Mining Human Brain Data: Analysis, Classification, and Visualization

Contact Information

List of Supported Students and Staff

• Prof. Fillia S. Makedon, PI (2 summer months per year).
• Dr. Andrew J. Saykin, Co-PI (2 summer months per year).
• Prof. Vasileios Megalooikonomou, Co-PI (1 summer month per year).
• James C. Ford (Dartmouth), Li Shen (Dartmouth), and Despina Kontos, (Temple), Ph.D. students.
• Prof. Christos Faloutsos, consultant (Carnegie Mellon).

Project Award Information

• Award Number: IDM 0083423
• Duration: 5/1/2001 - 4/30/2004
• Title: Mining Human Brain Data: Analysis, Classification, and Visualization
• Keywords: Data mining, fmri, human brain activation, functional-structural correlation

Project Summary

This project is a multidisciplinary effort to develop methods for content based retrieval, analysis, and visualization of probabilistic 3D spatial maps. In particular, we focus on the ability to mine associations in human brain activation data. We use the statistical parametric maps obtained from the analysis of fMRI (functional Magnetic Resonance Imaging) brain activations as an example of probabilistic 3D maps. Our goals are to make accurate classifications of activation maps into diagnosis groups, and to develop tools that enable queries of the form: "find an activation like this...".

Publications and Products

Year 1 (May 2001 - April 2002)

[1] A. Lazarevic, D. Pokrajac, V. Megalooikonomou and Z. Obradovic, "Distinguishing Among 3-D Distributions for Brain Image Data Classification", in Proceedings of the 4th International Conference on Neural Networks and Expert Systems in Medicine and Healthcare, Milos Island, Greece, pp. 389-396, June 2001.

[2] D. Pokrajac, A. Lazarevic, V. Megalooikonomou and Z. Obradovic, "Classification of brain image data using meaasures of distributional distance", presented at the 7th Annual Meeting of the Organization for Human Brain Mapping (OHBM01), Brighton, UK, June, 2001.

[3] J. Ford, F. Makedon, V. Megalooikonomou, A. Saykin, L. Shen, T. Steinberg, "Spatial Comparison of fMRI Activation Maps for Data Mining: A Methodology of Hierarchical Characterization and Classification", presented at the 7th Annual Meeting of the Organization for Human Brain Mapping (OHBM01), Brighton, UK, June 2001.

[4]L. Shen, L. Cheng, F. Teng, F. Makedon, J. Ford, T. Steinberg, and A. J. Saykin, "A Multimedia System for Tracing and Studying Regions-of-Interest in Brain Images", IEEE Multimedia Technology and Application Conference, Irvine, CA, pp 238-245, November 2001.

[5] V. Megalooikonomou, D. Pokrajac, A. Lazarevic, Z. Obradovic, "Effective classification of 3-D image data using partitioning methods", in Proceedings of the SPIE Conference on Visualization and Data Analysis, San Jose, CA, pp. 62-73, Jan. 2002.

[6] V. Megalooikonomou, "Evaluating the performance of association mining methods in 3-D medical image databases", in Proceedings of the 2nd SIAM International Conference on Data Mining, Arlington, VA, pp. 474-494, April 2002.

Year 2 (May 2002 - April 2003)

[7] A. J. Saykin, H. A. Wishart, L. A. Flashman, T. W. McAllister, T. McHugh, J. C. Ford, L. Shen, T. Steinberg, and F. Makedon, "Structure/function relationships in brain disorders: Strategies for mining volume, shape, lesion and BOLD fMRI activation data", paper presented at Society for Biological Psychiatry meeting, Philadelphia, PA, 2002.

[8] V. Megalooikonomou, D. Pokrajac, A. Lazarevic, D. Kontos and Z. Obradovic, "Analysis and Classification of Regions of Interest in 3D Medical Images", chapter in Bioimaging and Applications, N. Bourbakis (ed.), Kluwer Academic Publishers, 2003. Accepted.

[9] V. Megalooikonomou, H. Dutta, D. Kontos, "Fast and Effective Characterization of 3D Region Data", in Proceedings of the IEEE International Conference on Image Processing (ICIP), Rochester, NY, pp. 421-424, Sept. 2002.

[10] J. Ford, L. Shen, F. Makedon, L. A. Flashman, and A. J. Saykin, "A Combined Structural-Functional Classification of Schizophrenia using Hippocampal Volume plus fMRI Activation", EMBS-BMES2002 Second Joint Meeting of the IEEE Engineering in Medicine and Biology Society and the Biomedical Engineering Society, 2002.

[11] L. Shen, J. Ford, F. Makedon, and A. Saykin, "Hippocampal Shape Analysis: Surface-based Representation and Classification", SPIE Medical Imaging 2003: Conference 5032 - Image Processing, San Diego, California, February 2003.

[12] V. Megalooikonomou, D. Kontos, D. Pokrajac, A. Lazarevic, Z. Obradovic, O. Boyko, A. Saykin, J. Ford, F. Makedon, "Classification and Mining of Brain Image Data Using Adaptive Recursive Partitioning Methods: Application to Alzheimer Disease and Brain Activation Patterns", Human Brain Mapping Conference (OHBM'03), June 2003.

[13] L. Shen, J. Ford, F. Makedon, L. Flashman, and A. Saykin, "Surface-based Morphometric Analysis for Hippocampal Shape in Schizophrenia", Human Brain Mapping Conference (OHBM'03), June 2003.

[14] J. Ford, H. Farid, F. Makedon, L.A. Flashman, T.W. McAllister, V. Megalooikonomou, and A.J. Saykin, "Patient Classification of fMRI Activation Maps", 6th Annual International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI'03), 2003. Accepted.

[15] L. Shen, J. Ford, F. Makedon, Y. Wang, T. Steinberg, S. Ye, and A. Saykin, "Morphometric Analysis of Brain Structures for Improved Discrimination", 6th Annual International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI'03), 2003. Accepted.

[16] M. Sobel, V. Megalooikonomou, and D. Kontos, "Statistical Techniques for the Characterization of Regions in Noisy Settings", Journal of the American Statistical Association (JASA), 2003. In review.

[17] L. Shen, J. Ford, F. Makedon, and A. Saykin, "Effective Classification of 3D Closed Surfaces: Application to Modeling Neuroanatomical Structures", International Conference on Computer Vision, Pattern Recognition and Image Processing in conjunction with Seventh Joint Conference On Information Sciences (JCIS2003), Cary, North Carolina, Sept. 26 - 30, 2003. Accepted.

Year 3 (May 2003 - April 2004)

[18] J. Ford, Patient Classification from fMRI Brain Activation Patterns, Ph.D. thesis, Dartmouth College Department of Computer Science, July 2003.

[19] D. Kontos, V. Megalooikonomou, N. Ghubade, and C. Faloutsos, "Detecting discriminative functional MRI activation patterns using space filling curves", 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), Cancun, Mexico, 2003. Accepted.

[20] D. Kontos, V. Megalooikonomou, M. Sobel, and H. Dutta, "A Feature Selection Technique for Classfication and Similarity Searches of Spatial Region Data", IEEE International Conference on Data Mining (ICDM'03), Melbourne, Florida, USA, Nov. 19-22, 2003. In review.

[21] D. Pokrajac, V. Megalooikonomou, A. Lazarevic, D. Kontos, and Z. Obradovic, "Classification of 3D Medical Images based on Spatial Distribution Estimation", Artificial Intelligence in Medicine. In review.

Project Impact

It is clear that the future of neuroscience research may be affected by the ability to do large-scale mining of fMRI brain activations. This sort of endeavor can extract associations at multiple levels: among different activations for the same or different (groups of) individuals, between activations and tasks, between activations and ancillary data, and between tasks and ancillary data. Automatic characterization tools open the field to numerous new research opportunities, as they would offer common signature formats to do cross-modality brain data searches and correlation over multiple studies. In turn, the mined results would speed up discovery in the neuroscience field.

Goals, Objectives, and Targeted Activities

We have four major objectives:

1. Provide a common access to fMRI activation data with automatic characterization and classification tools for probabilistic 3D brain activation maps (p-volumes), and derive activation signatures to facilitate mining and the creation of taxonomies of activations.
2. Demonstrate these methods with the creation of a prototype robust database of brain activations that includes scan data as well as biographical, clinical, historical and other subject-centered ancillary data. This database will enable retrieval requests and subsequent analysis incorporating spatial statistics, classification, and visualization techniques.
3. Mine the prototype database and provide visualizations of the intermediate and final results to extract interesting new conclusions from existing data and simulated data.
4. Provide an evaluation of the system by using real and simulated data and making the tools and the prototype database accessible, with data mining facilities to be tested by the professional communities.

We will also eventually integrate structural analysis data that can be correlated seamlessly with the activation data and contribute to the ability to mine correlations between function and structure. This goal is served by shape analysis of brain structures which may contribute to the characterization and classification of brain activations.

Current and Future Activities

The following are some of the highlights of our ongoing work, with references to the Publications and Products section above:

1. Development of novel classification techniques for three-dimensional region data [1-3].
2. Development of efficient and effective techniques for characterizing image data and facilitating similarity queries or content-based retrieval [5].
3. Initial attempts to construct of a framework for evaluating many aspects of association mining methods in medical image databases, and in particular methods that can be used to discover associations between brain activations and tasks performed [6].
4. Investigation of Fisher Linear Discriminant (FLD) analysis as a non-threshold based tool for comparing brain activation maps [14,18].
5. Integration of structural information, including structure volume [6] as well as structure shape [11,13,15,17] and lesion data [7]. We are currently investigating the extension of our data mining approach to include structural brain information.

Additional future work will include the following:

1. Development of BRASS (BRain Access Support System), an online interactive database of brain anatomy and activations, will facilitate research in the field.

Other Project References

[22] J. Ford, F. Makedon, T. Steinberg, C. B. Owen, S. Johnson, and A. J. Saykin, "Stimulus tracking in functional magnetic resonance imaging (fMRI)", ACM Multimedia '98, September 1998.

[23] A. J. Saykin, L. A. Flashman, S. Frutiger, S. C. Johnson, A. C. Mamourian, C. H. Moritz, J. R. O'Jile, H. J. Riordan, R. B. Santulli, C. A. Smith, and J. B. Weaver, "Neuroanatomic Substrates of Semantic Memory Impairment in Alzheimer's Disease: Patterns of Functional MRI Activation", J. of the International Neuropsychological Society, 5:377-392, 1999.

[24] V. Megalooikonomou, C. Davatzikos, and E. H. Herskovits, "Mining Lesion-Deficit Associations in a Brain Image Database", in Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Diego, CA, pp. 347-351, Aug. 1999.

[25] E. H. Herskovits, V. Megalooikonomou, C. Davatzikos, A. Chen, R. N. Bryan, and J. Gerring, "Is the spatial distribution of brain lesions associated with closed-head injury predictive of subsequent development of attention-deficit hyperactivity disorder? Analysis with brain image database", Radiology, 213: 389-394, 1999.

[26] V. M. Megalooikonomou, J. Ford, L. Shen, F. Makedon, and A. Saykin, "Data Mining in Brain Imaging" (Invited Paper), Statistical Methods in Medical Research, vol. 9, pp. 359-394, 2000.

[27] V. Megalooikonomou, E. H. Herskovits, and C. Davatzikos, "A Simulator for Evaluating Methods for the Detection of Lesion-Deficit Associations", Human Brain Mapping, 10:61-73, 2000.

[28] S. C. Johnson, A. J. Saykin, L. C. Baxter, L. A. Flashman, R. B. Santulli, T. W. McAllister, and A. C. Mamourian, "The Relationship Between fMRI Activation and Cerebral Atrophy: Comparison of Normal Aging and Alzheimer Disease", Neuroimage, 11:179-187, 2000.

[29] F. Makedon, "Tools Towards Mining Human Brain Data" (Invited Talk), World Conference on the WWW and Internet, San Antonio, Texas, October 30-November 4, 2000.

[30] S. C. Johnson, A. J. Saykin, L. A. Flashman, T. W. McAllister, and M. Sparling, "Brain activation on fMRI and verbal memory ability: Functional neuroanatomic correlates of CVLT performance", in J. of the International Neuropsychological Society, 7, 55-62, 2001.

[31] A. J. Saykin, L. A. Flashman, L. Shen, J. Ashburner, M. Sparling, A. Donnelly, F. Makedon, D. Isecke, J. C. Ford, V. Megalooikonomou, and T. W. McAllister, "Hippocampal shape in schizophrenia: A deformation-based morphometric analysis", in Neuroimage, 13(6), S1096, June 2001.

[32] H. A. Wishart, A. J. Saykin, A. C. Mamourian, L. A. Flashman, C. E. Fadul, K. A. Ryan, J. C. Ford, and L. H. Kasper, "Working memory in multiple sclerosis: Relation of lesion burden to fMRI activation", in NeuroImage: Organization for Human Brain Mapping issue, 13(6), S762, June 2001.

[33] V. Megalooikonomou, and E. H. Herskovits, "Mining Structure-Function Associations in a Brain Image Database", chapter in Medical Data Mining and Knowledge Discovery , K.J. Cios (ed.), Springer-Verlag, 2001.

[34] L. Cheng, TROI - A Training and Research System For Tracing Regions of Interest in Brain Images, Master's thesis, Dartmouth College Dept. of Computer Science, June 2000.

[35] L. C. Baxter, A. J. Saykin, L. A. Flashman, S. C. Johnson, S. Guerin, and H. Wishart, "Sex differences in semantic processing: A functional MRI study", in Brain and Language, in press.

Area Background

Functional MR brain imaging is a non-invasive technique that has attracted enormous interest in recent years because of its potential to solve fundamental problems explaining cognitive processes and to find relations between brain structure and function. As a result, there is an explosion of neuroscience research results that wait to be characterized, classified, and made amenable to database and data mining technologies. Recognizing the high cost of building large repositories and their importance in understanding cognitive processes, our environment will fully utilize fMRI's potential by enabling inter- and intra-study data mining. Our techniques may also provide an invaluable diagnosis aid in numerous clinical fMRI applications.

Area References

1. G. M. Euripides, M. Petrakis, and C. Faloutsos, "Similarity Searching in Medical Image Databases," IEEE Transactions on Knowledge and Data Engineering, vol. 9, pp. 435-447, 1997.
2. M. S. Cohen and R. M. DuBois, "Stability, Repeatability, and the Expression of Signal Magnitude in Functional Magnetic Resonance Imaging," J. Magnetic Resonance Imaging, vol. 10, pp. 33-40, 1999.
3. K. J. Friston, A. P. Holmes, K. J. Worsley, J. P. Poline, C. D. Frith, and R. S. J. Frackowiak, "Statistical Parametric Maps in Functional Imaging: A General Linear Approach," Human Brain Mapping, vol. 2, pp. 189-210, 1995.

Potential Related Projects

We welcome contacts from other researchers interested in collaborations.