A Bayesian model of shape and appearance for subcortical brain segmentation
UMass Chan AffiliationsIntellectual and Developmental Disabilities Research Center
Department of Psychiatry
Image Processing, Computer-Assisted
Magnetic Resonance Imaging
Reproducibility of Results
Neuroscience and Neurobiology
Psychiatry and Psychology
MetadataShow full item record
AbstractAutomatic segmentation of subcortical structures in human brain MR images is an important but difficult task due to poor and variable intensity contrast. Clear, well-defined intensity features are absent in many places along typical structure boundaries and so extra information is required to achieve successful segmentation. A method is proposed here that uses manually labelled image data to provide anatomical training information. It utilises the principles of the Active Shape and Appearance Models but places them within a Bayesian framework, allowing probabilistic relationships between shape and intensity to be fully exploited. The model is trained for 15 different subcortical structures using 336 manually-labelled T1-weighted MR images. Using the Bayesian approach, conditional probabilities can be calculated easily and efficiently, avoiding technical problems of ill-conditioned covariance matrices, even with weak priors, and eliminating the need for fitting extra empirical scaling parameters, as is required in standard Active Appearance Models. Furthermore, differences in boundary vertex locations provide a direct, purely local measure of geometric change in structure between groups that, unlike voxel-based morphometry, is not dependent on tissue classification methods or arbitrary smoothing. In this paper the fully-automated segmentation method is presented and assessed both quantitatively, using Leave-One-Out testing on the 336 training images, and qualitatively, using an independent clinical dataset involving Alzheimer's disease. Median Dice overlaps between 0.7 and 0.9 are obtained with this method, which is comparable or better than other automated methods. An implementation of this method, called FIRST, is currently distributed with the freely-available FSL package.
SourcePatenaude B, Smith SM, Kennedy DN, Jenkinson M. A Bayesian model of shape and appearance for subcortical brain segmentation. Neuroimage. 2011 Jun 1;56(3):907-22. doi: 10.1016/j.neuroimage.2011.02.046. Epub 2011 Feb 23. PubMed PMID: 21352927; PubMed Central PMCID: PMC3417233. Link to article on publisher's site
Permanent Link to this Itemhttp://hdl.handle.net/20.500.14038/34809
Related ResourcesLink to Article in PubMed
Showing items related by title, author, creator and subject.
Flipping Biological Switches: Solving for Optimal Control: A DissertationChang, Joshua TsuKang (2015-03-30)Switches play an important regulatory role at all levels of biology, from molecular switches triggering signaling cascades to cellular switches regulating cell maturation and apoptosis. Medical therapies are often designed to toggle a system from one state to another, achieving a specified health outcome. For instance, small doses of subpathologic viruses activate the immune system’s production of antibodies. Electrical stimulation revert cardiac arrhythmias back to normal sinus rhythm. In all of these examples, a major challenge is finding the optimal stimulus waveform necessary to cause the switch to flip. This thesis develops, validates, and applies a novel model-independent stochastic algorithm, the Extrema Distortion Algorithm (EDA), towards finding the optimal stimulus. We validate the EDA’s performance for the Hodgkin-Huxley model (an empirically validated ionic model of neuronal excitability), the FitzHugh-Nagumo model (an abstract model applied to a wide range of biological systems that that exhibit an oscillatory state and a quiescent state), and the genetic toggle switch (a model of bistable gene expression). We show that the EDA is able to not only find the optimal solution, but also in some cases excel beyond the traditional analytic approaches. Finally, we have computed novel optimal stimulus waveforms for aborting epileptic seizures using the EDA in cellular and network models of epilepsy. This work represents a first step in developing a new class of adaptive algorithms and devices that flip biological switches, revealing basic mechanistic insights and therapeutic applications for a broad range of disorders.
Development of transplant immunosuppressive agents - considerations in the use of animal modelsCostello, Russell; Kissenpfennig, Adrien; Martins, Paulo N.A.; McDaid, James (2018-11-01)INTRODUCTION: The development of all immunosuppressant agents to date has involved the experimental use of large and small animal models. Over the last half-century, immunosuppressive drugs have extended the lives of transplant patients worldwide. However, the use of animal models in the development of these drugs is not perfect, and this has brought to light a number of issues including idiosyncratic reactions that are found in animal models but not in humans. The 2006 highly publicized case of the 'elephant man' TGN 1412 drug trial highlights the importance of being cogent of the limitations of animal models. Areas covered: This review covers the utility and limitations of the use of animal models for the development of immunosuppressant agents. This includes both large and small animal models, particularly rodent models in the transplant setting. Expert opinion: The use of animal models represents a critical stage in the development of immunosuppressive drugs. Limitations include physiological differences to humans; this is especially true of immunologically naïve lab rodents with small memory cell populations. Toxic drug levels may differ widely between species. Animal models are also costly and raise ethical concerns. However, there is currently no way to recreate the complex environment of the human immune system purely in vitro.
Modeling human T1D-associated autoimmune processesKhosravi-Maharlooei, Mohsen; Madley, Rachel; Borsotti, Chiara; Ferreira, Leonardo M. R.; Sharp, Robert C.; Brehm, Michael A.; Greiner, Dale L.; Parent, Audrey V.; Anderson, Mark S.; Sykes, Megan; et al. (2021-12-10)BACKGROUND: Type 1 diabetes (T1D) is an autoimmune disease characterized by impaired immune tolerance to beta-cell antigens and progressive destruction of insulin-producing beta-cells. Animal models have provided valuable insights for understanding the etiology and pathogenesis of this disease, but they fall short of reflecting the extensive heterogeneity of the disease in humans, which is contributed by various combinations of risk gene alleles and unique environmental factors. Collectively, these factors have been used to define subgroups of patients, termed endotypes, with distinct predominating disease characteristics. SCOPE OF REVIEW: Here, we review the gaps filled by these models in understanding the intricate involvement and regulation of the immune system in human T1D pathogenesis. We describe the various models developed so far and the scientific questions that have been addressed using them. Finally, we discuss the limitations of these models, primarily ascribed to hosting a human immune system (HIS) in a xenogeneic recipient, and what remains to be done to improve their physiological relevance. MAJOR CONCLUSIONS: To understand the role of genetic and environmental factors or evaluate immune-modifying therapies in humans, it is critical to develop and apply models in which human cells can be manipulated and their functions studied under conditions that recapitulate as closely as possible the physiological conditions of the human body. While microphysiological systems and living tissue slices provide some of these conditions, HIS mice enable more extensive analyses using in vivo systems.