OPtimization of Tractography algorithms through the intEgration of Anatomical knowledge of the brain white Matter
In the cutting-edge field of connectomics, the use of tractography derived from diffusion magnetic resonance imaging (dMRI) allows for the non-invasive mapping of white matter fibers in the brain. Despite its potential, the accuracy of this structural connectivity mapping is influenced by the current limitations of tractography algorithms and ongoing debates surrounding our understanding of white matter anatomy. To tackle these challenges head-on, we have established an International Research Program (IRP) operating as an associated international laboratory and called OpTeam (OPtimization of Tractography algorithms through the intEgration of Anatomical knowledge of the brain white Matter). OpTeam assembles a multidisciplinary team of experts in neuroanatomy and tractography with the goal of establishing an international hub dedicated to advancing the study of white matter anatomy and developing state-of-the-art tools for conducting advanced multimodal, multiscale, and multi-species tractography research in this domain.
Led by Dr. Laurent Petit from GIN-IMN (CNRS – University of Bordeaux, France) and Professor Maxime Descoteaux from SCIL (University of Sherbrooke, QC, Canada), who have a rich history of collaboration spanning ten years, OpTeam encompasses various projects designed to enhance our understanding of white matter anatomy and integrate it into tractography algorithms across different species, ranging from mice to humans. By doing so, the aim is to enhance the accuracy and reliability of human brain connectome reconstructions, benefiting both fundamental research and clinical applications. In addition to these scientific initiatives, OpTeam is committed to promoting collaboration and knowledge-sharing among students, postdoctoral researchers, and scholars from participating institutions. This collaborative framework not only encourages partnerships between French and Canadian academic institutions but also facilitates future funding opportunities and allows for joint student supervision across the universities involved.
OPTEAM has received the support of the CNRS with a funding of 50 000 € for 5 years up to 2028.
What are we interested in?
For almost 30 years, functional and diffusion MRI have opened a new era for a better understanding of the functional and structural connectivity of the human brain, respectively. Diffusion MRI tractography has thus been the first tool allowing non-invasive exploration of white matter in the human brain at an unprecedented level of detail (Jeurissen et al., 2019). It has shed new light on previously less accessible knowledge of cerebral anatomy. Beyond the various necessary improvements addressed in our project, tractography has led to a conceptual revolution by restoring a central role to the study of white matter organization. This revolution impacts both fundamental research and clinical and neurosurgical fields, and has ushered in the era of the “connectome” (Thiebaut de Schotten and Forkel, 2022). In addition to the millimetric resolution of mapping the white matter with dMRI (ranging from 0.1-1mm from rodents to humans), light sheet microscopy imaging (LSI) and optical coherence tomography (OCT) are emerging techniques for studying this structural WM organization at mesoscopic resolution (ranging from 0.2-10µm). The strengths and limitations of each of these techniques have partially overcome various technical challenges to address a number of open questions in neuroanatomy. However, no single method can be considered entirely reliable for a comprehensive study of brain connections.
In this context, the multiplicity of information provided by different techniques at different neural scales appears as the most reliable approach for studying the detailed organization of white matter anatomy. 
Exactly in this vein, our IRP proposal, OpTeam, builds an international pole for developing multimodal, multiscale, and multispecies tractography tools of brain white matter fibers. Ultimately, we aim to be able to provide a perfectly reliable representation of the white matter brain anatomy.
Why do we still need to study white matter anatomy in the human brain in 2024?
The idea that the organization and microstructure of brain white matter are crucial to understanding cognition and brain diseases emerged from the work of anatomists in the second half of the 19th century, e.g., Theodor Meynert (Meynert, 1885) and Jules, and Augusta Dejerine (Dejerine and Dejerine-Klumpke, 1895). Their meticulous dissections and studies of patients with brain lesions still shape our current understanding of how connections and disconnections influence brain functions. The advent of dMRI in the late 20th century has paved the way for non-invasive study of white matter anatomy and brain connections in healthy individuals and patients. The first images of white matter fiber reconstruction date back 25 years ago in an abstract of the 1st Conference on Human Brain Mapping held in Paris. Four years later, the first images of white matter bundles (Conturo et al., 1999) opened the floodgates to an incredible avalanche of publications (about 65,200 entries on Google Scholar for “diffusion MRI tractography” as of April 2023). This frenetic period of dMRI and tractography studies has allowed for the in vivo study of the organization of major white matter fiber bundles in the human brain with an unprecedented degree of precision since their first descriptions (e.g., (De Benedictis et al., 2016; Hau et al., 2017; Vavassori et al., 2021; Petit et al., 2023). However, the neuroscience community has explicitly recognized the limitations of current tractography algorithms and has simultaneously highlighted the limitations of our knowledge of white matter neuroanatomy. This has tempered the enthusiasm of those considering dMRI and tractography as the ultimate solution for reliably representing white matter neuroanatomy. There are thus two categories of people: the optimists who use fiber tractography without questioning, and the skeptics who consider that many results obtained from tractography should be interpreted with caution. It is indeed important to be aware of the pitfalls and potential limitations of fiber tracking, which can lead to misinterpretation of the resulting tractograms.
Prof. Descoteaux took the lead in organizing the “ISMRM 2015 Tractography Challenge,” with contributions from Dr. Petit (Maier-Hein et al., 2017). The main conclusion of this work, which continues to have a significant impact on the community (with over 950 citations as documented by Google Scholar in mid-April 2023), is that tractography algorithms should not be relied upon to accurately reconstruct white matter anatomy without prior knowledge of that anatomy. It emphasizes the importance of having a comprehensive understanding of white matter anatomy for reliable tractography results. More generally, the tractography community is therefore well aware of the limitations, common misconceptions and hidden biases of this technique, but also of all the potential it offers once these flaws will be better understood and corrected (Rheault et al., 2020).
If it is essential to integrate anatomical knowledge into tractography, it is necessary to actually know the ground truth. However, our understanding of anatomical details, especially in the human brain, is far from comprehensive. The most contentious debates revolve around the detailed description of white matter fibers that connect cortical structures. There is no general consensus on how to describe the characteristics of different types of white matter tracts, their crossing patterns, or their precise termination points (Mandonnet et al., 2018). This is why the current era of studying white matter connectivity is fascinating. Diffusion imaging and tractography offer a unique opportunity to study the structure and organization of this white matter in vivo. However, the inherent limitations of these techniques and the mandatory validation of their anatomical representations have opened a Pandora’s box in terms of our knowledge of human brain white matter anatomy.
From a metaphorical point of view, current tractography algorithms behave like Google Maps™ algorithm working without prior knowledge of road maps, but having only some geographic information (rivers, mountains, valleys…). To find the best route between two cities, a poorly informed Google Maps™ algorithm would propose numerous different solutions, including false positives and true roads corresponding to the ground truth of road maps. It would have to “rediscover” the logic of the organization of roads that connect different cities, such as the fact that roads do not pass anywhere but depend on rules derived from geographic constraints, for example, roads follow rivers, they generally pass through the valleys and not through the middle of mountains, etc. Once a road connecting the two cities is selected, it needs to be validated by the ground truth, which is the road map. The same applies to the anatomy of white matter and tractography algorithms. To further extend the metaphor, the current problem is not only that current tractography algorithms need to reconstruct the anatomy of white matter without prior knowledge, but also that our knowledge of white matter anatomy is far from being as precise as road maps!
Both Drs Petit and Descoteaux have been pioneers in the development of approaches aimed at better understanding WM anatomy and utilizing this knowledge for improved tractography. This IRP proposal, OpTeam, builds on this dual expertise in neuroanatomy and tractography to strengthen our international collaborations. These research efforts will improve the anatomical knowledge of WM at the mesoscopic and macroscopic scales and integrate them into tractography algorithms to construct better connectomes of the human brain.
How can these problems be solved and reliable connectomes be constructed?
Our scientific project focuses on improving the anatomical knowledge of white matter and integrating it into tractography algorithms. This is a necessary step to better integrate dMRI and tractography data into large multimodal population databases (such as the Human Connectome Project, UK Biobank, i-Share) as well as clinical and neurosurgical practices.
We propose to tackle this challenge at both the mesoscopic and macroscopic scales. This will be achieved through the continuation of ongoing collaborative projects, namely CROSS-TRACTS, X-TRACT and ALL-in-ONE. 
As such, the present IRP OpTeam project is fully based on the current research themes of its constituent laboratories.
The CROSS-TRACTS project aims to reveal the topology of the WM crossing fibers in the mouse brain at an unprecedented accuracy, thanks to advanced multimodal tractography approaches. It has received the support of the French ANR (ANR-22-CE45-0004) with a funding of 630 000 € for 48 months from October 2022 up to September 2026. CROSS-TRACTS first apply a gold-standard methodological approach for unraveling the WM’s anatomical features by combining viral tract labeling, whole-brain clearing, and light-sheet microscopy imaging (LSI). Second, in addition to advanced dMRI tractography, we use resting state fMRI (rs-fMRI) to investigate WM fibers’ fanning, bending, or 3-way asymmetric crossings at the macroscopic scale. CROSS-TRACTS will thus produce tractograms across different modalities (dMRI, rs-fMRI, LSI) and at multiple scales (macroscopic, mesoscopic) (Figure 2). We will tackle the compound challenge of integrating tractograms across modalities by applying a new deep neural network-based methodology that combines the dMRI-, rs-fMRI- and LSI-based tractograms within a correspondence autoencoding architecture (Legarreta et al., 2021; Legarreta et al., 2023).
CROSS-TRACTS will thus provide the neuroscience community with groundbreaking anatomical knowledge of the WM crossings’ topology. It will provide the “neural-tracing/LSI” community with an unprecedented LSI-based tractography tool to track their cleared fluorescent samples and, will give the fMRI community a dedicated fMRI-based asymmetrical tractography algorithm and, it will provide the dMRI community with a new generation of tractography algorithms that actually considers the ground truth of WM anatomy through a deep neural network autoencoder framework. To do so, CROSS-TRACTS allies scientific partners with very complementary expertise to resolve different technical issues including Drs. Petit, Joliot and Miraux are part of the current OpTeam IRP project.
CROSS-TRACTS has a strong potential for back-translation to human studies. Our preclinical approach is crucial in developing tools for integrating prior knowledge in dMRI tractography, and this can be extensively applied in human clinical studies and large multimodal neuroimaging databases.
In this respect, the X-TRACT project that allies scientific partners of the current OpTeam IRP project, namely Prof. Lefebvre, Prof. Descoteaux and Dr. Petit, is highly complementary to CROSS-TRACTS. X-TRACT also aims to study the organization of crossings between WM fibers, but this time on samples of human brain. It uses both MRI-based tractography and optical coherence tomography (OCT), and more specifically, the recently developed “serial blockface” histological approach by the project PI, Prof. Lefebvre (Lefebvre et al., 2019). This approach combines a tissue slicing device with an optical microscope. The brain is sequentially sliced to reveal tissue layers that are imaged under the microscope. The process is repeated until the entire sample has been imaged. Then, using advanced registration methods, all images are assembled into a single 3D volume (Lefebvre et al., 2017). Unlike the CROSS-TRACTS project, tissue labeling with a fluorescent dye linked to a viral injection is not required. Thus, OCT imaging and analysis methods can be easily adapted to experiments on postmortem human samples.
X-TRACT was initially supported by a Canadian QBIN Pilot Project grant (#35450) with a funding of 17 500 $CAD for 42 months from October 2021 to March 2023, and is currently also supported by a Canadian FRQNT Team Research Project grant (2021-PR-282231) with a funding of 240 500 $CAD for 48 months from June 2020 to May 2024. It will achieve its objectives by: 1) Improving a serial blockface OCT imager to integrate polarization sensitive detection (psOCT). This will improve the ability of the system to capture WM fibers parallel to the optical axis when performing microscopic imaging. 2) Validating the acquisition procedures and optimizing the multimodal imaging and analysis pipeline to align OCT to dMRI using mouse brains and 3) Investigating a fiber crossing area in human brain samples where multiple fiber bundles are crossing by performing ex vivo dMRI imaging and serial blockface histology. The rationale for using OCT is that it possesses intrinsic optical contrast in brain tissues, mainly due to myelinated fibers. Tissue labeling with fluorescent dye is not required, thus OCT imaging and analysis methods can be more easily translated from animal to human sample experiments. The expected outcome is an unprecedented mapping between the dMRI signal and the underlying myelin macro/meso-structure in the human brain obtained from OCT.
A third scientific project is currently emerging from the highly complementary expertise of the OpTeam partners. Dubbed ALL-in-ONE, it aims at describing the impact of WM microstructural abnormalities on the anatomo-functional connectivity of the healthy human brain. It allies Drs. Petit and Joliot, Prof. Rheault and Prof. Descoteaux engaged in the current OpTeam IRP project.
Different approaches based on deep learning allow today the automatic detection of WM abnormalities (perivascular spaces, white matter lesions) from multimodal magnetic resonance imaging (MRI T1, T2, FLAIR, …). In parallel, diffusion MRI and tractography allow the reconstruction of the fiber bundles of this white matter while resting state functional MRI reveals the networks of brain regions working in a synchronous way. All these data constitute the structural connectome of the human brain. Currently, the reconstruction of this connectome does not take into account the presence of WM abnormalities. ALL-in-ONE is combining these different connectomic fields within the same study so that the microstructural WM defects will be taken into account during the complete reconstruction of the structural connectome of the human brain. The objective is to apply deep learning tools for automatic detection of WM abnormalities on the UK-BioBank database (N=20,000) (Miller et al., 2016), and then to evaluate the effects of the presence of these abnormalities on the anatomo-functional connectivity obtained from diffusion MRI and resting-state functional MRI data of these same 20,000 subjects. ALL-in-ONE is based on the pooling of analysis tools already developed by the above-mentioned partners. It comprises the application of deep learning tools developed by Dr. Joliot to automatically segment PeriVascular Space (PVS) and White Matter Hyperintensities (WMH) segmentation in human MR images (Boutinaud et al., 2021; Duperron et al., 2023) and the dMRI and tractography pipelines developed by Prof. Rheault, Prof. Descoteaux and Dr. Petit (TractoFlow,(Theaud et al., 2020); ExTractorFlow (Petit et al., 2023); Tractometry (Cousineau et al., 2017)).
Who are we ?
Dr. Laurent Petit, Director of Research at the CNRS (Section 28) masters brain anatomy and develops innovative strategies for the tractography of brain white matter pathways. Dr. Petit has been cited more than 11770+ times, his h/i-10index are 51/75 and has 90+ journal publications (LP google scholar) including over 30+ dedicated to the diffusion MRI tractography since the last five years with 50% including ex vivo microscopic dissection data for validation of diffusion tractography in the human brain.
Prof. Maxime Descoteaux is a world expert in developing diffusion MRI and tractography processing tools. In 2021, he became a member of the college of the Royal Society of Canada. He is the founder and director of the Sherbrooke Connectivity Imaging Laboratory (SCIL), where there are 10-15 graduate and medical students, and postdocs. He holds the Research Chair in NeuroInformatics and his research focuses on brain connectivity from state-of-the-art diffusion MRI and multimodality acquisition, reconstruction, tractography, processing and visualization. The aim of the SCIL is to better understand structural connectivity, develop novel tractography algorithms, validate them and use them for human brain mapping and connectomics applications. Prof. Descoteaux has been cited more than 13600+ times, his h/i-10index are 57/168 and has 155+ journal publications (MD google scholar).
Dr. Marc Joliot, Director of Research at the CEA is an expert in intrinsic functional connectivity and human brain atlas. He is the academic leader of the LABCOM Ginesislab which aims is to conduct research in the field of management and analysis of large biomedical image databases of several hundred to several thousand individuals. He is currently leading the work-package RHU SHIVA (Stopping the cognitive decline and dementia by fighting covert cerebral small vessel disease) in charge of developing cutting-edge methods- based on deep-learning technologies, for automated quantification and differentiation of multiple cerebral coverts, white matter located, biomarkers on MRI.
Dr. Sylvain Miraux (CNRS Research Director, UMR5536-CNRS/Univ. Bordeaux), leading the CRMSB, has published more than 60 peer-reviewed articles in MRI methodology and applications and has experience in MR sequence development at a high magnetic field on preclinical small animal models. He has experience in project management in Technology for Health (ANR TecSAN, Ligue Contre le Cancer) and developed collaborations with industrial partners (Bruker, Siemens).
Dr. Nadège Corbin (CNRS Research Engineer, UMR5536-CNRS/Univ. Bordeaux)
Prof. François Rheault is a young Professor in Computer Science at the University of Sherbrooke. As the Director of the MINi (Medical Imaging and Neuroinformatics) Lab, he primarily focuses on research in diffusion MRI and tractography. He actively contributes to open-source software projects such as Dipy, Nibabel, COMMIT, and Scilpy, with a particular interest in white matter bundle segmentation, reproducibility evaluations, longitudinal analysis, and connectomics. He emphasizes the importance of engagement in open-source development and multidisciplinary collaborations.
Prof. Joël Lefebvre is a young Professor of digital imaging and computer vision in the computer science department of Université du Québec à Montréal (UQÀM), associate researcher at the Montréal Heart Institute and director of the Digital Imaging, Neurophotonics and Microscopy Laboratory (LINUM). He has an interdisciplinary background in physical and biomedical engineering, with a specialization in image processing and computer vision applied to neurophotonics. He received a master’s degree in applied sciences (biomedical engineering) and a doctorate in biomedical engineering from École Polytechnique de Montréal under the supervision of F. Lesage. He was a postdoctoral researcher at the Big Data Institute of the University of Oxford under the supervision of J. Rittscher. His main research activities focus on serial histology by optical coherence tomography of whole mouse brains, optical imaging of myelin, and deep learning applied to the analysis of 3D microscopy data.
References
Boutinaud P, Tsuchida A, Laurent A, Adonias F, Hanifehlou Z, Nozais V, Verrecchia V, Lampe L, Zhang J, Zhu Y-C, Tzourio C, Mazoyer B, Joliot M (2021) 3D segmentation of perivascular spaces on T1-weighted 3 Tesla MR images with a convolutional autoencoder and a U-shaped neural network. Frontiers in Neuroinformatics 15.
Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, McKinstry RC, Burton H, Raichle ME (1999) Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci U S A 96:10422-10427.
Cousineau M, Jodoin PM, Morency FC, Rozanski V, Grand’Maison M, Bedell BJ, Descoteaux M (2017) A test-retest study on Parkinson’s PPMI dataset yields statistically significant white matter fascicles. Neuroimage Clin 16:222-233.
De Benedictis A, Petit L, Descoteaux M, Marras CE, Barbareschi M, Corsini F, Dallabona M, Chioffi F, Sarubbo S (2016) New insights in the homotopic and heterotopic connectivity of the frontal portion of the human corpus callosum revealed by microdissection and diffusion tractography. Hum Brain Mapp 37:4718-4735.
Dejerine J, Dejerine-Klumpke A (1895) Anatomie des centres nerveux. Tome 1. Paris: Rueff et Cie.
Duperron MG et al. (2023) Genomics of perivascular space burden unravels early mechanisms of cerebral small vessel disease. Nat Med 29:950-962.
Hau J, Sarubbo S, Houde JC, Corsini F, Girard G, Deledalle C, Crivello F, Zago L, Mellet E, Jobard G, Joliot M, Mazoyer B, Tzourio-Mazoyer N, Descoteaux M, Petit L (2017) Revisiting the human uncinate fasciculus, its subcomponents and asymmetries with stem-based tractography and microdissection validation. Brain Struct Funct 222:1645-1662.
Jeurissen B, Descoteaux M, Mori S, Leemans A (2019) Diffusion MRI fiber tractography of the brain. NMR Biomed 32:e3785.
Lefebvre J, Delafontaine-Martel P, Lesage F (2019) A Review of Intrinsic Optical Imaging Serial Blockface Histology (ICI-SBH) for Whole Rodent Brain Imaging. Photonics 6:66.
Lefebvre J, Castonguay A, Pouliot P, Descoteaux M, Lesage F (2017) Whole mouse brain imaging using optical coherence tomography: reconstruction, normalization, segmentation, and comparison with diffusion MRI. Neurophotonics 4:1-14, 14.
Legarreta JH, Petit L, Jodoin PM, Descoteaux M (2023) Generative Sampling in Bundle Tractography using Autoencoders (GESTA). Med Image Anal 85:102761.
Legarreta JH, Petit L, Rheault F, Theaud G, Lemaire C, Descoteaux M, Jodoin PM (2021) Filtering in tractography using autoencoders (FINTA). Med Image Anal 72:102126.
Maier-Hein KH et al. (2017) The challenge of mapping the human connectome based on diffusion tractography. Nat Commun 8:1349.
Mandonnet E, Sarubbo S, Petit L (2018) The Nomenclature of Human White Matter Association Pathways: Proposal for a Systematic Taxonomic Anatomical Classification. Front Neuroanat 12:94.
Meynert T (1885) Psychiatry: Clinical Treatise on the Diseases of the Fore-Brain, trans. B. Sachs. New York & London: GP Putnam.
Miller KL et al. (2016) Multimodal population brain imaging in the UK Biobank prospective epidemiological study. Nat Neurosci 19:1523-1536.
Petit L, Ali KM, Rheault F, Bore A, Cremona S, Corsini F, De Benedictis A, Descoteaux M, Sarubbo S (2023) The structural connectivity of the human angular gyrus as revealed by microdissection and diffusion tractography. Brain Struct Funct 228:103-120.
Rheault F, Poulin P, Valcourt Caron A, St-Onge E, Descoteaux M (2020) Common misconceptions, hidden biases and modern challenges of dMRI tractography. J Neural Eng 17:011001.
Theaud G, Houde JC, Bore A, Rheault F, Morency F, Descoteaux M (2020) TractoFlow: A robust, efficient and reproducible diffusion MRI pipeline leveraging Nextflow & Singularity. Neuroimage 218:116889.
Thiebaut de Schotten M, Forkel SJ (2022) The emergent properties of the connected brain. Science 378:505-510.
Vavassori L, Sarubbo S, Petit L (2021) Hodology of the superior longitudinal system of the human brain: a historical perspective, the current controversies, and a proposal. Brain Struct Funct 226:1363-1384.