameter and axon-electrode distance allow for more selective stimulation of thicker target fibers using monophasic anodal pulses at higher axon-electrode distances. Cathodal monophasic pulses did not perform well due to blocking of the thicker fibers, which was only rarely seen for other pulse types., Significance: Sensitivities of stimulation thresholds to these parameters by numerical simulation reveal how the stimulation parameters can be changed in order to increase therapeutic effect and comfort during pVNS by enabling more selective stimulation.

Rationale: Positive surgical margins for invasive breast cancer (BC) treated with breast-conserving surgery (BCS) are defined as ink on tumor. The rate of positive margins is approximately 20%, since a time- and cost-effective method for margin assessment is lacking. In this study, we investigated margin status by intra-operative imaging using high-resolution 18 F-fluoro-deoxyglucose (FDG) positron emission tomography (PET) and X-ray computed tomography (CT). Methods: Twenty patients were enrolled and received 4   MBq/kg of FDG prior to surgery. Intra-operative imaging of the specimens was performed by the MOLECUBES β-CUBE (PET) and X-CUBE (CT). Margin status was assessed by three surgeons and compared with an algorithm. The sensitivity and specificity were calculated by using histopathological assessment as a gold standard. Results: A region with high FDG uptake was visualized in all specimens. Automated analysis showed a sensitivity of 90%, a specificity of 60%, and an area under the curve (AUC) of 0.86 after ROC analysis. Margin assessment by the surgeons resulted in a mean sensitivity and specificity of 79% and 72%, respectively. Conclusions: This proof-of-concept study demonstrates that high-resolution FDG-PET/CT can facilitate intra-operative margin assessment during BCS. This technique achieves good sensitivity and specificity and may therefore reduce re-operation rates in the future.

The intrinsic spatial resolution of clinical positron emission tomography (PET) detectors is ~3-4 mm. A further improvement of the resolution using pixelated detectors will not only result in a prohibitive cost, but is also inevitably accompanied by a strong degradation of important performance parameters like timing, energy resolution and sensitivity. Therefore, it is likely that future generation high resolution PET detectors will be based on continuous monolithic scintillation detectors. Monolithic detectors have attractive properties to reach superior 3D spatial resolution while outperforming pixelated detectors in timing, energy resolution and sensitivity. In this work, optical simulations including an advanced surface reflection model, allow us to investigate the influence of three parameters on the spatial resolution: silicon photomultiplier (SiPM) pixel size, photon detection efficiency (PDE) and the number of channels used to read out the SiPM array. A lutetium-yttrium oxyorthosilicate (LYSO) crystal with dimensions 50  ×  50  ×  16 mm 3 coupled to an SiPM array is calibrated and a nearest neighbor (NN) algorithm is used to position events. Findings show that the tested parameters affect the spatial resolution resulting in 0.40-0.66 mm full width at half maximum (FWHM). Best resolution could be obtained with smaller SiPM pixels, higher PDE, and an individual channel readout. However, it was shown that combining channels by adding their signals can significantly reduce the amount of readout channels while having small or no significant impact on the resolution. The mean depth of interaction (DOI) estimation error is 1.6 mm. This study demonstrates the ultimate spatial resolution that can be obtained with this detector without being constrained by practical limitations of experimental setups. In the future these optical simulations may be used as a more precise and fast method to obtain calibration data for real monolithic detectors.

The MOLECUBES β-CUBE scanner is the newest amongst commercially available preclinical PET scanners for dedicated small animal imaging. The scanner is compact, lightweight and utilizes a small footprint to facilitate bench-top imaging. It can be used individually, or in combination with the X-CUBE CT scanner, which provides the ability to perform all necessary PET data corrections and provide fully quantitative PET images. The PET detector comprises of an 8 mm thick monolithic LYSO scintillator read-out by an array of 3 mm  ×  3 mm Hamamatsu silicon photomultipliers. The monolithic scintillator provides the ability to measure depth-of-interaction which aids in the development of such a compact scanner. With a scanner diameter of 7.6 cm and axial length of 13 cm it is suitable for imaging both whole-body mice and rats. This paper presents the design and imaging performance of the β-CUBE scanner. NEMA NU4-2008 characterization and a variety of phantom and animal imaging studies to demonstrate the quantitative imaging performance of the PET scanner are presented. Spatial resolution of 1 mm is measured with a filtered-back projection reconstruction algorithm at the center of the scanner and DOI measurement helps maintain the excellent spatial resolution over the entire imaging FOV. An absolute peak sensitivity of 12.4% is measured with a 255-765 keV energy window. The scanner demonstrates good count-rate performance, with a peak NEC of 300 kcps and 160 kcps measured with ~900 µCi in the NEMA mouse and rat phantoms, respectively. Imaging data with the NEMA image quality phantom and Micro Derenzo phantoms demonstrate the ability to achieve good image quality and accurate quantitative data. Image uniformity of 7.4% and spill-over ratio of 8% were measured. The superior spatial resolution, excellent energy resolution and sensitivity also provide superior contrast recovery, with ~70% recovery for the 2 mm rods. While current commercial preclinical PET scanners have spatial resolution in the 1-2 mm range, the 1 mm 3 volumetric resolution presents significant improvement over current commercially available preclinical PET scanners. In combination with the X-CUBE scanner it provides the ability to perform fully quantitative imaging with spatially co-registered high-resolution 3D PET-CT images.

Brain tumour segmentation in medical images is a very challenging task due to the large variety in tumour shape, position, appearance, scanning modalities and scanning parameters. Most existing segmentation algorithms use information from four different MRI-sequences, but since this is often not available, there is need for a method able to delineate the different tumour tissues based on a minimal amount of data. We present a novel approach using a Random Forests model combining voxelwise texture and abnormality features on a contrast-enhanced T1 and FLAIR MRI. We transform the two scans into 275 feature maps. A random forest model next calculates the probability to belong to 4 tumour classes or 5 normal classes. Afterwards, a dedicated voxel clustering algorithm provides the final tumour segmentation. We trained our method on the BraTS 2013 database and validated it on the larger BraTS 2017 dataset. We achieve median Dice scores of 40.9% (low-grade glioma) and 75.0% (high-grade glioma) to delineate the active tumour, and 68.4%/80.1% for the total abnormal region including edema. Our fully automated brain tumour segmentation algorithm is able to delineate contrast enhancing tissue and oedema with high accuracy based only on post-contrast T1-weighted and FLAIR MRI, whereas for non-enhancing tumour tissue and necrosis only moderate results are obtained. This makes the method especially suitable for high-grade glioma., (Copyright © 2018 Elsevier Ltd. All rights reserved.)

Electrical source imaging (ESI) from interictal scalp EEG is increasingly validated and used as a valuable tool in the presurgical evaluation of epilepsy as a reflection of the irritative zone. ESI of ictal scalp EEG to localize the seizure onset zone (SOZ) remains challenging. We investigated the value of an approach for ictal imaging using ESI and functional connectivity analysis (FC). Ictal scalp EEG from 111 seizures in 27 patients who had Engel class I outcome at least 1 year following resective surgery was analyzed. For every seizure, an artifact-free epoch close to the seizure onset was selected and ESI using LORETA was applied. In addition, the reconstructed sources underwent FC using the spectrum-weighted Adaptive Directed Transfer Function. This resulted in the estimation of the SOZ in two ways: (i) the source with maximal power after ESI, (ii) the source with the strongest outgoing connections after combined ESI and FC. Next, we calculated the distance between the estimated SOZ and the border of the resected zone (RZ) for both approaches and called this the localization error ((i) LE pow and (ii) LE conn respectively). By comparing LE pow and LE conn , we assessed the added value of FC. The source with maximal power after ESI was inside the RZ (LE pow  = 0 mm) in 31% of the seizures and estimated within 10 mm from the border of the RZ (LE pow  ≤ 10 mm) in 42%. Using ESI and FC, these numbers increased to 72% for LE conn  = 0 mm and 94% for LE conn  ≤ 10 mm. FC provided a significant added value to ESI alone ( p  < 0.001). ESI combined with subsequent FC is able to localize the SOZ in a non-invasive way with high accuracy. Therefore it could be a valuable tool in the presurgical evaluation of epilepsy.

Purpose: In this study, we investigated fluorine-18 fluoromethylcholine (F-FCho) PET and contrast-enhanced MRI for predicting therapy response in glioblastoma (GB) patients according to the Response Assessment in Neuro-Oncology criteria. Our second aim was to investigate which imaging modality enabled prediction of treatment response first., Materials and Methods: Eleven GB patients who underwent no surgery or debulking only and received concomitant radiation therapy (RT) and temozolomide were included. The gold standard Response Assessment in Neuro-Oncology criteria were applied 6 months after RT to define responders and nonresponders. F-FCho PET and MRI were performed before RT, during RT (week 2, 4, and 6), and 1 month after RT. The contrast-enhancing tumor volume on T1-weighted MRI (GdTV) and the metabolic tumor volume (MTV) were calculated. GdTV, standardized uptake value (SUV)mean, SUVmax, MTV, MTV×SUVmean, and percentage change of these variables between all time-points were assessed to differentiate responders from nonresponders., Results: Absolute SUV values did not predict response. MTV must be taken into account. F-FCho PET could predict response with a 100% sensitivity and specificity using MTV×SUVmean 1 month after RT. A decrease in GdTV between week 2 and 6, week 4 and 6 during RT and week 2 during RT, and 1 month after RT of at least 31%, at least 18%, and at least 53% predicted response with a sensitivity and specificity of 100%. As such, the parameter that predicts therapy response first is MR derived, namely, GdTV., Conclusion: Our data indicate that both F-FCho PET and contrast-enhanced T1-weighted MRI can predict response early in GB patients treated with RT and temozolomide.

The substantia nigra pars reticulata (SNr) and external globus pallidus (GPe) constitute the two major output targets of the rodent striatum. Both the SNr and GPe converge upon thalamic relay nuclei (directly or indirectly, respectively), and are traditionally modeled as functionally antagonistic relay inputs. However, recent anatomical and functional studies have identified unanticipated circuit connectivity in both the SNr and GPe, demonstrating their potential as far more than relay nuclei. In the present study, we employed simultaneous deep brain stimulation and functional magnetic resonance imaging (DBS-fMRI) with cerebral blood volume (CBV) measurements to functionally and unbiasedly map the circuit- and network level connectivity of the SNr and GPe. Sprague-Dawley rats were implanted with a custom-made MR-compatible stimulating electrode in the right SNr (n=6) or GPe (n=7). SNr- and GPe-DBS, conducted across a wide range of stimulation frequencies, revealed a number of surprising evoked responses, including unexpected CBV decreases within the striatum during DBS at either target, as well as GPe-DBS-evoked positive modulation of frontal cortex. Functional connectivity MRI revealed global modulation of neural networks during DBS at either target, sensitive to stimulation frequency and readily reversed following cessation of stimulation. This work thus contributes to a growing literature demonstrating extensive and unanticipated functional connectivity among basal ganglia nuclei., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Purpose: To determine exposure to gradient switching fields of adults and children in a magnetic resonance imaging (MRI) scanner by evaluating internal electric fields within realistic models of adult male, adult female, and child inside transverse and longitudinal gradient coils, and to compare these results with compliance guidelines., Materials and Methods: Patients inside x-, y-, and z-gradient coils were simulated using anatomically realistic models of adult male, adult female, and child. The induced electric fields were computed for 1 kHz sinusoidal current with a magnitude of 1 A in the gradient coils. Rheobase electric fields were then calculated and compared to the International Commission on Non-Ionizing Radiation Protection (ICNIRP) 2004 and International Electrotechnical Commission (IEC) 2010 guidelines. The effect of the human body, coil type, and skin conductivity on the induced electric field was also investigated., Results: The internal electric fields are within the first level controlled operating mode of the guidelines and range from 2.7V m -1 to 4.5V m -1 , except for the adult male inside the y-gradient coil (induced field reaches 5.4V m -1 ).The induced electric field is sensitive to the coil type (electric field in the skin of adult male: 4V m -1 , 4.6V m -1 , and 3.8V m -1 for x-, y-, and z-gradient coils, respectively), the human body model (electric field in the skin inside y-gradient coil: 4.6V m -1 , 4.2V m -1 , and 3V m -1 for adult male, adult female, and child, respectively), and the skin conductivity (electric field 2.35-4.29% higher for 0.1S m -1 skin conductivity compared to 0.2S m -1 )., Conclusion: The y-gradient coil induced the largest fields in the patients. The highest levels of internal electric fields occurred for the adult male model. J. Magn. Reson. Imaging 2016;44:1360-1367., (© 2016 International Society for Magnetic Resonance in Medicine.)

The mouse model is widely used in a vast range of biomedical and preclinical studies. Thanks to the ability to detect and quantify biological processes at the molecular level in vivo, PET has become a well-established tool in these investigations. However, the need to visualize and quantify radiopharmaceuticals in anatomic structures of millimetre or less requires good spatial resolution and sensitivity from small-animal PET imaging systems.In previous work we have presented a proof-of-concept of a dedicated high-resolution small-animal PET scanner based on thin monolithic scintillator crystals and Digital Photon Counter photosensor. The combination of thin monolithic crystals and MLE positioning algorithm resulted in an excellent spatial resolution of 0.7 mm uniform in the entire field of view (FOV). However, the limitation of the scanner was its low sensitivity due to small thickness of the lutetium-yttrium oxyorthosilicate (LYSO) crystals (2 mm).Here we present an improved detector design for a small-animal PET system that simultaneously achieves higher sensitivity and sustains a sub-millimetre spatial resolution. The proposed detector consists of a 5 mm thick monolithic LYSO crystal optically coupled to a Digital Photon Counter. Mean nearest neighbour (MNN) positioning combined with depth of interaction (DOI) decoding was employed to achieve sub-millimetre spatial resolution. To evaluate detector performance the intrinsic spatial resolution, energy resolution and coincidence resolving time (CRT) were measured. The average intrinsic spatial resolution of the detector was 0.60 mm full-width-at-half-maximum (FWHM). A DOI resolution of 1.66 mm was achieved. The energy resolution was 23% FWHM at 511 keV and CRT of 529 ps were measured. The improved detector design overcomes the sensitivity limitation of the previous design by increasing the nominal sensitivity of the detector block and retains an excellent intrinsic spatial resolution.

Purpose: We investigated the temporal variation of the induced magnetic field due to the transverse and the longitudinal gradient coils in tungsten collimators arranged in hexagonal and pentagonal geometries with and without gaps between the collimators., Methods: We modeled x-, y-, and z-gradient coils and different arrangements of single-photon emission computed tomography (SPECT) collimators using FEKO, a three-dimensional electromagnetic simulation tool. A time analysis approach was used to generate the pulsed magnetic field gradient. The approach was validated with measurements using a 7T MRI scanner., Results: Simulations showed an induced magnetic field representing 4.66% and 0.87% of the applied gradient field (gradient strength = 500 mT/m) for longitudinal and transverse gradient coils, respectively. These values can be reduced by 75% by adding gaps between the collimators for the pentagonal arrangement, bringing the maximum induced magnetic field to less than 2% of the applied gradient for all of the gradient coils., Conclusion: Characterization of the maximum induced magnetic field shows that by adding gaps between the collimators for an integrated SPECT/MRI system, eddy currents can be corrected by the MRI system to avoid artifact. The numerical model was validated and was proposed as a tool for studying the effect of a SPECT collimator within the MRI gradient coils., (© 2014 Wiley Periodicals, Inc.)