Oct 25 – 27, 2022
Zadar, Croatia
Europe/Zagreb timezone

HIBALL Lecture in Brain Analytics and Learning

Ivica Kostović

Professor Ivica Kostović is the founder and Honorary Director of the Croatian Institute for Brain Research and a Fellow of the Croatian Academy of Sciences and Arts. His research interests consist in human developmental neurobiology and neuroanatomy, specifically the development of cortical circuitry, synaptogenesis, axonal pathways, dendrites, and the transient patterns of cortical organisation. His recent work looks at the correlation between cortical histogenesis and MR imaging in health and perinatal brain lesions.  Professor Kostović will deliver the first HIBALL Lecture in Brain Analytics and Learning. 

The Helmholtz International BigBrain Analytics and Learning Laboratory (HIBALL) establishes a long-standing international partnership around the BigBrain Project, and raises cross-disciplinary collaborations to the next level by reinforcing utilisation and co-development of the latest AI and high-performance computing (HPC) technologies for building highly detailed 3D brain models.
As an HBP partnering project, HIBALL is committed to HBP’s training activities and supports this with joint events and participation in the curriculum, through the work of our shared members, offering young researchers a great opportunity to showcase their research work and practise their mentoring skills.

From transient cellular compartments to brain's network architecture

The most challenging question in research of development of connectivity in the human brain is how neural networks emerge and mature during early fetal life. Current MR studies of developing structural and/or functional brain connectivity during early fetal months are limited by many constraints: in vivo studies are not performed before 20 weeks of gestation, most of projection pathways are still growing, neurons are migrating, most of pyramidal neurons giving rise to associative pathways are not yet born, cortical areas are not differentiated, convolutions are not formed, synaptic junctions are sparse and border between cortical grey and fetal white matter can’t be delineated.

The first aim of this presentation is to show advantages of developmental studies of early networks for integration of multimodal data, brain modeling, design of new computing platforms and new potential for AI research. The study of early network formation in the human brain shows the following advantages:
1. Step-by-step analysis is conceptually logical: begins from simple circuitry to construction of complex distributed and integrated networks.
2. The number of strategic points of interaction, that is chemical synapses, is small and accessible for quantification.
3. The biological programming by expression of hominini specific genes can be more efficiently identified,
4. Time parameters of neurogenetic events show excellent resolution (for example, main input system to cortex from thalamus takes 4 months before involvement in sensory processing,
5. Basic cortical organization is already disclosed from the early fetal life in form of two tangential connectivity networks which enclose modular (columnar) units composed of principal pyramidal neurons (which constitute 80 percent of all neuronal population). Analysis of intrinsic processing versus external influences on the formation of synaptic networks and their plasticity are favored by intrauterine conditions. 6. Alternations of circuitry leading to major mental disorders are thought to begin during early fetal life (First-hit hypothesis).

The second aim of this presentation is to show how spatiotemporal analysis of transient cellular compartments (subventricular and intermediate zone, subplate, cortical plate, marginal zone), visualized on histological sections processed for transcription factors of projection neurons, synaptic, fibrillar and ECM markers combined with post mortem and in vivo MRI 3D models, may explain how and when axons destined for neural networks form their trajectories being guided towards target areas (future nodes?) and select transient or permanent synaptic address with postsynaptic neurons. The early fetal cortex (8-11 postconceptional weeks - PCW) is composed of two tangential fibrillary-synaptic compartments which contain “modular” pathways from brain stem and basal forebrain and encompass densely cell-packed radially aligned cortical plate composed of deep projection neurons of layers 5, 6 and subplate which are born in the subventricular compartment. Next phase (12-14 PCW) is essential for the formation of long subcorticocortical (from thalamus and amygdala) and corticosubcortical (to striatum, pons and spinal cord pathways) and formation of primate characteristic, the most voluminous cerebral compartment – subplate. There is crucial geometrical difference between periventricular-limbic pathways and pathways for neocortex: limbic pathways form bundles with parallel arrangement of axons, while pathways approaching neocortex form sagittal strata, radiations and plexus-like termination within the subplate. Functionally, these early, sparsely synaptically connected circuitry shows spontaneous activity (oscillatory) and this phase may be described as pre-network phase in sense of connectoma. Between 15 and 22 PCW axons of somatosensory, auditory, visual and limbic pathways reach the subplate compartments in their target areas. These projection and limbic pathways, together with local randomly distributed glutamatergic and GABA-ergic neurons, form first structurally defined transient networks. Their functional connectivity remains unexplored. After 22 PCW projection axons make synapses within the cortical plate and first permanent “weak” networks appear, which respond to external stimuli (somatosensory, visual, auditory). In the interhemispheric cingular cortex, first resting state (rs-fc MRI) activity was recorded. Due to the fact that at that time long range associative axons did not penetrate the cortical plate, first resting state connectivity may depend on transient subplate and marginal zone nexus compartments. This phase can be described as co-existence between transient and initial permanent networks. Real long-range associative networks are being established after 28 PCW with involvement of axons of supragranular associative pyramidal neurons. Transient networks disappear during the first year with progressive strengthening of permanent pathways linking remote brain regions.

In conclusion, transient cellular compartments are essential constituency of earliest fetal connectivity and explain how neural components became incorporated into networks, outline relationships between early structural and functional connectivity and participate in the first long-range integration across the hemisphere and contribute to our understanding of brain’s network architecture . In addition, transient cellular compartments provide necessary continuity of neural functioning between transient and permanent neural networks. In the future research aggregated data obtained by analysis of dynamic development of transient cellular compartments, should be integrated with existing images from classical atlases and contemporary 3D models, new quantitative MR markers and spatiotemporal gene expression networks, using deep learning approaches. In addition, templates for each described phase should be constructed. Transient compartmentalization, as a foundation of model of brain development (“compartmentoma”), is essential for understanding of connectivity in utero and providing normative data for analysis of abnormal neurobiological development.

Keywords: neural connectivity of human fetal cortex, transient compartments, transient and permanent networks, 3D models, geometry of fiber systems, target finding, deep learning