Brain volume estimation from post-mortem newborn and fetal MRI

CortexODE: Learning Cortical Surface Reconstruction by Neural ODEs

We present CortexODE, a deep learning framework for cortical surface reconstruction. CortexODE leverages neural ordinary differential equations (ODEs) to deform an input surface into a target shape by learning a diffeomorphic flow. The trajectories of the points on the surface are modeled as ODEs, where the derivatives of their coordinates are parameterized via a learnable Lipschitz-continuous deformation network. This provides theoretical guarantees for the prevention of self-intersections. CortexODE can be integrated to an automatic learning-based pipeline, which reconstructs cortical surfaces efficiently in less than 5 seconds. The pipeline utilizes a 3D U-Net to predict a white matter segmentation from brain Magnetic Resonance Imaging (MRI) scans, and further generates a signed distance function that represents an initial surface. Fast topology correction is introduced to guarantee homeomorphism to a sphere. Following the isosurface extraction step, two CortexODE models are trained to deform the initial surface to white matter and pial surfaces respectively. The proposed pipeline is evaluated on large-scale neuroimage datasets in various age groups including neonates (25-45 weeks), young adults (22-36 years) and elderly subjects (55-90 years). Our experiments demonstrate that the CortexODE-based pipeline can achieve less than 0.2mm average geometric error while being orders of magnitude faster compared to conventional processing pipelines.

Fetal Liver Volume Assessment Using Magnetic Resonance Imaging in Fetuses With Cytomegalovirus Infection†

Objective

To assess fetal liver volume (FLV) by magnetic resonance imaging (MRI) in cytomegalovirus (CMV)-infected fetuses compared to a group of healthy fetuses.

Method

Most infected cases were diagnosed by the evidence of ultrasound abnormalities during routine scans and in some after maternal CMV screening. CMV-infected fetuses were considered severely or mildly affected according to prenatal brain lesions identified by ultrasound (US)/MRI. We assessed FLV, the FLV to abdominal circumference (AC) ratio (FLV/AC-ratio), and the FLV to fetal body volume (FBV) ratio (FLV/FBV-ratio). As controls, we included 33 healthy fetuses. Hepatomegaly was evaluated post-mortem in 11 cases of congenital CMV infection. Parametric trend and intraclass correlation analyses were performed.

Results

There were no significant differences in FLV between infected (n = 32) and healthy fetuses. On correcting the FLV for AC and FBV, we observed a significantly higher FLV in CMV-infected fetuses. There were no significant differences in the FLV, or the FLV/AC or FLV/FBV-ratios according to the severity of brain abnormalities. There was excellent concordance between the fetal liver weight estimated by MRI and liver weight obtained post-mortem. Hepatomegaly was not detected in any CMV-infected fetus.

Conclusion

In CMV-infected fetuses, FLV corrected for AC and FBV was higher compared to healthy controls, indicating relative hepatomegaly. These parameters could potentially be used as surrogate markers of liver enlargement.

Fetal postmortem imaging: an overview of current techniques and future perspectives

Fetal death because of miscarriage, unexpected intrauterine fetal demise, or termination of pregnancy is a traumatic event for any family. Despite advances in prenatal imaging and genetic diagnosis, conventional autopsy remains the gold standard because it can provide additional information not available during fetal life in up to 40% of cases and this by itself may change the recurrence risk and hence future counseling for parents. However, conventional autopsy is negatively affected by procedures involving long reporting times because the fetal brain is prone to the effect of autolysis, which may result in suboptimal examinations, particularly of the central nervous system. More importantly, fewer than 50%-60% of parents consent to invasive autopsy, mainly owing to the concerns about body disfigurement. Consequently, this has led to the development of noninvasive perinatal virtual autopsy using imaging techniques. Because a significant component of conventional autopsy involves the anatomic examination of organs, imaging techniques such as magnetic resonance imaging, ultrasound, and computed tomography are possible alternatives. With a parental acceptance rate of nearly 100%, imaging techniques as part of postmortem examination have become widely used in recent years in some countries. Postmortem magnetic resonance imaging using 1.5-Tesla magnets is the most studied technique and offers an overall diagnostic accuracy of 77%-94%. It is probably the best choice as a virtual autopsy technique for fetuses >20 weeks' gestation. However, for fetuses <20 weeks' gestation, its performance is poor. The use of higher magnetic resonance imaging magnetic fields such as 3-Tesla may slightly improve performance. Of note, in cases of fetal maceration, magnetic resonance imaging may offer diagnoses in a proportion of brain lesions wherein conventional autopsy fails. Postmortem ultrasound examination using a high-frequency probe offers overall sensitivity and specificity of 67%-77% and 74%-90%, respectively, with the advantage of easy access and affordability. The main difference between postmortem ultrasound and magnetic resonance imaging relates to their respective abilities to obtain images of sufficient quality for a confident diagnosis. The nondiagnostic rate using postmortem ultrasound ranges from 17% to 30%, depending on the organ examined, whereas the nondiagnostic rate using postmortem magnetic resonance imaging in most situations is far less than 10%. For fetuses ≤20 weeks' gestation, microfocus computed tomography achieves close to 100% agreement with autopsy and is likely to be the technique of the future in this subgroup. The lack of histology has always been listed as 1 limitation of all postmortem imaging techniques. Image-guided needle tissue biopsy coupled with any postmortem imaging can overcome this limitation. In addition to describing the diagnostic accuracy and limitations of each imaging technology, we propose a novel, stepwise diagnostic approach and describe the possible application of these techniques in clinical practice as an alternative or an adjunct or for triage to select cases that would specifically benefit from invasive examination, with the aim of reducing parental distress and pathologist workload. The widespread use of postmortem fetal imaging is inevitable, meaning that hurdles such as specialized training and dedicated financing must be overcome to improve access to these newer, well-validated techniques.

Latest developments in post-mortem foetal imaging

A sustained decline in parental consent rates for perinatal autopsies has driven the development of less-invasive methods for death investigation. A wide variety of imaging modalities have been developed for this purpose and include post-mortem whole body magnetic resonance imaging (MRI), ultrasound, computed tomography (CT) and micro-focus CT techniques. These are also vital for "minimally invasive" methods, which include potential for tissue sampling, such as image guidance for targeted biopsies and laparoscopic-assisted techniques. In this article, we address the range of imaging techniques currently in clinical practice and those under development. Significant advances in high-field MRI and micro-focus CT imaging show particular promise for smaller and earlier gestation foetuses. We also review how MRI biomarkers such as diffusion-weighted imaging and organ volumetric analysis may aid diagnosis and image interpretation in the absence of autopsy data. Three-dimensional printing and augmented reality may help make imaging findings more accessible to parents, colleagues and trainees.

A novel hybrid atlas-free hierarchical graph-based segmentation of newborn brain MRI using wavelet filter banks

Objective: The newborn brain MRI (magnetic resonance imaging) tissue segmentation plays a vital part in assessment of primary brain growth. In the newborn stage (nearly less than 28 days old), in T1- as well as T2-weighted MR images similar levels of intensity are exhibited by WM and GM, makes segmentation of the tissue extremely challenging. In this newborn stage for tissue segmentation, very few methods are developed. Hence the development of accurate brain tissue segmentation of neonate is prime objective of this paper.Methods: In this research work, we propose a novel hybrid atlas-free hierarchical graph-based tissue segmentation method for newborn infants. Wavelet filter banks are a class of deep models wherein filters and local neighborhood processes are used alternately for efficient segmentation on the raw input images, and fuzzy-based SVM (support vector machine) is used for precise tissue classification.Results: Specifically, from T1, T2 images multimodality information are used as inputs and then as outputs the segmentation maps are generated. The proposed approach considerably outperforms preceding methods of tissue segmentation as reflected in results. With this approach, the newborn MRI images that are even suffered from noise, poor resolution or the low contrasted images are also segmented more effectively with precision of 90% and sensitivity 98%.Conclusion: In addition, our findings indicate that the incorporation of multi-modality image led to significant improvements in performance. Thus, the proposed work effectively tackles the unreliability as well as the other issues faced with the prior methodologies with an interactive accurate segmentation outline.

Feasibility of Postmortem Imaging Assessment of Brain: Liver Volume Ratios with Pathological Validation

Introduction

Organ volumes at postmortem magnetic resonance imaging (PMMR) should reflect autopsy organ weights, and thus brain: liver volume ratios on imaging could be a surrogate for weight volume ratios at autopsy to indicate fetal growth restriction (FGR). This study aims to determine whether imaging-based organ volume ratios can replace autopsy organ weight ratios.

Materials and Methods

An unselected cohort of perinatal deaths underwent PMMR prior to autopsy. Semiautomated brain and liver volumes were compared to autopsy organ weights and ratios. Ratios were compared using Bland-Altman plots, and intra- and interobserver variability was assessed.

Results

A total 49 fetuses (25 male, 51%) at 17–42 weeks gestation were ­assessed. There was a reasonable correlation between autopsy-derived brain: liver weight ratios (AB: LwR) and imaging-derived brain: liver volume ratios (IB: LvR; r = 0.8). The mean difference between AB: LwR and IB: LvR was +0.7 (95% limits of agreement range −1.5 to +2.9). In a small subset where FGR was present, the optimal IB: LvR ≥5.5 gave 83.3% sensitivity and 86.0% specificity for diagnosis. There was acceptable agreement within readers (mean difference in IB: LvRs 0.77 ± 2.21) and between readers −0.36 ± 0.68.

Conclusion

IB: LvR provides a surrogate evaluation of AB: LwRs, and may be used as a marker of FGR where autopsy is declined.

Prenatal assessment of cerebellar vermian lobulation: fetal MRI with 3-Tesla postmortem validation

OBJECTIVES

To optimize the imaging assessment of fetal hindbrain malformations, this observational magnetic resonance imaging (MRI) study aimed to assess whether fetal vermian lobulation can be quantified accurately and whether the relative growth of vermian lobules is uniform.

METHODS

This retrospective study included singleton fetuses which underwent T2-weighted MRI in vivo with a 1.5-Tesla (T) scanner or within 24 h postmortem with a 3-T scanner between January 2007 and November 2016 at the Medical University of Vienna. We included only those showing normal structural brain development on ultrasound and MRI and which had image quality appropriate for quantitative analysis, i.e. good image quality and a precise midsagittal slice. Fetal brains were segmented and, for all discernible vermian lobules, we determined the mean relative area contribution (MRAC, the proportion of the lobule relative to the total vermian area, in terms of number of voxels). Inter- and intrarater measurement variability of a representative selection (21 cases) was determined by intraclass correlation coefficient (ICC) for voxel-based differences. A linear regression model was used to assess the correlation between the relative size of each vermian lobule (i.e. MRAC) and gestational age.

RESULTS

A total of 78 fetuses scanned in vivo aged 18-32 gestational weeks and seven fetuses scanned postmortem aged 16-30 weeks had a precise midsagittal slice and image quality sufficient for quantitative analysis. After 22 weeks of gestation, seven of the nine known vermian lobules could be discriminated reliably. The MRAC showed a mean ± SD difference of only 2.89 ± 3.01% between in-vivo and postmortem measurements. The ICC of voxel-based interrater differences was mean ± SD, 0.91 ± 0.05 and the intrarater ICC was 0.95 ± 0.03. Growth of cerebellar lobules was non-uniform: the MRAC of culmen and DFT (declive + folium + tuber) increased with gestational age, whereas that of lingula, centralis, pyramis and nodulus decreased. The growth of the uvula showed no significant correlation with gestational age.

CONCLUSIONS

Fetal vermian lobulation can be assessed accurately and reliably after 22 weeks on precise midsagittal sequences with 1.5-T T2-weighted MRI. Fetal vermian lobules show non-uniform growth, with expansion of DFT and culmen at the expense of the other vermian lobules. Evaluation and elucidation of vermian lobulation in normal fetuses should enable better characterization of fetuses with hindbrain malformations. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.

Complex Trajectories of Brain Development in the Healthy Human Fetus

This study characterizes global and hemispheric brain growth in healthy human fetuses during the second half of pregnancy using three-dimensional MRI techniques. We studied 166 healthy fetuses that underwent MRI between 18 and 39 completed weeks gestation. We created three-dimensional high-resolution reconstructions of the brain and calculated volumes for left and right cortical gray matter (CGM), fetal white matter (FWM), deep subcortical structures (DSS), and the cerebellum. We calculated the rate of growth for each tissue class according to gestational age and described patterns of hemispheric growth. Each brain region demonstrated major increases in volume during the second half of gestation, the most pronounced being the cerebellum (34-fold), followed by FWM (22-fold), CGM (21-fold), and DSS (10-fold). The left cerebellar hemisphere, CGM, and DSS had larger volumes early in gestation, but these equalized by term. It has been increasingly recognized that brain asymmetry evolves throughout the human life span. Advanced quantitative MRI provides noninvasive measurements of early structural asymmetry between the left and right fetal brain that may inform functional and behavioral laterality differences seen in children and young adulthood.

The developing human connectome project: A minimal processing pipeline for neonatal cortical surface reconstruction

The Developing Human Connectome Project (dHCP) seeks to create the first 4-dimensional connectome of early life. Understanding this connectome in detail may provide insights into normal as well as abnormal patterns of brain development. Following established best practices adopted by the WU-MINN Human Connectome Project (HCP), and pioneered by FreeSurfer, the project utilises cortical surface-based processing pipelines. In this paper, we propose a fully automated processing pipeline for the structural Magnetic Resonance Imaging (MRI) of the developing neonatal brain. This proposed pipeline consists of a refined framework for cortical and sub-cortical volume segmentation, cortical surface extraction, and cortical surface inflation, which has been specifically designed to address considerable differences between adult and neonatal brains, as imaged using MRI. Using the proposed pipeline our results demonstrate that images collected from 465 subjects ranging from 28 to 45 weeks post-menstrual age (PMA) can be processed fully automatically; generating cortical surface models that are topologically correct, and correspond well with manual evaluations of tissue boundaries in 85% of cases. Results improve on state-of-the-art neonatal tissue segmentation models and significant errors were found in only 2% of cases, where these corresponded to subjects with high motion. Downstream, these surfaces will enhance comparisons of functional and diffusion MRI datasets, supporting the modelling of emerging patterns of brain connectivity.

The developing human connectome project: A minimal processing pipeline for neonatal cortical surface reconstruction

The Developing Human Connectome Project (dHCP) seeks to create the first 4-dimensional connectome of early life. Understanding this connectome in detail may provide insights into normal as well as abnormal patterns of brain development. Following established best practices adopted by the WU-MINN Human Connectome Project (HCP), and pioneered by FreeSurfer, the project utilises cortical surface-based processing pipelines. In this paper, we propose a fully automated processing pipeline for the structural Magnetic Resonance Imaging (MRI) of the developing neonatal brain. This proposed pipeline consists of a refined framework for cortical and sub-cortical volume segmentation, cortical surface extraction, and cortical surface inflation, which has been specifically designed to address considerable differences between adult and neonatal brains, as imaged using MRI. Using the proposed pipeline our results demonstrate that images collected from 465 subjects ranging from 28 to 45 weeks post-menstrual age (PMA) can be processed fully automatically; generating cortical surface models that are topologically correct, and correspond well with manual evaluations of tissue boundaries in 85% of cases. Results improve on state-of-the-art neonatal tissue segmentation models and significant errors were found in only 2% of cases, where these corresponded to subjects with high motion. Downstream, these surfaces will enhance comparisons of functional and diffusion MRI datasets, supporting the modelling of emerging patterns of brain connectivity.

Diagnostic accuracy of postmortem imaging vs autopsy-A systematic review

Background Postmortem imaging has been used for more than a century as a complement to medico-legal autopsies. The technique has also emerged as a possible alternative to compensate for the continuous decline in the number of clinical autopsies. To evaluate the diagnostic accuracy of postmortem imaging for various types of findings, we performed this systematic literature review. Data sources The literature search was performed in the databases PubMed, Embase and Cochrane Library through January 7, 2015. Relevant publications were assessed for risk of bias using the QUADAS tool and were classified as low, moderate or high risk of bias according to pre-defined criteria. Autopsy and/or histopathology were used as reference standard. Findings The search generated 2600 abstracts, of which 340 were assessed as possibly relevant and read in full-text. After further evaluation 71 studies were finally included, of which 49 were assessed as having high risk of bias and 22 as moderate risk of bias. Due to considerable heterogeneity - in populations, techniques, analyses and reporting - of included studies it was impossible to combine data to get a summary estimate of the diagnostic accuracy of the various findings. Individual studies indicate, however, that imaging techniques might be useful for determining organ weights, and that the techniques seem superior to autopsy for detecting gas Conclusions and Implications In general, based on the current scientific literature, it was not possible to determine the diagnostic accuracy of postmortem imaging and its usefulness in conjunction with, or as an alternative to autopsy. To correctly determine the usefulness of postmortem imaging, future studies need improved planning, improved methodological quality and larger materials, preferentially obtained from multi-center studies.

Improving the clinical management of traumatic brain injury through the pharmacokinetic modeling of peripheral blood biomarkers

Background

Blood biomarkers of neurovascular damage are used clinically to diagnose the presence severity or absence of neurological diseases, but data interpretation is confounded by a limited understanding of their dependence on variables other than the disease condition itself. These include half-life in blood, molecular weight, and marker-specific biophysical properties, as well as the effects of glomerular filtration, age, gender, and ethnicity. To study these factors, and to provide a method for markers’ analyses, we developed a kinetic model that allows the integrated interpretation of these properties.

Methods

The pharmacokinetic behaviors of S100B (monomer and homodimer), Glial Fibrillary Acidic Protein and Ubiquitin C-Terminal Hydrolase L1 were modeled using relevant chemical and physical properties; modeling results were validated by comparison with data obtained from healthy subjects or individuals affected by neurological diseases. Brain imaging data were used to model passage of biomarkers across the blood–brain barrier.

Results

Our results show the following: (1) changes in biomarker serum levels due to age or disease progression are accounted for by differences in kidney filtration; (2) a significant change in the brain-to-blood volumetric ratio, which is characteristic of infant and adult development, contributes to variation in blood concentration of biomarkers; (3) the effects of extracranial contribution at steady-state are predicted in our model to be less important than suspected, while the contribution of blood–brain barrier disruption is confirmed as a significant factor in controlling markers’ appearance in blood, where the biomarkers are typically detected; (4) the contribution of skin to the marker S100B blood levels depends on a direct correlation with pigmentation and not ethnicity; the contribution of extracranial sources for other markers requires further investigation.

Conclusions

We developed a multi-compartment, pharmacokinetic model that integrates the biophysical properties of a given brain molecule and predicts its time-dependent concentration in blood, for populations of varying physical and anatomical characteristics. This model emphasizes the importance of the blood–brain barrier as a gatekeeper for markers’ blood appearance and, ultimately, for rational clinical use of peripherally-detected brain protein.

Electronic supplementary material

The online version of this article (doi:10.1186/s12987-016-0045-y) contains supplementary material, which is available to authorized users.

Post-mortem magnetic resonance foetal imaging: a study of morphological correlation with conventional autopsy and histopathological findings

The aim of the present study is to offer our experience concerning post-mortem magnetic resonance (PMMR) in foetal death cases and an evaluation of the differences between the findings acquired by PMMR and by forensic autopsy. Fifteen foetuses were recruited from July 2014 to December 2015. These had suffered intrauterine death in women in the 21st to 38th week of gestation who were treated in the emergency department for non-perception of foetal movements. We performed a PMMR on foetuses, 3 ± 1 days on average from the time of death, and then a complete forensic autopsy was performed. All 15 foetuses were examined with a whole-body study protocol, starting from the skull, down to and including the lower limbs. The total time of examination ranged from 20 to 30 min in each case. The external evaluation and description of post-mortem phenomena (maceration), record of the weight and detection and the various measurements of foetal diameters were evaluated before performing autopsy. A complete histopathological study was performed in each case. Out of 15 cases examined, eight were negative for structural anatomical abnormalities and/or diseases, both in the preliminary radiological examination and the traditional autopsy. In the remaining seven cases, pathological findings were detected by PMMR with corresponding results at autopsy. PMMR can provide useful information on foetal medical conditions and result in improved diagnostic classification. It may enable the planning of a more suitable technique before proceeding to autopsy, including focusing on certain aspects of organ pathology otherwise not detectable. The association between PMMR, post-mortem examination and related histological study of the foetus-placenta unit could help reduce the percentage of cases in which the cause of foetal death remains unexplained. Lastly, it may allow a selective sampling of the organ in order to target histological investigations.

Quantification of maceration changes using post mortem MRI in fetuses

BACKGROUND

Post mortem imaging is playing an increasingly important role in perinatal autopsy, and correct interpretation of imaging changes is paramount. This is particularly important following intra-uterine fetal death, where there may be fetal maceration. The aim of this study was to investigate whether any changes seen on a whole body fetal post mortem magnetic resonance imaging (PMMR) correspond to maceration at conventional autopsy.

METHODS

We performed pre-autopsy PMMR in 75 fetuses using a 1.5 Tesla Siemens Avanto MR scanner (Erlangen, Germany). PMMR images were reported blinded to the clinical history and autopsy data using a numerical severity scale (0 = no maceration changes to 2 = severe maceration changes) for 6 different visceral organs (total 12). The degree of maceration at autopsy was categorized according to severity on a numerical scale (1 = no maceration to 4 = severe maceration). We also generated quantitative maps to measure the liver and lung T2.

RESULTS

The mean PMMR maceration score correlated well with the autopsy maceration score (R(2) = 0.93). A PMMR score of ≥4.5 had a sensitivity of 91%, specificity of 64%, for detecting moderate or severe maceration at autopsy. Liver and lung T2 were increased in fetuses with maceration scores of 3-4 in comparison to those with 1-2 (liver p = 0.03, lung p = 0.02).

CONCLUSIONS

There was a good correlation between PMMR maceration score and the extent of maceration seen at conventional autopsy. This score may be useful in interpretation of fetal PMMR.

Brain extraction in pediatric ADC maps, toward characterizing neuro-development in multi-platform and multi-institution clinical images

Apparent Diffusion Coefficient (ADC) maps can be used to characterize myelination and to detect abnormalities in the developing brain. However, given the normal variation in regional ADC with myelination, detection of abnormalities is difficult when based on visual assessment. Quantitative and automated analysis of pediatric ADC maps is thus desired but requires accurate brain extraction as the first step. Currently, most existing brain extraction methods are optimized for structural T1-weighted MR images of fully myelinated brains. Due to differences in age and image contrast, these approaches do not translate well to pediatric ADC maps. To address this problem, we present a multi-atlas brain extraction framework that has 1) specificity: designed and optimized specifically for pediatric ADC maps; 2) generality: applicable to multi-platform and multi-institution data, and to subjects at various neuro-developmental stages across the first 6 years of life; 3) accuracy: highly accurate compared to expert annotations; and 4) consistency: consistently accurate regardless of sources of data and ages of subjects. We show how we achieve these goals, via optimizing major components in a multi-atlas brain extraction framework, and via developing and evaluating new criteria for its atlas ranking component. Moreover, we demonstrate that these goals can be achieved with a fixed set of atlases and a fixed set of parameters, which opens doors for our optimized framework to be used in large-scale and multi-institution neuro-developmental and clinical studies. In a pilot study, we use this framework in a dataset containing scanner-generated ADC maps from 308 pediatric patients collected during the course of routine clinical care. Our framework leads to successful quantifications of the changes in whole-brain volumes and mean ADC values across the first 6 years of life.