|
|
| (36 intermediate revisions by 6 users not shown) |
| Line 1: |
Line 1: |
| − | Back to [[DBP2]] | + | Back to [[DBP2:Main|NA-MIC DBP 2]] |
| | + | __NOTOC__ |
| | + | = Overview of UNC DBP 2 = |
| | + | == Longitudinal MRI study of early brain development in neuropsychiatric disorder: UNC Autism Study == |
| | | | |
| − | * '''Title:''' Longitudinal MRI study of early brain development in neuropsychiatric disorder: UNC Autism Study
| + | There is a lack of appropriate tools for processing of pediatric MRI, a challenging topic since pediatric MRI differs significantly from adult MRI due to variable brain shape and the process of maturation/myelination which are reflected in nonlinear shape/volume changes but also regional change of white matter. Working on a toolkit for the community would have a large impact, in particular also in view of existing and soon to be available databases of normative pediatric MRI (PI Alan Evans). Access to the UNC longitudinal pediatric MRI data representing a period of moderate but significant brain growth can spawn off interesting new software methodology developments from Core-1. Besides existing multi-modal MRI data, the UNC group has a very large set of segmented data (subcortical structures measured with very high reliability (0.92 up to 0.99) for over 140 MRI data sets(hippocampus, amygdale, putamen, pallide globe, caudate, ventricles) - to our knowledge the largest segmentation database of such high quality. These data could be used for shape analysis of growth trajectory and can also serve as a benchmark for novel semi-automated processing. The group has profound experience with the development of novel segmentation protocols (http://www.psychiatry.unc.edu/autismresearch/MRI_PAGE.htm) and the design of large-scale validation of segmentation methodology (see Yushkevich et al., 2006, NeuroImage, http://dx.doi.org/10.1016/j.neuroimage.2006.01.015). Moreover, the groups experience with state-of-the-art ITK/vtk processing tools will help to critically assess and improve the NA-MIC toolkit’s development from the viewpoint of users involved in large clinical studies. The processing of a relatively large database needs highly automated processing “pipelines”, i.e. co-registration of multi-modal data, atlas-to-template registration, automatic tissue segmentation, lobe parcellation, MRI-DTI registration, ROI analysis, and statistical analysis. This data therefore would be an excellent testbed for new automated Slicer 3 processing. A growth-rate analysis might have to include new methods for longitudinal image analysis, cortical thickness and cortical folding pattern analysis, methods not yet developed for the NA-MIC toolkit but required for human brain studies. [[DBP2:UNC:Introduction|More...]] |
| | | | |
| − | ==Team and Institute==
| + | Data is provided at the following link: '''[[Data:DBP2:UNC|UNC Data]]'''. |
| − | *Co-PI: Heather Cody Hazlett, PhD, (heather_cody at med.unc.edu, Ph: 919-966-4099)
| |
| − | *Co-PI: Joseph Piven, MD
| |
| − | *NA-MIC Engineering Contact: Jim Miller, GE Research
| |
| − | *NA-MIC Algorithms Contact: Martin Styner, UNC
| |
| | | | |
| − | * '''Affiliation/Institution:''' UNC Chapel Hill, Department of Psychiatry and the Neruodevelopmental Disorders Research Center NDRC
| + | = UNC Roadmap Project = |
| | | | |
| − | * '''Science:''' Multiple lines of converging evidence (from MRI, post-mortem and head circumference studies) indicate that brain enlargement in autism is a real phenomenon. However, the onset, trajectory and pattern of this enlargement (in brain tissues, regions and structures), relationship to developing neural circuitry and clinical features; and, pathogenesis, are not yet clear. Results from our longitudinal MRI study of brain development (2 years with follow-up at 4 years) demonstrate robust generalized enlargement of white and gray matter volume in cerebral cortex in autistic individuals (N= 51) by age 2 yrs. The MRI and earlier head circumference data strongly suggest a period of substantial brain overgrowth in autistic individuals between age 12 and 24 months, continuing into ages 4 and 5. This study will provide critical information about the trajectory of brain growth (regions, tissues, structures and fiber tracts) as measured on MRI and DTI, the potential relationship to clinical features; and underlying genetic etiology. These results will provide important insights into developmental brain and behavioral phenotypes, and neurobiological mechanisms in autism.
| + | {| cellpadding="10" border="1" style="background:lightblue;text-align:left;" |
| | | | |
| − | * ''' Benefits to NA-MIC'''<nowiki>: There is a lack of appropriate tools for processing of pediatric MRI, a challenging topic since pediatric MRI differs significantly from adult MRI due to variable brain shape and the process of maturation/myelination which are reflected in nonlinear shape/volume changes but also regional change of white matter:. Working on a toolkit for the community would have a large impact, in particular also in view of existing and soon to be available databases of normative pediatric MRI (PI Alan Evans). Access to the UNC longitudinal pediatric MRI data representing a period of moderate but significant brain growth can spawn off interesting new software methodology developments from Core-1. Besides existing multi-modal MRI data, the UNC group has a very large set of segmented data (subcortical structures measured with very high reliability (0.92 up to 0.99) for over 140 MRI data sets(hippocampus, amygdale, putamen, pallide globe, caudate, ventricles) - to our knowledge the largest segmentation database of such high quality. These data could be used for shape analysis of growth trajectory and can also serve as a benchmark for novel semi-automated processing. The group has profound experience with the development of novel segmentation protocols (</nowiki>http://www.psychiatry.unc.edu/autismresearch/MRI_PAGE.htm) and the design of large-scale validation of segmentation methodology (see Yushkevich et al., 2006, NeuroImage, http://dx.doi.org/10.1016/j.neuroimage.2006.01.015). Moreover, the groups experience with state-of-the-art ITK/vtk processing tools will help to critically assess and improve the NA-MIC toolkit’s development from the viewpoint of users involved in large clinical studies. The processing of a relatively large database needs highly automated processing “pipelines”, i.e. co-registration of multi-modal data, atlas-to-template registration, automatic tissue segmentation, lobe parcellation, MRI-DTI registration, ROI analysis, and statistical analysis. This data therefore would be an excellent testbed for new automated Slicer 3 processing. A growth-rate analysis might have to include new methods for longitudinal image analysis, cortical thickness and cortical folding pattern analysis, methods not yet developed for the NA-MIC toolkit but required for human brain studies.
| + | |style="width:15%" | [[Image:BrainDevelopment.jpg|200px]] |
| | + | |style="width:85%" | |
| | | | |
| − | * ''' Benefits to UNC NDRC group: ''' The UNC autism research group will have access to NAMIC tools not yet available for analysis, which will expose them to new tools and procedures beyond the ones locally developed. This will significantly expand their processing capabilities but also will allow them to do research within a larger team of leading image analysis research groups. New tools applied to the existing longitdudinal autism pediatric study, including raw image data and processed anatomical structures, are most likely lead to publications demonstrating the processing capabilities and versatility of the NAMIC toolkit.
| + | == [[DBP2:UNC:Cortical_Thickness_Roadmap|Cortical Thickness for Autism]] == |
| | | | |
| | + | Our goal is to begin a longitudinal study of early brain development by cortical thickness in autistic children and controls (2 years with follow-up at 4 years). We want to be able to make statistical group comparisons. [[DBP2:UNC:Cortical_Thickness_Roadmap|More...]] |
| | | | |
| − | ===Research Goals=== | + | <font color="red">'''New: '''</font> Participation in the NA-MIC Winter 2010 Project Week. |
| | | | |
| − | Multiple lines of converging evidence (from MRI, post-mortem and head circumference studies) indicate that brain enlargement in autism is a real phenomenon. However, the onset, trajectory and pattern of this enlargement (in brain tissues, regions and structures), relationship to developing neural circuitry and clinical features; and, pathogenesis, are not yet clear. Results from our longitudinal MRI study of brain development (2 years with follow-up at 4 years) demonstrate robust generalized enlargement of white and gray matter volume in cerebral cortex in autistic individuals (N= 51) by age 2 years. The MRI and earlier head circumference data strongly suggest a period of substantial brain overgrowth in autistic individuals between age 12 and 24 months, continuing into ages 4 and 5. This study will provide critical information about the trajectory of brain growth (regions, tissues, structures and fiber tracts) as measured on structural MRI and DTI, the potential relationship to clinical features; and underlying genetic etiology. These results will provide important insights into developmental brain and behavioral phenotypes, and neurobiological mechanisms in autism.
| + | |} |
| − | Our project has the following specific aims:
| |
| − | | |
| − | 1) To characterize the pattern of brain size and development at two cross-sectional time periods (ages 18-35 and 42-66 months) in autism and in comparison to controls (children with developmental delay and typical development).
| |
| − | | |
| − | 2) To examine the cross-sectional and longitudinal relationships between selected morphological brain features and the pattern of selected cognitive characteristics and behavioral abnormalities reported to be abnormal in autism.
| |
| − | | |
| − | ===Description of Data===
| |
| − | | |
| − | '''Subjects:''' Participants in this study include children with autism and controls. Controls include children with developmental delay (DD) and typical development (TYP). The initial age at entry for this longitudinal study (time point 1) is between 18 and 35 months. At this time a clinical and behavioral assessment (see below) and MRI scan is performed. A follow-up visit is conducted when subjects are between 42 and 66 months of age (approximately 24 months after their initial visit). At this time (time point 2) a repeat assessment and MRI scan is performed.
| |
| − | | |
| − | '''Inclusion/Exclusion Criteria for subjects with Autism:''' Subjects are eligible to enter the study at time 1 if they are between 18 and 35 months of age, and receive a clinical diagnosis of DSM-IV Autistic Disorder based on the following information: review of the patient’s symptoms, medical history and assessment data, observation (e.g., videotapes of Autism Diagnostic Observation Schedule-G (ADOS-G); and a physical exam. We exclude cases with: 1) medical conditions thought to be associated with autism (i.e., neurofibromatosis, tuberous sclerosis, PKU, and Fragile X Syndrome (FraX)); and, 2) gross CNS injury (e.g., Cerebral Palsy). Subjects are also excluded from the final analyses (and the longitudinal study and MRI scan at time 2) if they do not meet ADI-R criteria for autism at age 42-66 months or if they develop any of the exclusion criteria in the interval.
| |
| − | | |
| − | '''Inclusion/Exclusion Criteria for Controls:''' Controls are excluded at time 1 and 2 if they have evidence of a pervasive developmental disorder, a history of a neurological condition including CNS injury (e.g. Cerebral Palsy, severe closed head trauma) or significant perinatal suboptimality. Mentally retarded controls will be excluded if there is evidence of an identifiable or presumed etiology for their mental retardation (e.g., PKU, tuberous sclerosis), including Fragile X, a clear history of familial mental retardation, or if their mental retardation is part of a recognizable syndrome (e.g., Williams Syndrome).
| |
| − | | |
| − | '''MR Protocol:''' All scans are conducted on a 1.5 T Siemens scanner. Sequences include: a T1 weighted inversion recovery magnetization preparation (IR Prep) sequence with a slice thickness of 1.5 mms in the coronal plane, TE=5.4, TR=12.3, NEX=1, FOV=20 cm; flip angle = 40o; 256 X 192 matrix. PD/ T2 weighted fast spin-echo images will be acquired with the following parameters: 3.0 mm coronal slices, TE = 17/75, TR = 7200 ms, NEX = 1, FOV=20, 256 X 192 matrix. Tensor diffusion spin echo in the coronal and axial planes EPI, PSD-tensor, 4 shot, min TE, TR=12,000 ms, T1=2200, NEX=1, b value=1000, 128 X 64 matrix.
| |
| − | | |
| − | '''Assessment:''' For all the subjects with autism, diagnosis will be reviewed by interviewing the primary caretaker regarding the subject’s current and past behavior using the Autism Diagnostic Interview-Revised (ADI-R). The ADI-R is a semi-structured interview for autism. Items have been shown to be reliable and the accompanying algorithm adequately discriminates autistic individuals from mental-age matched non-autistic comparison subjects. Behavior: The Autism Diagnostic Observation Schedule -Generic (ADOS-G), is a semi-structured assessment of communication, social interaction and play or imaginative use of materials for individuals suspected of having autism or other pervasive developmental disorders (PDD). It will be administered to the subjects in the autism group. For purposes of this study, the ADOS-G will be used as a check on diagnoses from the ADI-R at both age periods and will provide additional information for the clinical diagnosis at age of entry. All controls will be screened (and excluded) for evidence of an autism spectrum disorder with the CARS. Cognitive: The Mullen Scales of Early Learning provide a single, reliable and valid instrument for estimation of IQ at both age periods in this study. At Time 2 cognitive development will also be assessed using the Differential Abilities Scale (DAS). Parents will also complete the Vineland Adaptive Behavior Scales with the clinician. This information will be incorporated into the clinical evaluation and the final judgment regarding diagnosis at entry. Language: All subjects will be administered the PreSchool Language Scale-4 (PLS-4) (at Time 1 and Time 2). The PLS-4 is a reliable and valid measure of receptive and expressive language from birth to 6 years of age. The MacArthur CDI Short Form Vocabulary Checklist (parent questionnaire) will be given at Time 1 and 2 to assess vocabulary use. Atypical Behavior: Behavior will be further characterized at Time 2 using the Aberrant Behavior Checklist (ABC), Repetitive Behavior Scale-Revised, Child Behavior Checklist (parent and teacher), and the Conners Parent and Teacher Rating Scales. These forms are parent questionnaires that report problem behaviors, such as hyperactivity and stereotyped behavior. Excellent convergence has also been shown between the CBCL and a number of discrete psychiatric syndromes demonstrating its usefulness as a rapid and useful screening instrument for selected psychiatric conditions. Use of this instrument will provide further descriptive information regarding the controls in this study. The Sensory Processing Assessment for Young Children (SPA) is a play-based observational assessment to be given at Time 1 to judge the child’s approach or avoidance to novel sensory toys, orientation and habituation to social and non-social sensory stimuli, and generation of novel action strategies with toys. The Sensory Experiences Questionnaire (SSQ) is a 25-item Likert-type parent questionnaire that asks about a child’s responses to various sensory stimuli (e.g., startling to loud noises, avoiding messy textures, etc.) in the context of functional daily activities. This will be given at T1 and T2. The SPA will be performed with the autism cases only, while the SSQ will be administered to all subjects. Physical Examination: A standardized physical examination will include examination for neurological abnormalities (asymmetries of the motor exam, hypotonia and reflexes), neurocutaneous abnormalities (including woods lamp examination) and dysmorphic features, and standardized measurement of head circumference and height. The ABC will be administered to controls, again as a way of obtaining further behavioral information on this sample. Similarly, data from the Connors Scales will provide comparable measures of attention, hyperactivity and impulsivity in both the case and control groups. Physical Exam: A standardized physical examination (including woods lamp) at entry and time 2 will include examination for neurological abnormalities (asymmetries of the motor exam, hypotonia and reflexes), neurocutaneous abnormalities (associated with CNS abnormalities such as tuberous sclerosis and neurofibromatosis), dysmorphic features (suggestive of a possible syndrome) and, standardized measurement of head circumference and height. In addition, handedness will be assessed at time 2, using a modification of the Edinborough Handedness Questionnaire. Laboratory: All autism and developmentally delayed children entering this study will have a Fragile X test (if not previously performed) to evaluate for any evidence of this chromosomal abnormality that will lead to exclusion from the study. DNA is collected at this time for autism and DD cases, and salivary DNA is collected from typical controls. Additional parent/family information: the Wechsler Abbreviated Scale of Intelligence (WASI) Verbal Scale only, Symptom Checklist-90-Revised, Family Economic/Educational Status, and the Family Environment Scale (FES) are used to obtain estimates of parental IQ, SES, and family functioning.
| |
| − | | |
| − | To date we have successfully scanned (i.e., usable MRIs) the following subjects:
| |
| − | | |
| − | Group Time1 Time2 (Follow-ups)
| |
| − | Autism 56 28
| |
| − | DD 10 6
| |
| − | TYP 18 8
| |
| − | | |
| − | Note that these cases have the behavioral and clinical assessment data described above in addition to their MRI data. Also, we are actively conducting follow-up (time 2) scans so these totals continue to increase.
| |
| − | | |
| − | ===Image Processing Needs===
| |
| − | There is a lack of appropriate tools for processing of pediatric MRI and DTI and our UNC lab has had to develop specific tools for this purpose. Pediatric datasets present a challenging since pediatric MRI differs significantly from adult MRI due to variable brain shape and the process of maturation (i.e., myelination) which is reflected not only in nonlinear shape/volume changes but also regional change of white matter. Working with a toolkit such as Slicer would have a large benefit to our group. Access to the UNC longitudinal pediatric MRI data can spawn interesting new software methodology developments from Core-1. Besides existing multi-modal MRI data, the UNC group already has a very large set of segmented data (subcortical structures measured with very high reliability (0.92 up to 0.99) for over 140 MRI data sets (hippocampus, amygdale, putamen, globus pallidus, caudate, ventricles) - to our knowledge the largest segmentation database of such high quality. These data could be used for shape analysis of growth trajectory and can also serve as a benchmark for novel semi-automated processing. Our group has experience with the development of segmentation tools (http://www.psychiatry.unc.edu/autismresearch/MRI_PAGE.htm) and the design of large-scale validation of segmentation methodology (see Yushkevich et al., 2006, NeuroImage, (http://dx.doi.org/10.1016/j.neuroimage.2006.01.015). Moreover, the group’s experience with state-of-the-art ITK/vtk processing tools will help to critically assess and improve the NA-MIC toolkit’s development from the viewpoint of users involved in large clinical studies.
| |
| − | | |
| − | Specific image analysis projects where we envision NA-MIC supporting our group include the following:
| |
| − | | |
| − | (1) Automated Segmentation. While we already have segmented data of many brain substructures, but only a handful of these are automated. The remaining protocols represent manual tracing projects. We would benefit from the development of semi-automated or fully automated segmentation tools for these structures. Given that we already have excellent reliability in the protocols we currently employ, our data could be useful to NA-MIC as a unique training, testing or validation dataset.
| |
| − | | |
| − | (2) Cortical Thickness. We currently have no means to perform localized cortical thickness analysis. Our existing method only provides a global or regional (based on cortical lobes) means to examine our data. Our project would benefit from NA-MIC support in that we could examine group differences in specific cortical regions of interest. The ability to study changes in cortical thickness over time, and between groups, would also be a needed area of support.
| |
| − | | |
| − | (3) DTI atlas matching. We currently have a tool that allows us to do automatic region based DTI analysis. However, we are restricted to white matter tracts that we have defined. With support from NA-MIC, we could add to our ability to examine differences in white matter between cases and controls by the development of an enhanced atlas-based tool that could allow us to do both regional and localized full brain DTI analysis.
| |
| − | | |
| − | (4) Shape analysis. Our existing shape analysis tools do not allow us to do quantitatively examine shape differences over time in correlation with clinical variables of interest (such as IQ, gender, and age). The ability to study shape changes in substructures along with the influence of clinical and/or behavioral differences in cases versus controls would be extremely valuable.
| |
| − | | |
| − | ===Current Image Processing===
| |
| − | | |
| − | A number of image processing tools have been created during the course of this study to assist in the image processing for this grant. These methods were created by our colleague’s (Dr. Guido Gerig) image processing lab at UNC through direct support from this project. The tools include:
| |
| − | | |
| − | 1. Imagine – a tool used to create image processing pipelines
| |
| − | | |
| − | 2. Intensity Rescaler – performs intensity windowing to make tissue values in two images the same range (e.g., class matching program)
| |
| − | | |
| − | 3. ImConvert – image converter to change images from one format to another
| |
| − | | |
| − | 4. Atlas Builder – generate average atlas, obtain deformation fields for individual images
| |
| − | | |
| − | 5. Circumference – to obtain circumference of perimeter labels (e.g. brain mask)
| |
| − | | |
| − | 6. DTI Checker – to check quality, correct minor alignment problems, remove slices with artifact
| |
| − | | |
| − | 7. FiberTracking - for tractography of Diffusion Tensor data (DTI)
| |
| − | | |
| − | 8. FiberViewer - for quantitative analysis of fibers
| |
| − | | |
| − | 9. CC Segmenter – an automatic corpus callosum segementation and parcellation tool. Currently segements the corpus callosum into 4 regions associated with cortical lobes.
| |
| − | | |
| − | 10. DicomImsel, DicomConvert, Xmedcon – other image conversion tools
| |
| − | | |
| − | 11. Substructure shape analysis – tools for conducting shape analysis of the hippocampus and caudate.
| |
| − | | |
| − | These tools work successfully to assist in the processing of our 2 and 4 year old image datasets. However, the generation of automated pipelines incorporating these tools would be advantageous. This would decrease labor needed and potential ‘user error” and increase productivity in our lab.
| |
| − | | |
| − | Current limitations of some tools, such as shape processing of hippocampus and caudate, include the inability to correlate shape with our clinical data and interpret this in a quantitative way. For example, we would be interested in examining longitudinal shape changes in relation to cognitive ability (IQ) and variables related to severity of autism (e.g., presence or absence of ritualistic repetitive behaviors).
| |
| − | | |
| − | ===Plans for the NA-MIC kit===
| |
| − | The processing of our relatively large pediatric database will benefit from the development of highly automated processing “pipelines” (i.e. co-registration of multi-modal data, atlas-to-template registration, automatic tissue segmentation, lobe parcellation, MRI-DTI registration, ROI analysis) that can generate quantitative data used in statistical analyses. This dataset therefore would be an excellent testing ground for the automated 3D Slicer version 3.
| |
| − | Plans to develop pipelines for growth-rate analysis might include new methods for longitudinal image analysis, cortical thickness and cortical folding pattern analysis, methods not yet developed for the NA-MIC toolkit but required for human brain studies. Additional projects would include a more regionally defined DTI analysis where properties of specific regions of interest could be measured. Automating the DTI processing pipeline currently used by our lab would also be desirable. Finally, the development of new segmentation protocols for specific cortical regions, such as the dorsolateral prefrontal cortex, would also advance the goals of this project.
| |
| − | As part of this effort, we would support a computer science engineer who will be responsible for communicating with NA-MIC developers and applying new tools to our data. We also have the support of image processing research assistants in our lab to assist with the work of this CS engineer. We anticipate efficient and timely communication between our lab and NA-MIC so that new tools are being tested in a collaborative working relationship.
| |
| − | | |
| − | ===Future Directions===
| |
| − | | |
| − | We are in the process of submitting new grant proposals to conduct follow-up assessments and scans on the sample described in this proposal (2 and 4 year olds with autism, DD, and TYP). This will provide a third longitudinal data point between ages 6-8 years old for this existing sample. For this endeavor, we will require new tools to perform automated segmentation (e.g., brain tissue, substructures) of the MRI scans of the 6-8 year olds, as well as the creation of new tools to perform longitudinal data analysis (e.g., shape, cortical thickness) from the 2 to 8 year old age range.
| |
| − | | |
| − | ===VII. Summary===
| |
| − | | |
| − | The UNC autism research group will benefit from access to NA-MIC tools and support from the NA-MIC group. This will significantly expand our image processing capabilities and will allow our team to do research within a larger team of leading image analysis research groups. NA-MIC will benefit from our unique longitudinal dataset of structural and DTI data. Our site will work with NA-MIC to produce image processing tools that (1) are applicable for pediatric datasets, (2) address the need to examine longitudinal brain development, (3) automate pipelines for image processing, and (3) incorporate existing tools into the NA-MIC toolkit.
| |
