Search

The zebrafish Danio rerio has become a popular model host to explore disease pathology caused by infectious agents. A main advantage is its transparency at an early age, which enables live imaging of infection dynamics. While multispecies infections are common in patients, the zebrafish model is rarely used to study them, although the model would be ideal for investigating pathogen-pathogen and pathogen-host interactions. This may be due to the absence of an established multispecies infection protocol for a defined organ and the lack of suitable image analysis pipelines for automated image processing. To address these issues, we developed a protocol for establishing and tracking single and multispecies bacterial infections in the inner ear structure (otic vesicle) of the zebrafish by imaging. Subsequently, we generated an image analysis pipeline that involved deep learning for the automated segmentation of the otic vesicle, and scripts for quantifying pathogen frequencies through fluorescence intensity measures. We used Pseudomonas aeruginosa, Acinetobacter baumannii, and Klebsiella pneumoniae, three of the difficult-to-treat ESKAPE pathogens, to show that our infection protocol and image analysis pipeline work both for single pathogens and pairwise pathogen combinations. Thus, our protocols provide a comprehensive toolbox for studying single and multispecies infections in real-time in zebrafish., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2024 Schmitz et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Automated detection of specific cells in three-dimensional datasets such as whole-brain light-sheet image stacks is challenging. Here, we present DELiVR, a virtual reality-trained deep-learning pipeline for detecting c-Fos + cells as markers for neuronal activity in cleared mouse brains. Virtual reality annotation substantially accelerated training data generation, enabling DELiVR to outperform state-of-the-art cell-segmenting approaches. Our pipeline is available in a user-friendly Docker container that runs with a standalone Fiji plugin. DELiVR features a comprehensive toolkit for data visualization and can be customized to other cell types of interest, as we did here for microglia somata, using Fiji for dataset-specific training. We applied DELiVR to investigate cancer-related brain activity, unveiling an activation pattern that distinguishes weight-stable cancer from cancers associated with weight loss. Overall, DELiVR is a robust deep-learning tool that does not require advanced coding skills to analyze whole-brain imaging data in health and disease., (© 2024. The Author(s).)

The translation of AI-generated brain metastases (BM) segmentation into clinical practice relies heavily on diverse, high-quality annotated medical imaging datasets. The BraTS-METS 2023 challenge has gained momentum for testing and benchmarking algorithms using rigorously annotated internationally compiled real-world datasets. This study presents the results of the segmentation challenge and characterizes the challenging cases that impacted the performance of the winning algorithms. Untreated brain metastases on standard anatomic MRI sequences (T1, T2, FLAIR, T1PG) from eight contributed international datasets were annotated in stepwise method: published UNET algorithms, student, neuroradiologist, final approver neuroradiologist. Segmentations were ranked based on lesion-wise Dice and Hausdorff distance (HD95) scores. False positives (FP) and false negatives (FN) were rigorously penalized, receiving a score of 0 for Dice and a fixed penalty of 374 for HD95. The mean scores for the teams were calculated. Eight datasets comprising 1303 studies were annotated, with 402 studies (3076 lesions) released on Synapse as publicly available datasets to challenge competitors. Additionally, 31 studies (139 lesions) were held out for validation, and 59 studies (218 lesions) were used for testing. Segmentation accuracy was measured as rank across subjects, with the winning team achieving a LesionWise mean score of 7.9. The Dice score for the winning team was 0.65 ± 0.25. Common errors among the leading teams included false negatives for small lesions and misregistration of masks in space. The Dice scores and lesion detection rates of all algorithms diminished with decreasing tumor size, particularly for tumors smaller than 100 mm3. In conclusion, algorithms for BM segmentation require further refinement to balance high sensitivity in lesion detection with the minimization of false positives and negatives. The BraTS-METS 2023 challenge successfully curated well-annotated, diverse datasets and identified common errors, facilitating the translation of BM segmentation across varied clinical environments and providing personalized volumetric reports to patients undergoing BM treatment., Competing Interests: Conflicts of Interest No conflicts of interest to disclose.

Optical coherence tomography angiography (OCTA) is a non-invasive imaging modality that can acquire high-resolution volumes of the retinal vasculature and aid the diagnosis of ocular, neurological and cardiac diseases. Segmenting the visible blood vessels is a common first step when extracting quantitative biomarkers from these images. Classical segmentation algorithms based on thresholding are strongly affected by image artifacts and limited signal-to-noise ratio. The use of modern, deep learning-based segmentation methods has been inhibited by a lack of large datasets with detailed annotations of the blood vessels. To address this issue, recent work has employed transfer learning, where a segmentation network is trained on synthetic OCTA images and is then applied to real data. However, the previously proposed simulations fail to faithfully model the retinal vasculature and do not provide effective domain adaptation. Because of this, current methods are unable to fully segment the retinal vasculature, in particular the smallest capillaries. In this work, we present a lightweight simulation of the retinal vascular network based on space colonization for faster and more realistic OCTA synthesis. We then introduce three contrast adaptation pipelines to decrease the domain gap between real and artificial images. We demonstrate the superior segmentation performance of our approach in extensive quantitative and qualitative experiments on three public datasets that compare our method to traditional computer vision algorithms and supervised training using human annotations. Finally, we make our entire pipeline publicly available, including the source code, pretrained models, and a large dataset of synthetic OCTA images.

Statistical shape models are an essential tool for various tasks in medical image analysis, including shape generation, reconstruction and classification. Shape models are learned from a population of example shapes, which are typically obtained through segmentation of volumetric medical images. In clinical practice, highly anisotropic volumetric scans with large slice distances are prevalent, e.g., to reduce radiation exposure in CT or image acquisition time in MR imaging. For existing shape modeling approaches, the resolution of the emerging model is limited to the resolution of the training shapes. Therefore, any missing information between slices prohibits existing methods from learning a high-resolution shape prior. We propose a novel shape modeling approach that can be trained on sparse, binary segmentation masks with large slice distances. This is achieved through employing continuous shape representations based on neural implicit functions. After training, our model can reconstruct shapes from various sparse inputs at high target resolutions beyond the resolution of individual training examples. We successfully reconstruct high-resolution shapes from as few as three orthogonal slices. Furthermore, our shape model allows us to embed various sparse segmentation masks into a common, low-dimensional latent space - independent of the acquisition direction, resolution, spacing, and field of view. We show that the emerging latent representation discriminates between healthy and pathological shapes, even when provided with sparse segmentation masks. Lastly, we qualitatively demonstrate that the emerging latent space is smooth and captures characteristic modes of shape variation. We evaluate our shape model on two anatomical structures: the lumbar vertebra and the distal femur, both from publicly available datasets., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Stefan Zachow reports a relationship with 1000shapes GmbH that includes: board membership and equity or stocks. Bjoern Menze is a member of the Editorial Board of Medical Image Analysis., (Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved.)

Background: Many automatic approaches to brain tumor segmentation employ multiple magnetic resonance imaging (MRI) sequences. The goal of this project was to compare different combinations of input sequences to determine which MRI sequences are needed for effective automated brain metastasis (BM) segmentation., Methods: We analyzed preoperative imaging (T1-weighted sequence ± contrast-enhancement (T1/T1-CE), T2-weighted sequence (T2), and T2 fluid-attenuated inversion recovery (T2-FLAIR) sequence) from 339 patients with BMs from seven centers. A baseline 3D U-Net with all four sequences and six U-Nets with plausible sequence combinations (T1-CE, T1, T2-FLAIR, T1-CE + T2-FLAIR, T1-CE + T1 + T2-FLAIR, T1-CE + T1) were trained on 239 patients from two centers and subsequently tested on an external cohort of 100 patients from five centers., Results: The model based on T1-CE alone achieved the best segmentation performance for BM segmentation with a median Dice similarity coefficient (DSC) of 0.96. Models trained without T1-CE performed worse (T1-only: DSC = 0.70 and T2-FLAIR-only: DSC = 0.73). For edema segmentation, models that included both T1-CE and T2-FLAIR performed best (DSC = 0.93), while the remaining four models without simultaneous inclusion of these both sequences reached a median DSC of 0.81-0.89., Conclusions: A T1-CE-only protocol suffices for the segmentation of BMs. The combination of T1-CE and T2-FLAIR is important for edema segmentation. Missing either T1-CE or T2-FLAIR decreases performance. These findings may improve imaging routines by omitting unnecessary sequences, thus allowing for faster procedures in daily clinical practice while enabling optimal neural network-based target definitions., 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 © 2023 The Authors. Published by Elsevier B.V. All rights reserved.)

Collateral circulation results from specialized anastomotic channels which are capable of providing oxygenated blood to regions with compromised blood flow caused by arterial obstruction. The quality of collateral circulation has been established as a key factor in determining the likelihood of a favorable clinical outcome and goes a long way to determining the choice of a stroke care model. Though many imaging and grading methods exist for quantifying collateral blood flow, the actual grading is mostly done through manual inspection. This approach is associated with a number of challenges. First, it is time-consuming. Second, there is a high tendency for bias and inconsistency in the final grade assigned to a patient depending on the experience level of the clinician. We present a multi-stage deep learning approach to predict collateral flow grading in stroke patients based on radiomic features extracted from MR perfusion data. First, we formulate a region of interest detection task as a reinforcement learning problem and train a deep learning network to automatically detect the occluded region within the 3D MR perfusion volumes. Second, we extract radiomic features from the obtained region of interest through local image descriptors and denoising auto-encoders. Finally, we apply a convolutional neural network and other machine learning classifiers to the extracted radiomic features to automatically predict the collateral flow grading of the given patient volume as one of three severity classes - no flow (0), moderate flow (1), and good flow (2). Results from our experiments show an overall accuracy of 72% in the three-class prediction task. With an inter-observer agreement of 16% and a maximum intra-observer agreement of 74% in a similar experiment, our automated deep learning approach demonstrates a performance comparable to expert grading, is faster than visual inspection, and eliminates the problem of grading bias., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2023 Tetteh, Navarro, Meier, Kaesmacher, Paetzold, Kirschke, Zimmer, Wiest and Menze.)

Background: Intratumoral heterogeneity at the cellular and molecular level is a hallmark of glioblastoma (GB) that contributes to treatment resistance and poor clinical outcome. Little is known regarding epigenetic heterogeneity and intratumoral phylogeny and their implication for molecular classification and targeted therapies., Patients and Methods: Multiple tissue biopsies (238 in total) were sampled from 56 newly-diagnosed, treatment-naive GB patients from a prospective in-house cohort and publicly available data and profiled for DNA methylation using the Illumina MethylationEPIC array. Methylation-based classification using the glioma classifier developed by Ceccarelli et al. and estimation of the MGMT promoter methylation status via the MGMT-STP27 model were carried out. In addition, copy number variations (CNVs) and phylogeny were analyzed., Results: Almost half of the patients (22/56, 39%) harbored tumors composed of heterogeneous methylation subtypes. We found two predominant subtype combinations: classic-/mesenchymal-like, and mesenchymal-/pilocytic astrocytoma-like. Nine patients (16%) had tumors composed of subvolumes with and without MGMT promoter methylation, whereas 20 patients (36%) were homogeneously methylated, and 27 patients (48%) were homogeneously unmethylated. CNV analysis revealed high variations in many genes, including CDKN2A/B, EGFR, and PTEN. Phylogenetic analysis correspondingly showed a general pattern of CDKN2A/B loss and gain of EGFR, PDGFRA, and CDK4 during early stages of tumor development., Conclusions: (Epi)genetic intratumoral heterogeneity is a hallmark of GB, both at DNA methylation and CNV level. This intratumoral heterogeneity is of utmost importance for molecular classification as well as for defining therapeutic targets in this disease, as single biopsies might underestimate the true molecular diversity in a tumor., Competing Interests: Disclosure The authors have declared no conflicts of interests., (Copyright © 2022 The Author(s). Published by Elsevier Ltd.. All rights reserved.)

Purpose: Although treatment planning and individualized dose application for emerging prostate-specific membrane antigen (PSMA)-targeted radioligand therapy (RLT) are generally recommended, it is still difficult to implement in practice at the moment. In this study, we aimed to prove the concept of pretherapeutic prediction of dosimetry based on imaging and laboratory measurements before the RLT treatment., Methods: Twenty-three patients with metastatic castration-resistant prostate cancer (mCRPC) treated with 177 Lu-PSMA I&T RLT were included retrospectively. They had available pre-therapy 68  Ga-PSMA-HEBD-CC PET/CT and at least 3 planar and 1 SPECT/CT imaging for dosimetry. Overall, 43 cycles of 177 Lu-PSMA I&T RLT were applied. Organ-based standard uptake values (SUVs) were obtained from pre-therapy PET/CT scans. Patient dosimetry was calculated for the kidney, liver, spleen, and salivary glands using Hermes Hybrid Dosimetry 4.0 from the planar and SPECT/CT images. Machine learning methods were explored for dose prediction from organ SUVs and laboratory measurements. The uncertainty of these dose predictions was compared with the population-based dosimetry estimates. Mean absolute percentage error (MAPE) was used to assess the prediction uncertainty of estimated dosimetry., Results: An optimal machine learning method achieved a dosimetry prediction MAPE of 15.8 ± 13.2% for the kidney, 29.6% ± 13.7% for the liver, 23.8% ± 13.1% for the salivary glands, and 32.1 ± 31.4% for the spleen. In contrast, the prediction based on literature population mean has significantly larger MAPE (p < 0.01), 25.5 ± 17.3% for the kidney, 139.1% ± 111.5% for the liver, 67.0 ± 58.3% for the salivary glands, and 54.1 ± 215.3% for the spleen., Conclusion: The preliminary results confirmed the feasibility of pretherapeutic estimation of treatment dosimetry and its added value to empirical population-based estimation. The exploration of dose prediction may support the implementation of treatment planning for RLT., (© 2022. The Author(s).)

Registration methods facilitate the comparison of multiparametric magnetic resonance images acquired at different stages of brain tumor treatments. Image-based registration solutions are influenced by the sequences chosen to compute the distance measure, and the lack of image correspondences due to the resection cavities and pathological tissues. Nonetheless, an evaluation of the impact of these input parameters on the registration of longitudinal data is still missing. This work evaluates the influence of multiple sequences, namely T1-weighted (T1), T2-weighted (T2), contrast enhanced T1-weighted (T1-CE), and T2 Fluid Attenuated Inversion Recovery (FLAIR), and the exclusion of the pathological tissues on the non-rigid registration of pre- and post-operative images. We here investigate two types of registration methods, an iterative approach and a convolutional neural network solution based on a 3D U-Net. We employ two test sets to compute the mean target registration error (mTRE) based on corresponding landmarks. In the first set, markers are positioned exclusively in the surroundings of the pathology. The methods employing T1-CE achieves the lowest mTREs, with a improvement up to 0.8 mm for the iterative solution. The results are higher than the baseline when using the FLAIR sequence. When excluding the pathology, lower mTREs are observable for most of the methods. In the second test set, corresponding landmarks are located in the entire brain volumes. Both solutions employing T1-CE obtain the lowest mTREs, with a decrease up to 1.16 mm for the iterative method, whereas the results worsen using the FLAIR. When excluding the pathology, an improvement is observable for the CNN method using T1-CE. Both approaches utilizing the T1-CE sequence obtain the best mTREs, whereas the FLAIR is the least informative to guide the registration process. Besides, the exclusion of pathology from the distance measure computation improves the registration of the brain tissues surrounding the tumor. Thus, this work provides the first numerical evaluation of the influence of these parameters on the registration of longitudinal magnetic resonance images, and it can be helpful for developing future algorithms., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2022 Canalini, Klein, Waldmannstetter, Kofler, Cerri, Hering, Heldmann, Schlaeger, Menze, Wiestler, Kirschke and Hahn.)

Exploiting 4D-flow magnetic resonance imaging (MRI) data to quantify hemodynamics requires an adequate spatio-temporal vector field resolution at a low noise level. To address this challenge, we provide a learned solution to super-resolve in vivo 4D-flow MRI data at a post-processing level. We propose a deep convolutional neural network (CNN) that learns the inter-scale relationship of the velocity vector map and leverages an efficient residual learning scheme to make it computationally feasible. A novel, direction-sensitive, and robust loss function is crucial to learning vector-field data. We present a detailed comparative study between the proposed super-resolution and the conventional cubic B-spline based vector-field super-resolution. Our method improves the peak-velocity to noise ratio of the flow field by 10 and 30% for in vivo cardiovascular and cerebrovascular data, respectively, for 4 × super-resolution over the state-of-the-art cubic B-spline. Significantly, our method offers 10x faster inference over the cubic B-spline. The proposed approach for super-resolution of 4D-flow data would potentially improve the subsequent calculation of hemodynamic quantities., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2022 Shit, Zimmermann, Ezhov, Paetzold, Sanches, Pirkl and Menze.)

Automatic localization and segmentation of organs-at-risk (OAR) in CT are essential pre-processing steps in medical image analysis tasks, such as radiation therapy planning. For instance, the segmentation of OAR surrounding tumors enables the maximization of radiation to the tumor area without compromising the healthy tissues. However, the current medical workflow requires manual delineation of OAR, which is prone to errors and is annotator-dependent. In this work, we aim to introduce a unified 3D pipeline for OAR localization-segmentation rather than novel localization or segmentation architectures. To the best of our knowledge, our proposed framework fully enables the exploitation of 3D context information inherent in medical imaging. In the first step, a 3D multi-variate regression network predicts organs' centroids and bounding boxes. Secondly, 3D organ-specific segmentation networks are leveraged to generate a multi-organ segmentation map. Our method achieved an overall Dice score of 0.9260 ± 0.18% on the VISCERAL dataset containing CT scans with varying fields of view and multiple organs.

Self-supervised pretext tasks have been introduced as an effective strategy when learning target tasks on small annotated data sets. However, while current research focuses on exploring novel pretext tasks for meaningful and reusable representation learning for the target task, the study of its robustness and generalizability has remained relatively under-explored. Specifically, it is crucial in medical imaging to proactively investigate performance under different perturbations for reliable deployment of clinical applications. In this work, we revisit medical imaging networks pre-trained with self-supervised learnings and categorically evaluate robustness and generalizability compared to vanilla supervised learning. Our experiments on pneumonia detection in X-rays and multi-organ segmentation in CT yield conclusive results exposing the hidden benefits of self-supervision pre-training for learning robust feature representations.

Background: Hyperpolarization enhances the sensitivity of nuclear magnetic resonance experiments by between four and five orders of magnitude. Several hyperpolarized sensor molecules have been introduced that enable high sensitivity detection of metabolism and physiological parameters. However, hyperpolarized magnetic resonance spectroscopy imaging (MRSI) often suffers from poor signal-to-noise ratio and spectral analysis is complicated by peak overlap. Here, we study measurements of extracellular pH (pH e ) by hyperpolarized zymonic acid, where multiple pH e compartments, such as those observed in healthy kidney or other heterogeneous tissue, result in a cluster of spectrally overlapping peaks, which is hard to resolve with conventional spectroscopy analysis routines., Methods: We investigate whether deep learning methods can yield improved pH e prediction in hyperpolarized zymonic acid spectra of multiple pH e compartments compared to conventional line fitting. As hyperpolarized 13 C-MRSI data sets are often small, a convolutional neural network (CNN) and a multilayer perceptron (MLP) were trained with either a synthetic or a mixed (synthetic and augmented) data set of acquisitions from the kidneys of healthy mice., Results: Comparing the networks' performances compartment-wise on a synthetic test data set and eight real kidney data shows superior performance of CNN compared to MLP and equal or superior performance compared to conventional line fitting. For correct prediction of real kidney pH e values, training with a mixed data set containing only 0.5% real data shows a large improvement compared to training with synthetic data only. Using a manual segmentation approach, pH maps of kidney compartments can be improved by neural network predictions for voxels including three pH compartments., Conclusion: The results of this study indicate that CNNs offer a reliable, accurate, fast and non-interactive method for analysis of hyperpolarized 13 C MRS and MRSI data, where low amounts of acquired data can be complemented to achieve suitable network training., (© 2022. The Author(s).)

Voluntary and involuntary patient motion is a major problem for data quality in clinical routine of Magnetic Resonance Imaging (MRI). It has been thoroughly investigated and, yet it still remains unresolved. In quantitative MRI, motion artifacts impair the entire temporal evolution of the magnetization and cause errors in parameter estimation. Here, we present a novel strategy based on residual learning for retrospective motion correction in fast 3D whole-brain multiparametric MRI. We propose a 3D multiscale convolutional neural network (CNN) that learns the non-linear relationship between the motion-affected quantitative parameter maps and the residual error to their motion-free reference. For supervised model training, despite limited data availability, we propose a physics-informed simulation to generate self-contained paired datasets from a priori motion-free data. We evaluate motion-correction performance of the proposed method for the example of 3D Quantitative Transient-state Imaging at 1.5T and 3T. We show the robustness of the motion correction for various motion regimes and demonstrate the generalization capabilities of the residual CNN in terms of real-motion in vivo data of healthy volunteers and clinical patient cases, including pediatric and adult patients with large brain lesions. Our study demonstrates that the proposed motion correction outperforms current state of the art, reliably providing a high, clinically relevant image quality for mild to pronounced patient movements. This has important implications in clinical setups where large amounts of motion affected data must be discarded as they are rendered diagnostically unusable., Competing Interests: Declaration of Competing Interest Carolin M. Pirkl and Marion I. Menzel are employees at GE Healthcare, Munich, Germany. Matteo Cencini and Michela Tosetti receive research funding from GE Healthcare. All other authors declare that they do not have any financial or non-financial conflict of interests., (Copyright © 2022. Published by Elsevier B.V.)

An important factor for the development of spinal degeneration, pain and the outcome of spinal surgery is known to be the balance of the spine. It must be analyzed in an upright, standing position to ensure physiological loading conditions and visualize load-dependent deformations. Despite the complex 3D shape of the spine, this analysis is currently performed using 2D radiographs, as all frequently used 3D imaging techniques require the patient to be scanned in a prone position. To overcome this limitation, we propose a deep neural network to reconstruct the 3D spinal pose in an upright standing position, loaded naturally. Specifically, we propose a novel neural network architecture, which takes orthogonal 2D radiographs and infers the spine's 3D posture using vertebral shape priors. In this work, we define vertebral shape priors using an atlas and a spine shape prior, incorporating both into our proposed network architecture. We validate our architecture on digitally reconstructed radiographs, achieving a 3D reconstruction Dice of 0.95, indicating an almost perfect 2D-to-3D domain translation. Validating the reconstruction accuracy of a 3D standing spine on real data is infeasible due to the lack of a valid ground truth. Hence, we design a novel experiment for this purpose, using an orientation invariant distance metric, to evaluate our model's ability to synthesize full-3D, upright, and patient-specific spine models. We compare the synthesized spine shapes from clinical upright standing radiographs to the same patient's 3D spinal posture in the prone position from CT.

A multitude of image-based machine learning segmentation and classification algorithms has recently been proposed, offering diagnostic decision support for the identification and characterization of glioma, Covid-19 and many other diseases. Even though these algorithms often outperform human experts in segmentation tasks, their limited reliability, and in particular the inability to detect failure cases, has hindered translation into clinical practice. To address this major shortcoming, we propose an unsupervised quality estimation method for segmentation ensembles. Our primitive solution examines discord in binary segmentation maps to automatically flag segmentation results that are particularly error-prone and therefore require special assessment by human readers. We validate our method both on segmentation of brain glioma in multi-modal magnetic resonance - and of lung lesions in computer tomography images. Additionally, our method provides an adaptive prioritization mechanism to maximize efficacy in use of human expert time by enabling radiologists to focus on the most difficult, yet important cases while maintaining full diagnostic autonomy. Our method offers an intuitive and reliable uncertainty estimation from segmentation ensembles and thereby closes an important gap toward successful translation of automatic segmentation into clinical routine., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2021 Kofler, Ezhov, Fidon, Pirkl, Paetzold, Burian, Pati, El Husseini, Navarro, Shit, Kirschke, Bakas, Zimmer, Wiestler and Menze.)

Purpose: To evaluate diagnostic accuracy of fully automated analysis of multimodal imaging data using [ 18 F]-FET-PET and MRI (including amide proton transfer-weighted (APTw) imaging and dynamic-susceptibility-contrast (DSC) perfusion) in differentiation of tumor progression from treatment-related changes in patients with glioma., Material and Methods: At suspected tumor progression, MRI and [ 18 F]-FET-PET data as part of a retrospective analysis of an observational cohort of 66 patients/74 scans (51 glioblastoma and 23 lower-grade-glioma, 8 patients included at two different time points) were automatically segmented into necrosis, FLAIR-hyperintense, and contrast-enhancing areas using an ensemble of deep learning algorithms. In parallel, previous MR exam was processed in a similar way to subtract preexisting tumor areas and focus on progressive tumor only. Within these progressive areas, intensity statistics were automatically extracted from [ 18 F]-FET-PET, APTw, and DSC-derived cerebral-blood-volume (CBV) maps and used to train a Random Forest classifier with threefold cross-validation. To evaluate contribution of the imaging modalities to the classifier's performance, impurity-based importance measures were collected. Classifier performance was compared with radiology reports and interdisciplinary tumor board assessments., Results: In 57/74 cases (77%), tumor progression was confirmed histopathologically (39 cases) or via follow-up imaging (18 cases), while remaining 17 cases were diagnosed as treatment-related changes. The classification accuracy of the Random Forest classifier was 0.86, 95% CI 0.77-0.93 (sensitivity 0.91, 95% CI 0.81-0.97; specificity 0.71, 95% CI 0.44-0.9), significantly above the no-information rate of 0.77 (p = 0.03), and higher compared to an accuracy of 0.82 for MRI (95% CI 0.72-0.9), 0.81 for [ 18 F]-FET-PET (95% CI 0.7-0.89), and 0.81 for expert consensus (95% CI 0.7-0.89), although these differences were not statistically significant (p > 0.1 for all comparisons, McNemar test). [ 18 F]-FET-PET hot-spot volume was single-most important variable, with relevant contribution from all imaging modalities., Conclusion: Automated, joint image analysis of [ 18 F]-FET-PET and advanced MR imaging techniques APTw and DSC perfusion is a promising tool for objective response assessment in gliomas., (© 2021. The Author(s).)

Purpose: Advanced MRI-based biomarkers offer comprehensive and quantitative information for the evaluation and characterization of brain tumors. In this study, we report initial clinical experience in routine glioma imaging with a novel, fully 3D multiparametric quantitative transient-state imaging (QTI) method for tissue characterization based on T1 and T2 values., Methods: To demonstrate the viability of the proposed 3D QTI technique, nine glioma patients (grade II-IV), with a variety of disease states and treatment histories, were included in this study. First, we investigated the feasibility of 3D QTI (6:25 min scan time) for its use in clinical routine imaging, focusing on image reconstruction, parameter estimation, and contrast-weighted image synthesis. Second, for an initial assessment of 3D QTI-based quantitative MR biomarkers, we performed a ROI-based analysis to characterize T1 and T2 components in tumor and peritumoral tissue., Results: The 3D acquisition combined with a compressed sensing reconstruction and neural network-based parameter inference produced parametric maps with high isotropic resolution (1.125 × 1.125 × 1.125 mm 3 voxel size) and whole-brain coverage (22.5 × 22.5 × 22.5 cm 3 FOV), enabling the synthesis of clinically relevant T1-weighted, T2-weighted, and FLAIR contrasts without any extra scan time. Our study revealed increased T1 and T2 values in tumor and peritumoral regions compared to contralateral white matter, good agreement with healthy volunteer data, and high inter-subject consistency., Conclusion: 3D QTI demonstrated comprehensive tissue assessment of tumor substructures captured in T1 and T2 parameters. Aiming for fast acquisition of quantitative MR biomarkers, 3D QTI has potential to improve disease characterization in brain tumor patients under tight clinical time-constraints., (© 2021. The Author(s).)

Vertebral labelling and segmentation are two fundamental tasks in an automated spine processing pipeline. Reliable and accurate processing of spine images is expected to benefit clinical decision support systems for diagnosis, surgery planning, and population-based analysis of spine and bone health. However, designing automated algorithms for spine processing is challenging predominantly due to considerable variations in anatomy and acquisition protocols and due to a severe shortage of publicly available data. Addressing these limitations, the Large Scale Vertebrae Segmentation Challenge (VerSe) was organised in conjunction with the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) in 2019 and 2020, with a call for algorithms tackling the labelling and segmentation of vertebrae. Two datasets containing a total of 374 multi-detector CT scans from 355 patients were prepared and 4505 vertebrae have individually been annotated at voxel level by a human-machine hybrid algorithm (https://osf.io/nqjyw/, https://osf.io/t98fz/). A total of 25 algorithms were benchmarked on these datasets. In this work, we present the results of this evaluation and further investigate the performance variation at the vertebra level, scan level, and different fields of view. We also evaluate the generalisability of the approaches to an implicit domain shift in data by evaluating the top-performing algorithms of one challenge iteration on data from the other iteration. The principal takeaway from VerSe: the performance of an algorithm in labelling and segmenting a spine scan hinges on its ability to correctly identify vertebrae in cases of rare anatomical variations. The VerSe content and code can be accessed at: https://github.com/anjany/verse., 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 © 2021 Elsevier B.V. All rights reserved.)

The aim of this paper is to provide a comprehensive overview of the MICCAI 2020 AutoImplant Challenge. The approaches and publications submitted and accepted within the challenge will be summarized and reported, highlighting common algorithmic trends and algorithmic diversity. Furthermore, the evaluation results will be presented, compared and discussed in regard to the challenge aim: seeking for low cost, fast and fully automated solutions for cranial implant design. Based on feedback from collaborating neurosurgeons, this paper concludes by stating open issues and post-challenge requirements for intra-operative use. The codes can be found at https://github.com/Jianningli/tmi.

Background: In patients with soft-tissue sarcomas, tumor grading constitutes a decisive factor to determine the best treatment decision. Tumor grading is obtained by pathological work-up after focal biopsies. Deep learning (DL)-based imaging analysis may pose an alternative way to characterize STS tissue. In this work, we sought to non-invasively differentiate tumor grading into low-grade (G1) and high-grade (G2/G3) STS using DL techniques based on MR-imaging., Methods: Contrast-enhanced T1-weighted fat-saturated (T1FSGd) MRI sequences and fat-saturated T2-weighted (T2FS) sequences were collected from two independent retrospective cohorts (training: 148 patients, testing: 158 patients). Tumor grading was determined following the French Federation of Cancer Centers Sarcoma Group in pre-therapeutic biopsies. DL models were developed using transfer learning based on the DenseNet 161 architecture., Results: The T1FSGd and T2FS-based DL models achieved area under the receiver operator characteristic curve (AUC) values of 0.75 and 0.76 on the test cohort, respectively. T1FSGd achieved the best F1-score of all models (0.90). The T2FS-based DL model was able to significantly risk-stratify for overall survival. Attention maps revealed relevant features within the tumor volume and in border regions., Conclusions: MRI-based DL models are capable of predicting tumor grading with good reproducibility in external validation.

We propose a dictionary-matching-free pipeline for multi-parametric quantitative MRI image computing. Our approach has two stages based on compressed sensing reconstruction and deep learned quantitative inference. The reconstruction phase is convex and incorporates efficient spatiotemporal regularisations within an accelerated iterative shrinkage algorithm. This minimises the under-sampling (aliasing) artefacts from aggressively short scan times. The learned quantitative inference phase is purely trained on physical simulations (Bloch equations) that are flexible for producing rich training samples. We propose a deep and compact encoder-decoder network with residual blocks in order to embed Bloch manifold projections through multi-scale piecewise affine approximations, and to replace the non-scalable dictionary-matching baseline. Tested on a number of datasets we demonstrate effectiveness of the proposed scheme for recovering accurate and consistent quantitative information from novel and aggressively subsampled 2D/3D quantitative MRI acquisition protocols., Competing Interests: Declaration of Competing Interest Authors declare that they have no conflict of interest., (Copyright © 2020. Published by Elsevier B.V.)

Nucleosomes, the fundamental units of chromatin, regulate readout and expression of eukaryotic genomes. Single-molecule experiments have revealed force-induced nucleosome accessibility, but a high-resolution unwrapping landscape in the absence of external forces is currently lacking. Here, we introduce a high-throughput pipeline for the analysis of nucleosome conformations based on atomic force microscopy and automated, multi-parameter image analysis. Our data set of ∼10 000 nucleosomes reveals multiple unwrapping states corresponding to steps of 5 bp DNA. For canonical H3 nucleosomes, we observe that dissociation from one side impedes unwrapping from the other side, but in contrast to force-induced unwrapping, we find only a weak sequence-dependent asymmetry. Notably, centromeric CENP-A nucleosomes do not unwrap anti-cooperatively, in stark contrast to H3 nucleosomes. Finally, our results reconcile previous conflicting findings about the differences in height between H3 and CENP-A nucleosomes. We expect our approach to enable critical insights into epigenetic regulation of nucleosome structure and stability and to facilitate future high-throughput AFM studies that involve heterogeneous nucleoprotein complexes.

The purpose of this study was to analyze size and growth dynamics of focal lesions (FL) as well as to quantify diffuse infiltration (DI) in untreated smoldering multiple myeloma (SMM) patients and correlate those MRI features with timepoint and cause of progression. We investigated 199 whole-body magnetic resonance imaging (wb-MRI) scans originating from longitudinal imaging of 60 SMM patients and 39 computed tomography (CT) scans for corresponding osteolytic lesions (OL) in 17 patients. All FLs >5 mm were manually segmented to quantify volume and growth dynamics, and DI was scored, rating four compartments separately in T1- and fat-saturated T2-weighted images. The majority of patients with at least two FLs showed substantial spatial heterogeneity in growth dynamics. The volume of the largest FL ( p = 0.001, c-index 0.72), the speed of growth of the fastest growing FL ( p = 0.003, c-index 0.75), the DI score (DIS, p = 0.014, c-index 0.67), and its dynamic over time (DIS dynamic, p < 0.001, c-index 0.67) all significantly correlated with the time to progression. Size and growth dynamics of FLs correlated significantly with presence/appearance of OL in CT within 2 years after the respective MRI assessment ( p = 0.016 and p = 0.022). DIS correlated with decrease of hemoglobin ( p < 0.001). In conclusion, size and growth dynamics of FLs correlate with prognosis and local bone destruction. Connections between MRI findings and progression patterns (fast growing FL-OL; DIS-hemoglobin decrease) might enable more precise diagnostic and therapeutic approaches for SMM patients in the future.

Layered, two-dimensional (2D) materials are promising for next-generation photonics devices. Typically, the thickness of mechanically cleaved flakes and chemical vapor deposited thin films is distributed randomly over a large area, where accurate identification of atomic layer numbers is time-consuming. Hyperspectral imaging microscopy yields spectral information that can be used to distinguish the spectral differences of varying thickness specimens. However, its spatial resolution is relatively low due to the spectral imaging nature. In this work, we present a 3D deep learning solution called DALM (deep-learning-enabled atomic layer mapping) to merge hyperspectral reflection images (high spectral resolution) and RGB images (high spatial resolution) for the identification and segmentation of MoS 2 flakes with mono-, bi-, tri-, and multilayer thicknesses. DALM is trained on a small set of labeled images, automatically predicts layer distributions and segments individual layers with high accuracy, and shows robustness to illumination and contrast variations. Further, we show its advantageous performance over the state-of-the-art model that is solely based on RGB microscope images. This AI-supported technique with high speed, spatial resolution, and accuracy allows for reliable computer-aided identification of atomically thin materials.

Resting-state functional MRI (rs-fMRI) has provided important insights into brain physiology. It has become an increasingly popular method for presurgical mapping, as an alternative to task-based functional MRI wherein the subject performs a task while being scanned. However, there is no commonly acknowledged gold standard approach for detecting eloquent brain areas using rs-fMRI data in clinical settings. In this study, a general linear model-based machine learning (GLM-ML) approach was tested to predict individual motor task activation based on rs-fMRI data. Its accuracy was then compared to a conventional independent component analysis (ICA) approach. 47 healthy subjects were scanned using resting state, active and passive motor task fMRI experiments using a clinically applicable low-resolution fMRI protocol. The model was trained to associate rs-fMRI network maps with that of hand movement task fMRI, then used to predict task activation maps for unseen subjects solely based on their rs-fMRI data. Our results showed that the GLM-ML approach can accurately predict individual differences in task activation using rs-fMRI data and outperform conventional ICA to detect task activation in the primary sensorimotor region. Furthermore, the predicted activation maps using the GLM -ML model matched well with the activation of passive hand movement fMRI on an individual basis. These results suggest that GLM-ML approach can robustly predict individual differences of task activation based on conventional low-resolution rs-fMRI data and has important implications for future clinical applications.

We present DeepVesselNet, an architecture tailored to the challenges faced when extracting vessel trees and networks and corresponding features in 3-D angiographic volumes using deep learning. We discuss the problems of low execution speed and high memory requirements associated with full 3-D networks, high-class imbalance arising from the low percentage (<3%) of vessel voxels, and unavailability of accurately annotated 3-D training data-and offer solutions as the building blocks of DeepVesselNet. First, we formulate 2-D orthogonal cross-hair filters which make use of 3-D context information at a reduced computational burden. Second, we introduce a class balancing cross-entropy loss function with false-positive rate correction to handle the high-class imbalance and high false positive rate problems associated with existing loss functions. Finally, we generate a synthetic dataset using a computational angiogenesis model capable of simulating vascular tree growth under physiological constraints on local network structure and topology and use these data for transfer learning. We demonstrate the performance on a range of angiographic volumes at different spatial scales including clinical MRA data of the human brain, as well as CTA microscopy scans of the rat brain. Our results show that cross-hair filters achieve over 23% improvement in speed, lower memory footprint, lower network complexity which prevents overfitting and comparable accuracy that does not differ from full 3-D filters. Our class balancing metric is crucial for training the network, and transfer learning with synthetic data is an efficient, robust, and very generalizable approach leading to a network that excels in a variety of angiography segmentation tasks. We observe that sub-sampling and max pooling layers may lead to a drop in performance in tasks that involve voxel-sized structures. To this end, the DeepVesselNet architecture does not use any form of sub-sampling layer and works well for vessel segmentation, centerline prediction, and bifurcation detection. We make our synthetic training data publicly available, fostering future research, and serving as one of the first public datasets for brain vessel tree segmentation and analysis., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2020 Tetteh, Efremov, Forkert, Schneider, Kirschke, Weber, Zimmer, Piraud and Menze.)

Whole-body imaging of mice is a key source of information for research. Organ segmentation is a prerequisite for quantitative analysis but is a tedious and error-prone task if done manually. Here, we present a deep learning solution called AIMOS that automatically segments major organs (brain, lungs, heart, liver, kidneys, spleen, bladder, stomach, intestine) and the skeleton in less than a second, orders of magnitude faster than prior algorithms. AIMOS matches or exceeds the segmentation quality of state-of-the-art approaches and of human experts. We exemplify direct applicability for biomedical research for localizing cancer metastases. Furthermore, we show that expert annotations are subject to human error and bias. As a consequence, we show that at least two independently created annotations are needed to assess model performance. Importantly, AIMOS addresses the issue of human bias by identifying the regions where humans are most likely to disagree, and thereby localizes and quantifies this uncertainty for improved downstream analysis. In summary, AIMOS is a powerful open-source tool to increase scalability, reduce bias, and foster reproducibility in many areas of biomedical research.

This study aims to develop a silent, fast and 3D method for T1 and proton density (PD) mapping, while generating time series of T1-weighted (T1w) images with bias-field correction. Undersampled T1w images at different effective inversion times (TIs) were acquired using the inversion recovery prepared RUFIS sequence with an interleaved k-space trajectory. Unaliased images were reconstructed by constraining the signal evolution to a temporal subspace which was learned from the signal model. Parameter maps were obtained by fitting the data to the signal model, and bias-field correction was conducted on T1w images. Accuracy and repeatability of the method was accessed in repeated experiments with phantom and volunteers. For the phantom study, T1 values obtained by the proposed method were highly consistent with values from the gold standard method, R 2 = 0.9976. Coefficients of variation (CVs) ranged from 0.09% to 0.83%. For the volunteer study, T1 values from gray and white matter regions were consistent with literature values, and peaks of gray and white matter can be clearly delineated on whole-brain T1 histograms. CVs ranged from 0.01% to 2.30%. The acoustic noise measured at the scanner isocenter was 2.6 dBA higher compared to the in-bore background. Rapid and with low acoustic noise, the proposed method is shown to produce accurate T1 and PD maps with high repeatability by reconstructing sparsely sampled T1w images at different TIs using temporal subspace. Our approach can greatly enhance patient comfort during examination and therefore increase the acceptance of the procedure.

The purpose of this study was to assess how different MRI protocols (spinal vs. spinal plus pelvic vs. whole-body (wb)-MRI) affect staging in patients with smoldering multiple myeloma (SMM), according to the SLiM-CRAB-criterion '>1 focal lesion (FL) in MRI'. In this retrospective study, a baseline cohort of 147 SMM patients with wb-MRI at initial diagnosis was investigated, including prognostic data regarding development of CRAB-criteria. Fifty-two patients formed a follow-up cohort with a median of three wb-MRIs. The locations of all FLs were determined and it was calculated how staging decisions regarding the criterion '>1 FL in MRI' would have been made if only a limited anatomic area (spine vs. spine plus pelvis) would have been covered by the MRI protocol. Furthermore, subgroups of patients selected by different cutoff-protocol-combinations were compared regarding their prognosis for development of CRAB-criteria. With an MRI protocol limited to spine/spine plus pelvis, only 28%/64% of patients who actually had >1 FL in wb-MRI would have been rated correctly as having '>1 FL in MRI'. Fifty-four percent/36% of patients with exactly 1 FL in spine/spine plus pelvis revealed >1 FL when the entire wb-MRI was analyzed. During follow-up, four more patients developed >1 FL in wb-MRI; both limited MRI protocols would have detected only one of these four patients as having >1 FL at the correct timepoint. Having >1 FL in spine/in spine plus pelvis/in the whole body was associated with a 43%/57%/49% probability of developing CRAB-criteria within 2 years. Patients with >3 FL in spine plus pelvis and patients with >4 FL in the whole body had an 80% probability to develop CRAB-criteria within 2 years. MRI protocols limited to the spine or to spine plus pelvis lead to substantial underdiagnoses of patients who actually have >1 FL in wb-MRI at baseline and during follow-up, which influences staging and treatment decisions according to the current SLiM-CRAB criteria. However, given the spatial distribution of FLs and the analysis on clinical course of patients indicates that the cutoff for the number of FLs should be adopted according to the MRI protocol when using MRI for staging in SMM.

Novel methods for quantitative, transient-state multiparametric imaging are increasingly being demonstrated for assessment of disease and treatment efficacy. Here, we build on these by assessing the most common Non-Cartesian readout trajectories (2D/3D radials and spirals), demonstrating efficient anti-aliasing with a k-space view-sharing technique, and proposing novel methods for parameter inference with neural networks that incorporate the estimation of proton density. Our results show good agreement with gold standard and phantom references for all readout trajectories at 1.5 T and 3 T. Parameters inferred with the neural network were within 6.58% difference from the parameters inferred with a high-resolution dictionary. Concordance correlation coefficients were above 0.92 and the normalized root mean squared error ranged between 4.2 and 12.7% with respect to gold-standard phantom references for T1 and T2. In vivo acquisitions demonstrate sub-millimetric isotropic resolution in under five minutes with reconstruction and inference times < 7 min. Our 3D quantitative transient-state imaging approach could enable high-resolution multiparametric tissue quantification within clinically acceptable acquisition and reconstruction times.

Imaging techniques such as computed tomographies (CT) play a major role in clinical imaging and diagnosis of malignant lesions. In recent years, metal nanoparticle platforms enabled effective payload delivery for several imaging techniques. Due to the possibility of surface modification, metal nanoparticles are predestined to facilitate molecular tumor targeting. In this work, we demonstrate the feasibility of anti-plasma membrane Heat shock protein 70 (Hsp70) antibody functionalized gold nanoparticles (cmHsp70.1-AuNPs) for tumor-specific multimodal imaging. Membrane-associated Hsp70 is exclusively presented on the plasma membrane of malignant cells of multiple tumor entities but not on corresponding normal cells, predestining this target for a tumor-selective in vivo imaging. In vitro microscopic analysis revealed the presence of cmHsp70.1-AuNPs in the cytosol of tumor cell lines after internalization via the endo-lysosomal pathway. In preclinical models, the biodistribution as well as the intratumoral enrichment of AuNPs were examined 24 h after i.v. injection in tumor-bearing mice. In parallel to spectral CT analysis, histological analysis confirmed the presence of AuNPs within tumor cells. In contrast to control AuNPs, a significant enrichment of cmHsp70.1-AuNPs has been detected selectively inside tumor cells in different tumor mouse models. Furthermore, a machine-learning approach was developed to analyze AuNP accumulations in tumor tissues and organs. In summary, utilizing mHsp70 on tumor cells as a target for the guidance of cmHsp70.1-AuNPs facilitates an enrichment and uniform distribution of nanoparticles in mHsp70-expressing tumor cells that enables various microscopic imaging techniques and spectral-CT-based tumor delineation in vivo.

Despite great advances in brain tumor segmentation and clear clinical need, translation of state-of-the-art computational methods into clinical routine and scientific practice remains a major challenge. Several factors impede successful implementations, including data standardization and preprocessing. However, these steps are pivotal for the deployment of state-of-the-art image segmentation algorithms. To overcome these issues, we present BraTS Toolkit. BraTS Toolkit is a holistic approach to brain tumor segmentation and consists of three components: First, the BraTS Preprocessor facilitates data standardization and preprocessing for researchers and clinicians alike. It covers the entire image analysis workflow prior to tumor segmentation, from image conversion and registration to brain extraction. Second, BraTS Segmentor enables orchestration of BraTS brain tumor segmentation algorithms for generation of fully-automated segmentations. Finally, Brats Fusionator can combine the resulting candidate segmentations into consensus segmentations using fusion methods such as majority voting and iterative SIMPLE fusion. The capabilities of our tools are illustrated with a practical example to enable easy translation to clinical and scientific practice., (Copyright © 2020 Kofler, Berger, Waldmannstetter, Lipkova, Ezhov, Tetteh, Kirschke, Zimmer, Wiestler and Menze.)

Purpose: To use and test a labeling algorithm that operates on two-dimensional reformations, rather than three-dimensional data to locate and identify vertebrae., Materials and Methods: The authors improved the Btrfly Net, a fully convolutional network architecture described by Sekuboyina et al, which works on sagittal and coronal maximum intensity projections (MIPs) and augmented it with two additional components: spine localization and adversarial a priori learning. Furthermore, two variants of adversarial training schemes that incorporated the anatomic a priori knowledge into the Btrfly Net were explored. The superiority of the proposed approach for labeling vertebrae on three datasets was investigated: a public benchmarking dataset of 302 CT scans and two in-house datasets with a total of 238 CT scans. The Wilcoxon signed rank test was employed to compute the statistical significance of the improvement in performance observed with various architectural components in the authors' approach., Results: On the public dataset, the authors' approach using the described Btrfly Net with energy-based prior encoding (Btrfly pe-eb ) network performed as well as current state-of-the-art methods, achieving a statistically significant ( P < .001) vertebrae identification rate of 88.5% ± 0.2 (standard deviation) and localization distances of less than 7 mm. On the in-house datasets that had a higher interscan data variability, an identification rate of 85.1% ± 1.2 was obtained., Conclusion: An identification performance comparable to existing three-dimensional approaches was achieved when labeling vertebrae on two-dimensional MIPs. The performance was further improved using the proposed adversarial training regimen that effectively enforced local spine a priori knowledge during training. Spine localization increased the generalizability of our approach by homogenizing the content in the MIPs. Supplemental material is available for this article. © RSNA, 2020., Competing Interests: Disclosures of Conflicts of Interest: A.S. Activities related to the present article: grant/grants pending and travel support from the European Research Council relationships. Activities not related to the present article: disclosed no relevant relationships. Other relationships: disclosed no relevant relationships. M.R. Activities related to the present article: German Excellence Initiative (and the European Union Seventh Framework Program under grant agreement n291763. Activities not related to the present article: currently employed by the Friedrich Miescher Institute for Biomedical Research. Other relationships: disclosed no relevant relationships. A.V. disclosed no relevant relationships. B.H.M. disclosed no relevant relationships. J.S.K. Activities related to the present article: grant from European Research Council StG iback and Nvidia. Activities not related to the present article: payment for lectures from Philips Healthcare. Other relationships: disclosed no relevant relationships., (2020 by the Radiological Society of North America, Inc.)

The advent of medical imaging and automatic image analysis is bringing the full quantitative assessment of lesions and tumor burden at every clinical examination within reach. This opens avenues for the development and testing of functional disease models, as well as their use in the clinical practice for personalized medicine. In this paper, we introduce a Bayesian statistical framework, based on mixed-effects models, to quantitatively test and learn functional disease models at different scales, on population longitudinal data. We also derive an effective mathematical model for the crossover between initially detected lesions and tumor dissemination, based on the Iwata-Kawasaki-Shigesada model. We finally propose to leverage this descriptive disease progression model into model-aware biomarkers for personalized risk-assessment, taking all available examinations and relevant covariates into account. As a use case, we study Multiple Myeloma, a disseminated plasma cell cancer, in which proper diagnostics is essential, to differentiate frequent precursor state without end-organ damage from the rapidly developing disease requiring therapy. After learning the best biological models for local lesion growth and global tumor burden evolution on clinical data, and computing corresponding population priors, we use individual model parameters as biomarkers, and can study them systematically for correlation with external covariates, such as sex or location of the lesion. On our cohort of 63 patients with smoldering Multiple Myeloma, we show that they perform substantially better than other radiological criteria, to predict progression into symptomatic Multiple Myeloma. Our study paves the way for modeling disease progression patterns for Multiple Myeloma, but also for other metastatic and disseminated tumor growth processes, and for analyzing large longitudinal image data sets acquired in oncological imaging. It shows the unprecedented potential of model-based biomarkers for better and more personalized treatment decisions and deserves being validated on larger cohorts to establish its role in clinical decision making., (Copyright © 2019 Elsevier B.V. All rights reserved.)

Purpose: Glioblastoma is routinely treated by concomitant radiochemotherapy. Current target definition guidelines use anatomic MRI (magnetic resonance imaging) scans, taking into account contrast enhancement and the rather unspecific hyperintensity on the fluid-attenuated inversion recovery (FLAIR) sequence., Methods and Materials: We applied deep learning based free water correction of diffusion tensor imaging (DTI) scans to estimate the infiltrative gross tumor volume (iGTV) inside of the FLAIR hyperintense region. We analyzed the resulting iGTVs and their impact on target volume definition in a retrospective cohort of 33 GBM patients., Results: iGTVs were significantly smaller compared to standard pre- and post-operative gross tumor volume (GTV) definitions. Two novel infiltrative tumor GTVs (nGTV PRE-OP and nGTV POST-OP ) defined as the conjunction volume of the standard GTV and the iGTV showed only a moderate increase in size compared to standard GTV definitions. On postoperative scans, the iGTV was predominantly covered by the two clinical target volume (CTV) concepts CTV EORTC and CTV ROTG1 . A novel infiltrative tumor CTV (nCTV) [nGTV POST-OP  + 2 cm margin] was significantly smaller compared to CTV ROTG1 but larger than CTV EORTC . The overlap volume and conformity index demonstrated a distinct spatial configuration of the nCTV. Tumor recurrences overlapped with the iGTV in all but one patients and were completely covered by the nCTV in all patients. After reducing the margin to 1 cm recurrences coverage was at least in-field in all patients., Conclusion: To conclude, free water corrected DTI scans may help to define infiltrative tumor areas of GBM that could ultimately be used to individualize RT treatment planning in terms of dose sparing or dose escalation., (Copyright © 2019 Elsevier B.V. All rights reserved.)

Magnetic resonance imaging (MRI) has evolved into an outstandingly versatile diagnostic modality, as it has the ability to non-invasively produce detailed information on a tissue's structure and function. Complementary data is normally obtained in separate measurements, either as contrast-weighted images, which are fast and simple to acquire, or as quantitative parametric maps, which offer an absolute quantification of underlying biophysical effects, such as relaxation times or flow. Here, we demonstrate how to acquire and reconstruct data in a transient-state with a dual purpose: 1 - to generate contrast-weighted images that can be adjusted to emphasise clinically relevant image biomarkers; exemplified with signal modulation according to flow to obtain angiography information, and 2 - to simultaneously infer multiple quantitative parameters with a single, highly accelerated acquisition. This is achieved by introducing three novel elements: a model that accounts for flowing blood, a method for sequence design using smooth flip angle excitation patterns that incorporates both parameter encoding and signal contrast, and the reconstruction of temporally resolved contrast-weighted images. From these images we simultaneously obtain angiography projections and multiple quantitative maps. By doing so, we increase the amount of clinically relevant data without adding measurement time, creating new dimensions for biomarker exploration and adding value to MR examinations for patients and clinicians alike.

Purpose: The in vivo probing of restricted diffusion effects in large lipid droplets on a clinical MR scanner remains a major challenge due to the need for high b-values and long diffusion times. This work proposes a methodology to probe mean lipid droplet sizes using diffusion-weighted MRS (DW-MRS) at 3T., Methods: An analytical expression for restricted diffusion was used. Simulations were performed to evaluate the noise performance and the influence of particle size distribution. To validate the method, oil-in-water emulsions were prepared and examined using DW-MRS, laser deflection and light microscopy. The tibia bone marrow was scanned in volunteers to test the method repeatability and characterize microstructural differences at different locations., Results: The simulations showed accurate and precise droplet size estimation when a sufficient SNR is reached with minor dependence on the size distribution. In phantoms, a good correlation between the measured droplet sizes by DW-MRS and by laser deflection (R 2 = 0.98; P = 0.01) and microscopy (R 2 = 0.99; P < 0.01) measurements was obtained. A mean coefficient of variation of 11.5 % was found for the lipid droplet diameter in vivo. The average diameter was smaller at a proximal (50.1 ± 7.3 µm) compared with a distal tibia location (61.1 ± 6.8 µm) (P < 0.01)., Conclusion: The presented methods were able to probe restricted diffusion effects in lipid droplets using DW-MRS and to estimate lipid droplet size. The methodology was validated using phantoms and the in vivo feasibility in bone marrow was shown based on a good repeatability and findings in agreement with literature., (© 2019 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine.)

In the original version of this Article the values in the rightmost column of Table 1 were inadvertently shifted relative to the other columns. This has now been corrected in the PDF and HTML versions of the Article.

Background and Purpose: The T1-weighted dynamic contrast enhanced (DCE)-MRI is an imaging technique that provides a quantitative measure of pharmacokinetic (PK) parameters characterizing microvasculature of tissues. For the present study, we propose a new machine learning (ML) based approach to directly estimate the PK parameters from the acquired DCE-MRI image-time series that is both more robust and faster than conventional model fitting. Materials and Methods: We specifically utilize deep convolutional neural networks (CNNs) to learn the mapping between the image-time series and corresponding PK parameters. DCE-MRI datasets acquired from 15 patients with clinically evident mild ischaemic stroke were used in the experiments. Training and testing were carried out based on leave-one-patient-out cross- validation. The parameter estimates obtained by the proposed CNN model were compared against the two tracer kinetic models: (1) Patlak model, (2) Extended Tofts model, where the estimation of model parameters is done via voxelwise linear and nonlinear least squares fitting respectively. Results: The trained CNN model is able to yield PK parameters which can better discriminate different brain tissues, including stroke regions. The results also demonstrate that the model generalizes well to new cases even if a subject specific arterial input function (AIF) is not available for the new data. Conclusion: A ML-based model can be used for direct inference of the PK parameters from DCE image series. This method may allow fast and robust parameter inference in population DCE studies. Parameter inference on a 3D volume-time series takes only a few seconds on a GPU machine, which is significantly faster compared to conventional non-linear least squares fitting.