MEDUSA: A GPU-based tool to create realistic phantoms of the brain microstructure using tiny spheres

Ginsburger, Kévin; Matuschke, Felix; Poupon, Fabrice; Mangin, Jean‐François; Axer, Markus; Poupon, Cyril

doi:10.1016/j.neuroimage.2019.02.055

Cited by 39 publications

(54 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Existing frameworks have been developed to model morphological features such as fiber undulation43,51 (although we do not observe periodic undulations in our data), and diameter variations52,53 . Others allow for the generation of a more complex WM environment with beaded axons and cells54 . None, however, have so far imposed anatomically realistic trajectory patterns or longitudinal ADDs.…”

mentioning

confidence: 99%

Axon morphology is modulated by the local environment and impacts the non-invasive investigation of its structure-function relationship

Andersson

Kjer

Rafael-Patiño

et al. 2020

Preprint

View full text Add to dashboard Cite

Axonal conduction velocity, which ensures efficient function of the brain network, is related to axon diameter. Non-invasive, in vivo axon diameter estimates can be made with diffusion magnetic resonance imaging, but the technique requires 3D validation. Here, high resolution, 3D synchrotron X-ray Nano-Holotomography images of white matter samples from the corpus callosum of a monkey brain reveal that blood vessels, cells and vacuoles affect axonal diameter and trajectory. Within single axons, we find that the variance in diameter and conduction velocity correlates with the mean diameter, contesting the value of precise diameter determination in larger axons. These complex 3D axon morphologies drive previously reported 2D trends in axon diameter and g-ratio. Furthermore, we find that these morphologies bias the estimates of axon diameter with diffusion magnetic resonance imaging and, ultimately, impact the investigation and formulation of the axon structure-function relationship. 4 acquired diffusion signal 15 . However, diffusion MRI-based AD estimates 14,16,17 are larger than those obtained by histology 15,18 . A potential cause is inaccurate modeling of the WM compartments, including the century old representation of myelinated axons as cylinders. A validation of the 3D WM anatomy could thus improve diffusion MRI-based AD estimations 17 and shed light on the validity of enforcing a cylindrical geometry and constant g-ratio in axonal structure-function relations. Recent 3D electron microscopy (EM) studies on axon morphology of the mouse reveal, in high resolution, non-uniform ADs and trajectories 19,20 . However, axons are only tracked for up to 20 µm, a fraction of their length in MRI voxels. Here, we characterize the long-range micro-morphologies of axons against the backdrop of the complex 3D WM environment consisting of blood vessels, cells and vacuoles. With synchrotron X-ray Nano-Holotomography (XNH), we acquire MRI measurements of the WM from the same monkey brain as in Alexander et al. (2010) 14 and Dyrby et al. (2013) 21 , in which the MRIderived AD estimates were larger than those estimated by histology. The 3D WM environment is mapped at a voxel size of 75 nm and volume of approximately 150 ⨉ 150 ⨉ 150 µm 3 . By combining adjacent XNH volumes, we extract axons >660 µm in length and show that AD, axon trajectory and g-ratio depend on the local microstructural environment. The 3D measurements shed light on the interpretation of 2D measurements, highlighting the importance of the third dimension for a robust description of single-axon structure and function. Lastly, by performing Monte Carlo (MC) diffusion simulations on axonal substrates with morphological features deriving from the XNH-segmented axons, we show that geometrical deviations from cylinders cause an overestimation of AD with diffusion MRI.

show abstract

mentioning

confidence: 99%

Axon morphology is modulated by the local environment and impacts the non-invasive investigation of its structure-function relationship

Andersson

Kjer

Rafael-Patiño

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…When combining all of the proposed mechanisms together, the achievable density is higher than any of the improvements individually. This improved performance is comparable to the state of the art, MEDUSA (Ginsburger et al, 2019), with particularly good performance relative to MEDUSA in the crossing fibre configurations. This improvement in density can be appreciated visually in Figure 11 which demonstrates virtual histology of a parallel fibre phantom for each of the mechanisms.…”

Section: Impact Of Biological Mechanismsmentioning

confidence: 62%

“…The most common approach to this is the packing of fibres into densely packed configurations (Close et al, 2009;Ginsburger et al, 2019Ginsburger et al, , 2018Rafael-Patino et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

“…The typical approach, as taken in the state-of-the-art MEDUSA algorithm (Ginsburger et al, 2019), is to generate a set of overlapping fibres decomposed into small segments and iteratively refine their positions to remove the overlap between them. Despite their recent progress, further advance of this class of techniques may be limited, because nature does not create fibres before attempting to pack them together.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Contextual Fibre Growth to Generate Realistic Axonal Packing for Diffusion MRI Simulation

Callaghan

Alexander

Zhang

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

This paper presents Contextual Fibre Growth (ConFiG), an approach to generate white matter numerical phantoms by mimicking natural fibre genesis. ConFiG grows fibres one-byone, following simple rules motivated by real axonal guidance mechanisms. These simple rules enable ConFiG to generate phantoms with tuneable microstructural features by growing fibres while attempting to meet morphological targets such as user-specified density and orientation distribution. We compare ConFiG to the state-of-the-art approach based on packing fibres together by generating phantoms in a range of fibre configurations including crossing fibre bundles and orientation dispersion. Results demonstrate that ConFiG produces phantoms with up to 20% higher densities than the state-of-the-art, particularly in complex configurations with crossing fibres. We additionally show that the microstructural morphology of ConFiG phantoms is comparable to real tissue, producing diameter and orientation distributions close to electron microscopy estimates from real tissue as well as capturing complex fibre cross sections. Signals simulated from ConFiG phantoms match real diffusion MRI data well, showing that ConFiG phantoms can be used to generate realistic diffusion MRI data. This demonstrates the feasibility of ConFiG to generate realistic synthetic diffusion MRI data for developing and validating microstructure modelling approaches.

show abstract

“…The segmentation of brain tissue ultrastructures in 3D-EM datasets can substitute simplistic biophysical models by more realistic models. Such realistic 3D tissue models open the possibility to investigate underlying reasons for the diffusion magnetic resonance imaging contrast and its macroscopic changes in brain diseases 42–45 or investigate conduction velocity in myelinated and unmyelinated axons in electrophysiology 46,47 .…”

Section: Discussionmentioning

confidence: 99%

DeepACSON: Automated Segmentation of White Matter in 3D Electron Microscopy

Abdollahzadeh

Belevich

Jokitalo

et al. 2019

Preprint

View full text Add to dashboard Cite

Advances in 3-dimensional electron microscopy (3D-EM) have made possible imaging brain ultrastructures with nanoscopic resolutions from hundreds of micrometers tissue. To quantify the acquired big datasets, it requires automated segmentation techniques that annotate the entirety of ultrastructural components. Recently, deep convolutional neural networks (DCNNs) have emerged as the key to automated human-level segmentation of neurites in 3D-EM datasets. Current techniques, however, address the segmentation of high-resolution datasets and for small numbers of neurites. Therefore, we present DeepACSON.DeepACSON performs DCNN-based semantic segmentation and 3D shape analysis based instance segmentation. We used DeepACSON to segment ten low-resolution serial block-face scanning electron microscopy datasets of the white matter of rats after sham-operation and traumatic brain injury. DeepACSON segmented hundreds of thousands of long-span myelinated axons, thousands of cell nuclei, and millions of mitochondria with excellent evaluation scores. DeepACSON quantified the morphological aspects of myelinated axons, such as axonal diameter, eccentricity, and tortuosity, as well as the spatial organization of ultrastructures, such as inter-mitochondrial distance, and the density of cells and myelinated axons.15 × 15 × 15 µm 3 white matter ultrastructures requires 2 400 h of an expert's time 4 . Semi-automated segmentation methods based on machine learning approaches 3, 5 have improved the rate of segmentation. However, these methods still require a considerable amount of manual interaction as the segmentation is driven on the manually extracted skeletons of neurites, proofreading, and correction of errors.1 Deep convolutional neural networks (DCNNs) have emerged as the key to the state-of-the-art automated segmentation techniques 6-9 . DCNN-based segmentation methods generally comprise two steps: generating a membrane probability map from 3D-EM datasets referred to as semantic segmentation, followed by instance segmentation that turns probability maps into individual object instances. DCNNs are typically used for semantic segmentation and more traditional image analysis techniques favoring bottom-up design, i.e., over-segmentation and subsequent merge, are used for instance segmentation. For example, DeepEM3D 7 applied 3D watershed transform on the membrane probability maps. Funke et al. 8 proposed learning voxel affinities, followed by 3D watershed and iterative region agglomeration on the watershed segmentation. Only recently, Januszewski et al. 9 suggested flood-filling networks merging the semantic and instance segmentation into a recurrent neural network. Flood-filling networks have a bottom-up design, offering a sophisticated training-based alternative for over-segmentation and subsequent merge.Bottom-up design of the state-of-the-art EM segmentation techniques 6-9 is subjected to local optimization. Hence, segmentation requires distinctive image features such as high contrast between cellular membranes and other struc...

show abstract

MEDUSA: A GPU-based tool to create realistic phantoms of the brain microstructure using tiny spheres

Cited by 39 publications

References 59 publications

Axon morphology is modulated by the local environment and impacts the non-invasive investigation of its structure-function relationship

Axon morphology is modulated by the local environment and impacts the non-invasive investigation of its structure-function relationship

Contextual Fibre Growth to Generate Realistic Axonal Packing for Diffusion MRI Simulation

DeepACSON: Automated Segmentation of White Matter in 3D Electron Microscopy

Contact Info

Product

Resources

About