Diffusion Tractography

Tuesday 13 May 2014

Daan Christiaens^1,2, Frederik Maes^1,2, Stefan Sunaert^1,3, and Paul Suetens^1,2
1Medical Imaging Research Center, KU Leuven, Leuven, Vlaams-Brabant, Belgium, ²Electrical Engineering, KU Leuven, Leuven, Vlaams-Brabant, Belgium,³Translational MRI, KU Leuven, Leuven, Vlaams-Brabant, Belgium

We extend the global fibre reconstruction framework of Reisert et al. (2011) for multi-shell HARDI data and allow to use any fibre response function represented in spherical harmonics. Additionally, we introduce two isotropic terms that model the fractions of grey matter and cerebrospinal fluid. Our approach successfully recovers the main white matter tracts on the high-resolution, multi-shell data of the human connectome project, using a fibre response function estimated from the data.

Matteo Figini^1,2, Giuseppe Baselli¹, Marco Riva³, Lorenzo Bello^4,5, Hui Zhang⁶, and Alberto Bizzi^7
1Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Milano, MI, Italy, ²Scientific Direction, Fondazione IRCCS Istituto Neurologico "Carlo Besta", Milano, MI, Italy, ³Neurosurgery, Università degli Studi di Milano, Milano, MI, Italy, ⁴Surgical Neuro-Oncology Unit, Humanitas Clinical and Research Center, Rozzano, MI, Italy, ⁵Department of Medical Biotechnology and Translational Medicine, Università degli Studi di Milano, Milano, MI, Italy, ⁶Department of Computer Science and Centre for Medical Image Computing, University College London, London, United Kingdom,⁷Neuroradiology, Humanitas Clinical and Research Center, Rozzano, MI, Italy

In this work we developed a procedure to perform NODDI-based tractography with an ODI threshold equivalent to the commonly used FA = 0.2 in order to fairly compare the results with those from DTI-based tractography; the aim is to investigate the advantages of performing tractography based on fiber directions and microstructural parameters estimated by NODDI. Our preliminary results showed a higher track density in areas infiltrated by brain tumours with NODDI than with DTI. The advantage of NODDI-based tractography appears to be most evident in areas of vasogenic edema.

Emmanuel Caruyer¹, Luke Bloy², Birkan Tunç¹, Jérémy Lecoeur¹, Varsha Shankar¹, and Ragini Verma^1
1Section of Biomedical Image Analysis, Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Children's Hospital of Pennsylvania, Philadelphia, PA, United States

We present a comparative study of 16 fiber tractography algorithms on a test-retest dataset of 9 healthy subjects. We evaluate the ability of the tractography methods to provide reproducible tracts geometries, as well as to reconstruct subject-specific white matter anatomy. The dataset consists in 9 healthy subjects, scanned at three time points separated by a 2 weeks interval. We reconstructed both left and right corticospinal tracts. Tractography algorithms were tested using either their freely available implementations or in house software. We report contrasted results across tracking methods.

Jean-Christophe Houde¹, Emmanuel Caruyer², Alessandro Daducci³, and Maxime Descoteaux^1
1Computer Science Departement, Université de Sherbrooke, Sherbrooke, Québec, Canada, ²Department of Radiology, University of Pennsylvania, Pennsylvania, United States, ³Signal Processing Lab (LTS5), École polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland

The Tractometer evaluation system is presented. We show how it was used to evaluate entries in the ISBI 2013 HARDI reconstruction challenge. From this evaluation, important questions about local modeling and tractography arose, such as: for the tractography end user, what is really important? More precise local reconstructions? Tracking algorithms with the least errors, or with the least missed bundles? Is angular error the best quality metric for local model development? How should we deal with errors in tractography? And most critically, how can we evaluate tractography pipelines on real data? These questions, and some insights, are presented in this work.

Sean Foxley¹, Saad Jbabdi¹, Wilfred Lam¹, Olaf Ansorge², Gwenaelle Douaud¹, and Karla Miller^1
1FMRIB Centre, University of Oxford, Oxford, OXON, United Kingdom, ²University of Oxford, Oxford, OXON, United Kingdom

Post-mortem human brain imaging at 3T using diffusion-weighted steady state free precession (DWSSFP) has been demonstrated to provide improved data for tractography compared with diffusion weighted spin echo data. Here, DWSSFP was used to acquire diffusion data of post-mortem human brain at 7T. Tractography results produced from this data demonstrated increased prevalence of secondary fiber estimates compared with equivalent data acquired at 3T. Better estimates from 7T data provided improved results in tracts that encounter intersecting pathways; specifically those that project through the centrum semiovale. These projecting tracts detected in 7T data are largely unseen in 3T data.

Luis Manuel Colon-Perez¹, Mansi Parekh², Michelle Couret³, Rosemary Klassen³, Michael King⁴, Paul Carney⁵, and Thomas Mareci^1
1Biochemistry and Molecular Biology, University of Florida, Gainesville, FL, United States, ²Radiology, Stanford University, Stanford, CA, United States,³University of Florida, FL, United States, ⁴Pharmacology and Therapeutics, University of Florida, FL, United States, ⁵Pediatrics, University of Florida, FL, United States

The perforant pathway is a bundle of axons that provide direct connections from the entorhinal cortex (EC) to all subfields of the hippocampus (HC). This pathway is a complicated structure difficult to visualize in clinical data. In order to study this pathway, we acquired high spatial resolution diffusion weighted images of excised human tissue, which reduce volume averaging enhancing the ability to resolve this small fiber pathway and employ an advanced model of diffusion displacement which improve tracking for smaller fiber bundles by resolving crossing and kissing fibers. In this study we visualize this pathway using probabilistic and deterministic tractography.

Yurui Gao^1,2, Andrew J Plassard³, Ann Choe^2,4, Iwona Stepniewskwa⁵, Xia Li⁴, Bennett A Landman^4,6, and Adam W Anderson^2,4
1VUIIS, Vanderbilt University, Nashville, Tennessee, United States, ²BME, Vanderbilt University, TN, United States, ³Computer Science, Vanderbilt University, TN, United States, ⁴VUIIS, Vanderbilt University, TN, United States, ⁵Psychology, Vanderbilt University, TN, United States, ⁶Electrical Engineering, Vanderbilt University, TN, United States

Diffusion MRI has been proposed as a method to parcellate cortex into functionally distinct regions based on tractography-derived connectivity features. However, the accuracy of the parcellation has not been investigated extensively. This study aimed to validate DTI connectivity-based cortical parcellation by comparison to a histological gold-standard. The comparison reveals the potential and limitations of connectivity-based parcellation.

Daniel Barazany^1,2, Silvia De Santis^1,2, Mara Cercignani³, Yaniv Assaf², and Derek K Jones^1
1School of Psychology, Cardiff University, Cardiff, United Kingdom, ²Department of Neurobiology, Tel Aviv University, Tel Aviv, Israel, ³Clinical Imaging Sciences Centre, Brighton and Sussex Medical School, Falmer, East Sussex, United Kingdom

Tractometry framework aims to establish more accurate and informative description of white matter brain tissue by on exploiting imaging methods which provide higher sensitivity to the underlying tissue subcomponents such as the composite hindered and restricted model of diffusion (CHARMED) and qMT. In this work we utilized tractometry microstructural metrics to examine and characterize two different white matter pathways of the living rat brain, the corpus callosum and the fimbria. We show the high sensitivity of these metrics to differentiate between the two fibers.

Robert Elton Smith¹, J-Donald Tournier^1,2, Fernando Calamante^1,3, and Alan Connelly^1,3
1The Florey Institute of Neuroscience and Mental Health, Heidelberg, Victoria, Australia, ²Centre for the Developing Brain, King's College London, London, London, United Kingdom, ³Medicine (Austin Health / Northern Health), The University of Melbourne, Heidelberg, Victoria, Australia

Accurate evaluation of brain white matter connectivity using diffusion MRI tractography requires development of robust quantitative methods. We propose a successor to the "Spherical-deconvolution Informed Filtering of Tractograms (SIFT)" method, entitled SIFT2, which determines an appropriate weighting factor for every streamline in a tractogram in a manner that matches the streamline densities to the underlying fibre volumes as estimated by spherical deconvolution. The result is a dense whole-brain streamlines reconstruction where the sum of streamline weights between two areas is a proportional estimate of the cross-sectional area of the axonal pathway connecting those regions; a truly quantitative measure of white matter connectivity.

Mark Drakesmith¹, Karen Caeyenberghs², Anirban Dutt³, Glyn Lewis⁴, Anthony S David³, and Derek K Jones^1
1CUBRIC, Cardiff University, Cardiff, Wales, United Kingdom, ²Department of Physical therapy and motor rehabilitation, Ghent University, Gent, Belgium,³Institute of Psychiatry, Kings College London, London, United Kingdom, ⁴Academic Unit of Psychiatry, University of Bristol, Bristol, United Kingdom

Graph theory (GT), a powerful tool for quantifying network properties from tractography, is subject to bias and instability due to false positives (FPs). This study illustrates this bias in GT metrics and examines the effects of thresholding to reduce this bias. Thresholding does reduce the effects of FPs but also introduce their own biases. Statistical comparisons of GT metrics are also shown to be highly unstable across thresholds compared to non-GT metrics, although genuine group differences tend to be more stable. A multi-threshold permutation correction strategy is suggested to improve sensitivity of statistical comparisons of GT metrics to genuine group differences.