CAREER: Extracting Patterns from Medical Image Databases

Supported by the National Science Foundation (NSF) - IIS-0237921

Duration: 9/2003-9/2008


Principal Investigator

Vasileios Megalooikonomou
Department of Computer and Information Sciences

Temple University
314 Wachman Hall

1805 N. Broad Street

Philadelphia PA 19122

215-204-5774
vasilis@cis.temple.edu
http://www.cis.temple.edu/~vasilis

List of Supported Students and Staff

Other collaborators

  • Prof. Christos Faloutsos (CMU)
  • Prof. Zoran Obradovic, Marcus Sobel, Scott faro, Feroze Mohamed, Longin Jan Latecki, Rolf Lakamper (Temple University)
  • Prof. Andrew Saykin (Indiana Univ.)
  • Prof. Fillia Makedon (Univ. of Texas at Arlington)
  • Prof. Nick Bryan, James Gee, Andrew Maidment, Predrag Bakic (Univ. of Pennsylvania)
  • Prof. Dragoljub Pokrajac (Delaware State Univ.)
  • Postdoc: Dr. Alexandar Lazarevic (Univ. of Minnesota)
  • Postdoc: Dr. James Ford (Dartmouth College)
  • Graduate students: Guoqiang Shan, Qiang Wang, Guo Li, Erickson Miranda, Michael Barnathan, Rui Li, Li An, Krishna Kumaraswamy, Nilesh Ghubade (Temple Univeristy and CMU)
  • Undergraduate students: Alexander Cohn, Ailar Javadi, Dan Mulhern, Joseph Dangelmaier

Keywords

Data mining

medical image analysis
image retrieval

image classification

similarity searches

Project Summary

The focus of this career development plan is to build an education and research program that will focus on the discovery of patterns and relations between anatomy (structure) and function through the effective and efficient analysis of large repositories of medical images and other clinical data. Medical centers almost everywhere today are facing an interesting challenge in analyzing the huge volumes of image and associated clinical data collected daily as part of several ongoing studies. By focusing on the regions of interest (ROIs), the approach uses novel techniques to extract their most discriminative features and uses them in classification and similarity searches. New representations of the information content of medical images are also provided. Moreover, spatial data mining tools are developed to efficiently discover associations between image data and non-image (functional) data. The approaches have applicability to medical images from a wide range of modalities (e.g., CT, MRI, fMRI, angiography, confocal microscopy, etc) showing normal and abnormal conditions of various structures. Information about the function of structures related to various medical conditions is extracted from clinical assessment. The educational plan focuses on providing students with a solid theoretical background and practical experience on data mining and its applications in medicine, promote interdisciplinary learning, and enable the training of a more versatile type of scientist.  

Publications and Products

Q. Wang, V. Megalooikonomou, C. Faloutsos, "Time Series Analysis with Multiple Resolutions", Information Systems, Vol. 35, No. 1, pp. 56074, 2010.

V. Megalooikonomou, M. Barnathan, D. Kontos, P. R. Bakic, A. D.A. Maidment, "A Representation and Classification Scheme for Tree-like Structures in Medical Images: Analyzing the Branching Pattern of Ductal Trees in X-ray Galactograms", IEEE Transactions on Medical Imaging, Vol. 28, Issue 4, pp. 487-493, 2009.

Q. Wang, V. Megalooikonomou, "A Performance Evaluation Framework for Association Mining in Spatial Data", Journal of Intelligent Information Systems, (accepted).

D. Kontos, V. Megalooikonomou, J. Gee, "Morphometric analysis of brain images with reduced number of statistical tests: a study on the gender-related differentiation of the corpus callosum", Artificial Intelligence in Medicine, Vol. 47, No. 1, pp. 75-86, 2009.

L. An, H. Xie, M.H. Chin, Z. Obradovic, D.J. Smith and V. Megalooikonomou, "Analysis of multiplex gene expression maps obtained by voxelation", BMC Bioinformatics 2009, Vol. 10 (Suppl 4):S10, (in press).

L. An, Z. Obradovic, D. Smith, O. Bodenreider and V. Megalooikonomou, "Mining Association Rules among Gene Functions in Clusters of Similar Gene Expression Maps", Proceedings of the Workshop on Data Mining in Functional Genomics, IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Washington, D.C., Nov. 2009.

H. Ling, M. Barnathan, V. Megalooikonomou, P. R. Bakic, and A.D.A. Maidment, "Probabilistic Branching Node Detection Using Hybrid Local Features", IEEE International Symposium on Biomedical Imaging (ISBI), Boston, MA, 2009.

A. Skoura, M. Barnathan, and V. Megalooikonomou, "Classification of Ductal Tree Structures in Galactograms", IEEE International Symposium on Biomedical Imaging (ISBI), Boston, MA, 2009.

E. Miranda, G. Shan, and V. Megalooikonomou, "Performing Vector Quantization Using Reduced Data Representation", Proceedings of the Data Compression Conference, Salt Lake City, Utah, 2009.

Q. Wang and V. Megalooikonomou, "A Dimensionality Reduction Technique for Efficient Time Series Similarity Analysis", Information Systems, Vol. 33, No. 1, pp. 115-132, 2008.

V. Megalooikonomou, D. Kontos, D. Pokrajac, A. Lazarevic and Z. Obradovic, "An adaptive partitioning approach for mining discriminant regions in 3D image data", Journal of Intelligent Information Systems, Vol. 31, No. 3, pp. 217-242, 2008.

L. An, H. Xie, M. Chin, Z. Obradovic, D. Smith, V. Megalooikonomou, "Analysis of Multiplex Gene Expression Maps Obtained By Voxelation", Proceedings of the IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Philadelphia, USA, 2008.

M. Barnathan, J. Zhang, V. Megalooikonomou, "A Web-Accessible Framework for the Automated Storage and Texture Analysis of Biomedical Images", Proceedings of the 5th IEEE International Symposium on Biomedical Imaging (ISBI), Paris, France, 2008.

M. Barnathan, J. Zhang, E. Miranda, V. Megalooikonomou, S. Faro, H. Hensley, L. D. Valle, K. Khalili, J. Gordon, F. B. Mohamed, "A Texture-Based Methodology For Identifying Tissue Type in Magnetic Resonance Images", Proceedings of the 5th IEEE International Symposium on Biomedical Imaging (ISBI), Paris, France, 2008.

M. Barnathan, R. Li, V. Megalooikonomou, F. Mohamed, S. Faro, "Wavelet Analysis of 4D Motor Task fMRI Data", Proceedings of Computer Assisted Radiology and Surgery (CARS), Barcelona, Spain, 2008.

M. Barnathan, J. Zhang, D. Kontos, P. Bakic, A. Maidment, V. Megalooikonomou, "Analyzing Tree-Like Structures In Biomedical Images Based On Texture And Branching: An Application To Breast Imaging", Proceedings of the International Workshop on Digital Mammography (IWDM), Tucson, AZ, 2008.

C. Faloutsos and V. Megalooikonomou, "On Data Mining, Compression, and Kolmogorov Complexity", Data Mining and Knowledge Discovery, Tenth Anniversary Issue, Vol. 15, No. 1, pp. 3-20(18), 2007.

Kontos, D., Megalooikonomou, V., Sobel, M., "A Statistical Approach for Selecting Discriminative Features of Spatial Regions of Interest", Intelligent Data Analysis, vol. 11, No. 2, pp. 111-135, 2007.

L. Latecki, V. Megalooikonomou, Q. Wang, D. Yu, "An Elastic Partial Shape Matching Technique", Pattern Recognition, Vol. 40, No. 11, pp. 3069-3080, 2007.

Megalooikonomou, V., Kontos, D., "A model for distributed analysis of medical image data across clinical information repositories", IEEE Engineering in Medicine and Biology Magazine, Vol. 26, No. 5, pp. 36-42, 2007.

L. J. Latecki, Q. Wang, S. Koknar-Tezel, and V. Megalooikonomou, "Optimal Subsequence Bijection", Proceedings of the IEEE International Conference on Data Mining, Omaha, NE, pp. 565-570, 2007.

J. Zhang and V. Megalooikonomou, "An effective and efficient technique for searching for similar brain activation patterns", Proceedings of the IEEE International Symposium on Biomedical Imaging, 2007.

Q. Wang, E. Karamani-Liacouras, E. Miranda, U.S. Kanamala, V. Megalooikonomou, "Classification of brain tumors using MRI and MRS", Proceedings of the SPIE Conference on Medical Imaging, 2007.

V. Megalooikonomou, J. Zhang, D. Kontos, P.R. Bakic, "Analysis of texture patterns in medical images with an application to breast imaging", Proceedings of the SPIE Conference on Medical Imaging, 2007.

D. Kontos, V. Megalooikonomou, A. Javadi, P. Bakic, A. Maidment, "Classification of Galactograms Using Fractal Properties of the Breast Ductal Network", Proceedings of the IEEE International Symposium on Biomedical Imaging: From Nano to Macro (ISBI), Arlington, Virginia, April 6-9, 2006.

V. Megalooikonomou, D. Kontos, J. Danglemaier, A. Javadi, P. A. Bakic, A.D.A. Maidment, "A representation and classification scheme for tree-like structures in medical images: An application on branching pattern analysis of ductal trees in x-ray galactograms", Proceedings of the SPIE Conference on Medical Imaging, San Diego, California, Feb. 2006.

P. Bakic, D. Kontos, V. Megalooikonomou, A. Maidment, "Comparison of Methods for Classification of Breast Ductal Branching Patterns" Proceedings of the 8th International Workshop on Digital Mammography (IWDM), Manchester, England, 2006.

Pokrajac, D., Megalooikonomou, V., Lazarevic, A., Kontos, D., Obradovic, Z., "Applying Spatial Distribution Analysis Techniques to Classification of 3D Medical Images", Artificial Intelligence in Medicine, Vol. 33, No. 3, pp. 261-280, Mar. 2005.

Kontos, D. and Megalooikonomou, V., "Fast and Effective Characterization for Classification and Similarity Searches of 2D and 3D Spatial Region Data", Pattern Recognition, Vol. 38, No. 11, pp. 1831-1846, 2005.

V. Megalooikonomou, Q. Wang, G. Li, C. Faloutsos, "A Multiresolution Symbolic Representation of Time Series", Proceedings of the 21st IEEE International Conference on Data Engineering (ICDE05), Tokyo, Japan, April 5-8, 2005, pp. 668-679.

V. Megalooikonomou, D. Kontos, "Integrating clinical information repositories: A framework for distributed analysis of medical image data", Proceedings of the 5th International Network Conference (INC 2005), Special Session on Image, Signal and Distributed Data Processing for Networked eHealth Applications, Samos Island, Greece, July 2005, pp. 545-552.

Q. Wang, V. Megalooikonomou, "A clustering algorithm for intrusion detection", in Proccedings of the SPIE Conference on Data Mining, Intrusion Detection, Information Assurance, and Data Networks Security, Orlando, Florida, USA, March 28 - April 1, Vol. 5812, pp. 31-38, 2005.

D. Kontos, V. Megalooikonomou and J. Gee, "Reducing the computational cost for statistical medical image analysis: An MRI study on the sexual morphological differentiation of the corpus callosum", in Proceedings of the 18th IEEE International Symposium on Computer-Based Medical Systems (CBMS05), Trinity College Dublin, Ireland, June 23-24, pp. 282-287, 2005.

Q. Wang, V. Megalooikonomou, G. Li, "A Symbolic Representation of Time Series", Proceedings of the IEEE Eighth International Symposium on Signal Processing and Its Applications (ISSPA'05), Sydney, Australia, Aug. 28-31, 2005, pp. 655-658.

V. Megalooikonomou, D. Kontos, N. DeClaris and P. Cano, "Utilizing Domain Knowledge in Neural Network Models for Peptide-Allele Binding Prediction", Proceedings of the 2005 IEEE Symposium on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB'05), San Diego, California, Nov. 2005.

L. J. Latecki, V. Megalooikonomou, Q. Wang, R. Lakaemper, C. A. Ratanamahatana, and E. Keogh, "Elastic Partial Matching of Time Series", Proceedings of the 9th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD'05), Porto, Portugal, Lecture Notes in Computer Science, Vol. 3721, pp. 577-584, 2005.

Kumaraswamy, K., Megalooikonomou, V. and Faloutsos, C., "Fractal Dimension and Vector Quantization", Information Processing Letters, vol. 91, no. 3, pp. 107-113, 2004.

Kontos, D. and Megalooikonomou, V., "Fast and Effective Characterization of 3D Region of Interest in Medical Image Data," in Proceedings of the SPIE International Symposium on Medical Imaging, San Diego, CA, Volume 5370 Medical Imaging 2004, pp. 1324-1331, 2004.

Wang, Q., Kontos, D., Li, G. and Megalooikonomou, V., "Application of Time Series Techniques to Data Mining and Analysis of Spatial Patterns in 3D images," in Proceedings of the International Conference on Acoustics, Speech and Signal Processing, (ICASSP), Montreal, Canada, pp.525-528, 2004.

Kumaraswamy, K., Faloutsos, C., Shan, G. and Megalooikonomou, V., "Relation between Fractal Dimension and Performance of Vector Quantization," in Proceedings of the Data Compression Conference (DCC'04), Salt Lake City, UT, pp. 547, 2004.

Kontos, D., Megalooikonomou, V., Sobel, M., Wang, Q., "An MCMC Feature Selection Technique for Characterizing and Classifying Spatial Region Data," in Lecture Notes in Computer Science 3138, Proceedings of the International Workshop on Syntactical and Structural Pattern Recognition (SSPR 2004) and Statistical Pattern Recognition (SPR 2004), Lisbon, Portugal, pp. 379-387, 2004.

Megalooikonomou, V., Wang, Q., Kontos, D., Li, G., Ford, J., Saykin, A., "Analysis of Brain Image Data using Sequence Analysis Techniques," in Proceedings of the Human Brain Mapping Conference (OHBM'04), Budapest, Hungary, June 13-17, 2004.

Kontos, D., Megalooikonomou, V., Wang, Q., Ford, J., Makedon, F., Saykin, A., "Identifying Discriminative fMRI Activation Signatures in Alzheimer's Disease: Studying a Series of Semantic Decision Tasks," in Proceedings of the Human Brain Mapping Conference (OHBM'04), Budapest, Hungary, June 13-17, 2004.

Kontos, D., Megalooikonomou, V., Pokrajac, D., Lazarevic, A., Obradovic, Z., Boyko, O.B., Ford, J., Makedon, F., Saykin, A.J., "Extraction of Discriminative Functional MRI Activation Patterns and an Application to Alzheimer's Disease," in Proceedings of the 7th Annual International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI'04), Rennes-Saint Malo, Sept. 26-30, 2004, Part II, Lecture Notes in Computer Science 3217, Vol. 2, pp. 727-735, 2004.

Lakamper, R., Latecki, L.J., Megalooikonomou, V., Wang, Q., Wang, X., "Learning Descriptive and Distinctive Parts of Objects with a Part-Based Shape Similarity Measure", in Proceedings of the 6th International Conference on Signal and Image Processing (SIP'04), Honolulu, Hawaii, Aug. 2004.

V. Megalooikonomou, G. Li, Q. Wang, "A Dimensionality Reduction Technique for Efficient Similarity Analysis of Time Series Databases", in Proceedings of the 13th ACM Conference on Information and Knowledge Management (ACM CIKM 04), Washington, DC, USA, November 8-13, pp. 160-161, 2004.

Kontos, D., Megalooikonomou, V., Ghubade, N., Faloutsos, C., "Detecting discriminative functional MRI activation patterns using space filling curves," in Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), Cancun, Mexico, pp. 963-967, 2003.

Other specific products

Data or databases: We have collected medical image data (mammography, MRI, fMRI, MRS) and associated clinical assessment. Some of these data have been populated to a database accessible through the web page of the Data Engineering Laboratory (denlab.temple.edu) or they are directly available at denlab.temple.edu/data_repository. In addition to datasets, we provide code that we have developed as part of this project. Datasets and software is available through the web site of the Data Engineering Laboratory (denlab.temple.edu).

Teaching aids: Course material for the new courses we have developed (Knowledge Discovery and Data Mining, Advanced Topics in Databases - Multimedia Databases) and the courses we have refined (Principles of Data Management) is already shared through the web. Material is already available under specific courses on the web page of the PI (knight.cis.temple.edu/~vasilis).

Software: We provide source code that we have developed as part of this project. The programs include dynamic recursive partitioning, a spatial data simulator and Matlab routines for generation of time series data. Sharing Information: The software is available through the DEnLab web site. The source code for dynamic recursive partitioning is available through http://denlab.temple.edu/data_repository/. Additional source code for a spatial data simulator and routines for generation of time series data is available through http://denlab.temple.edu/tool&data.htm.

Project Impact

This work is expected to facilitate the process of medical decision making by providing tools for automatic extraction of the most discriminative features of regions of interest in medical images of various modalities, the efficient retrieval of similar regions in large collections of such images and the elucidation of associations and patterns. Solving these problems will enable researchers to integrate, manipulate and analyze large volumes of image data conducting large-scale epidemiological trials of many afflictions. Analysis of image-based clinical trials using these tools is expected to facilitate advances in both diagnosis, and treatment and provide new insight into the relation of anatomy and function. It is expected that this work will be widely used within the medical imaging community. The project will also provide an excellent resource for graduate and undergraduate work in data mining, data compression, multimedia databases, and other areas. As part of this work we have developed two graduate courses: Data Mining and Knowledge Discovery, Medical Image Data Mining.

 

Goals, Objectives and Targeted Activities

The main emphasis of the research activities is on the efficient management of 3-D medical image data and the extraction of patterns between changes in structure, changes in function and other variables. Tools that are helpful to achieve these goals include similarity retrieval, classification, associations mining, and indexing. Although some work has been done in similarity retrieval and classification for general types of images, progress in medical images is slow because of various reasons including the unavailability of large repositories of medical image data, the difficulty of handling medical images since regions that are of interest occupy a small portion of the image and the need for additional robustness in the techniques developed due to the application of these tools in critical domains. Our objectives are to (a) provide efficient methods for the quantitative characterization of regions of interest in medical images employing data and dimensionality reduction techniques to select the most discriminative features, (b) to develop effective classification techniques as well as indexing tools for facilitating the retrieval of similar regions of interest from large medical image databases, (c) to develop data mining tools for discovering patterns between the structure of certain organs and their functionality. The objectives of the education program are: (a) to create a unified multidisciplinary curriculum for data mining and knowledge discovery with a focus on the discovery of medical knowledge through the analysis of medical images and associated clinical data, (b) to create an advanced research group dedicated to data mining in medical image databases.

 

Current and Future Activities

Research:

1. Development of efficient and effective quantitative techniques for characterization of spatial regions of interest (ROIs) of various types: homogeneous and non-homogeneous.

2. Investigation of ROI classification methods based on distributional distances and maximum likelihood methods.

3. Evaluation of the developed techniques on initial real data sets collected from medical collaborators.

4. Investigation of a feature selection technique based on a Markov Chain Monte Carlo (MCMC) framework.

5. Study of the relationship between performance of data compression algorithms, data mining, and fractal properties of data.

Teaching:

Further development of a graduate course on Data Mining and Knowledge Discovery and a graduate seminar on Advanced Topics in Databases - Multimedia Databases. Continue the development of of a graduate course on medical image data mining.

Laboratory:

Establishment of a laboratory that is devoted to research and teaching in data mining and data engineering.

 

Area Background

The project being interdisciplinary in nature requires familiarity with image databases, data mining, machine learning, pattern recognition, information retrieval, data compression, data visualization, medical image analysis as well as specific knowledge from medical/biological domains.  

Area References

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.

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.

F. Korn, N. Sidiropoulos, C. Faloutsos, E. Siegel, and Z. Protopapas, "Fast and effective retrieval of medical tumor shapes," IEEE Transactions on Knowledge and Data Engineering, vol. 10, pp. 889-904, 1998.

C. E. Brodley, A. C. Kak, J. G. Dy, C. R. Shyu, A. Aisen, and L. Broderick, "Content-based retrieval from medical image databases: A synergy of human interaction, machine learning and computer vision," in Proceedings of the The Sixteenth National Conference on Artificial Intelligence, pp. 760-767, Orlando, FL, 1999.

K. J. Cios, (Ed.), Medical Data Mining and Knowledge Discovery, Springer-Verlag, 2001.

C. Shyu, C. E. Brodley, A. Kak, A. Kosaka, A. Aisen, and L. Broderick, "ASSERT: A physician-in-the-loop content-based image retrieval system for HRCT image databases," Computer Vision and Image Understanding, vol. 75, pp. 111-132, 1999.

Y. Liu and F. Dellaert, "A Classification Based Euclidean Similarity Metric for 3D Image Retrieval," in Proceedings of the Computer Vision and Pattern Recognition (CVPR'98), 1998.

V. Megalooikonomou, J. Ford, L. Shen, F. Makedon, and A. Saykin, "Data Mining in Brain Imaging", Statistical Methods in Medical Research, vol. 9, pp. 359-394, 2000.

Y. Liu and F. Dellaert, "Classification Driven Medical Image Retrieval," in Proceedings of the DARPA Image Understanding Workshop (IUW'98), 1998.

H. Tagare, C. Jaffe, and J. Duncan, "Medical Image Databases: A Content-based Retrieval Approach," Journal of the American Medical Informatics Association, vol. 4, pp. 184-198, 1997.

A. Berman and L. G. Shapiro, "A Flexible Image Database System for Content-Based Retrieval," Computer Vision and Image Understanding, Vol. 75, Nos. 1-2, pp. 175-195, 1999.

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.

M. Flickner, H. Sawhnew, W. Niblack, J. Ashley, Q. Huang, B. Dom, M. Gorkani, J. Hafner, D. Lee, D. Petkovic, D. Steel, P. Yanker,"Query by image and video content: the QBIC system," Computer, pp 23-32, Vol 3, number 9, 1995.

 

Project Websites

http://knight.cis.temple.edu/~vasilis/research/career.html

This is the main website for our project.

http://denlab.temple.edu

This is the website for the Data Engineering Laboratory (DEnLab).