Support Feature Machine for Classification of Abnormal Brain Activity

Slides:



Advertisements
Similar presentations
ECG Signal processing (2)
Advertisements

Image classification Given the bag-of-features representations of images from different classes, how do we learn a model for distinguishing them?
W. Art Chaovalitwongse Industrial & Systems Engineering
 Introduction and motivation  Comparitive investigation: Predictive performance of measures of synchronization  Statistical validation of seizure predictions:
Northeast Regional Epilepsy Group Christos Lambrakis M.D. 1.
Discriminative and generative methods for bags of features
Standard electrode arrays for recording EEG are placed on the surface of the brain. Detection of High Frequency Oscillations Using Support Vector Machines:
Seizure prediction by non- linear time series analysis of brain electrical activity Ilana Podlipsky.
Data Mining – Intro.
Advanced Database Applications Database Indexing and Data Mining CS591-G1 -- Fall 2001 George Kollios Boston University.
Optimization and Data Mining in Epilepsy Research W. Art Chaovalitwongse Assistant Professor Industrial and Systems Engineering Rutgers University.
A Machine Learning Framework for Space Medicine Predictive Diagnostics with Physiological Signals Ning Wang, Michael R. Lyu Dept. of Computer Science &
This week: overview on pattern recognition (related to machine learning)
Data Processing Machine Learning Algorithm The data is processed by machine algorithms based on hidden Markov models and deep learning. They are then utilized.
Optimization-Based Data Mining Approaches in Neuroscience Research Panos M. Pardalos University of Florida.
Kernel Methods A B M Shawkat Ali 1 2 Data Mining ¤ DM or KDD (Knowledge Discovery in Databases) Extracting previously unknown, valid, and actionable.
Data Mining Knowledge on rough set theory SUSHIL KUMAR SAHU.
Current work at UCL & KCL. Project aim: find the network of regions associated with pleasant and unpleasant stimuli and use this information to classify.
Exploration of Instantaneous Amplitude and Frequency Features for Epileptic Seizure Prediction Ning Wang and Michael R. Lyu Dept. of Computer Science and.
Classifiers Given a feature representation for images, how do we learn a model for distinguishing features from different classes? Zebra Non-zebra Decision.
Acknowledgements Contact Information Anthony Wong, MTech 1, Senthil K. Nachimuthu, MD 1, Peter J. Haug, MD 1,2 Patterns and Rules  Vital signs medoids.
Copyright © 2012, SAS Institute Inc. All rights reserved. ANALYTICS IN BIG DATA ERA ANALYTICS TECHNOLOGY AND ARCHITECTURE TO MANAGE VELOCITY AND VARIETY,
CSSE463: Image Recognition Day 11 Lab 4 (shape) tomorrow: feel free to start in advance Lab 4 (shape) tomorrow: feel free to start in advance Test Monday.
Classifying Event-Related Desynchronization in EEG, ECoG, and MEG Signals Kim Sang-Hyuk.
Introduction: Brain Dynamics Jaeseung Jeong, Ph.D Department of Bio and Brain Engineering, KAIST.
CSSE463: Image Recognition Day 11 Due: Due: Written assignment 1 tomorrow, 4:00 pm Written assignment 1 tomorrow, 4:00 pm Start thinking about term project.
WHAT IS DATA MINING?  The process of automatically extracting useful information from large amounts of data.  Uses traditional data analysis techniques.
Next, this study employed SVM to classify the emotion label for each EEG segment. The basic idea is to project input data onto a higher dimensional feature.
A recurring neurological disorder characterized by random firing of nerve cells in the brain which cause a temporary shutdown of normal brain function.
Feature learning for multivariate time series classification Mustafa Gokce Baydogan * George Runger * Eugene Tuv † * Arizona State University † Intel Corporation.
Automated Evaluation of Physical Therapy Exercises by Multi-Template Dynamic Time Warping of Wearable Sensor Signals Aras Yurtman and Billur Barshan.
P.Demestichas (1), S. Vassaki(2,3), A.Georgakopoulos(2,3)
Data Mining, Machine Learning, Data Analysis, etc. scikit-learn
Data Mining – Intro.
Transaction Processing System (TPS)
Machine Learning – Classification David Fenyő
Scalable EEG interpretation using Deep Learning and Schema Descriptors
WP2 INERTIA Distributed Multi-Agent Based Framework
Guillaume-Alexandre Bilodeau
[Ran Manor and Amir B.Geva] Yehu Sapir Outlines Review
A Personal Tour of Machine Learning and Its Applications
An Artificial Intelligence Approach to Precision Oncology
School of Computer Science & Engineering
Ways of investigating the brain
CSSE463: Image Recognition Day 11
Epileptic Seizure Prediction
Damiano Bolzoni, Sandro Etalle, Pieter H. Hartel
Transaction Processing System (TPS)
Capturing the Secret Dances in the Brain
CSSE463: Image Recognition Day 11
Automated Evaluation of Physical Therapy Exercises by Multi-Template Dynamic Time Warping of Wearable Sensor Signals Aras Yurtman and Billur Barshan.
Data Warehousing and Data Mining
ECE539 final project Instructor: Yu Hen Hu Fall 2005
Big Data Resources for EEGs: Enabling Deep Learning Research
To learn more, visit The Neural Engineering Data Consortium Mission: To focus the research community on a progression of research questions.
Transaction Processing System (TPS)
Data Mining: Concepts and Techniques
Data Mining: Concepts and Techniques
Northeast Regional Epilepsy Group Christos Lambrakis M.D.
Data Mining, Machine Learning, Data Analysis, etc. scikit-learn
Data Mining, Machine Learning, Data Analysis, etc. scikit-learn
Data Mining: Concepts and Techniques
Model Enhanced Classification of Serious Adverse Events
CSSE463: Image Recognition Day 11
Topic 5: Cluster Analysis
CSSE463: Image Recognition Day 11
Assignment 1: Classification by K Nearest Neighbors (KNN) technique
Exploiting the Power of Group Differences to Solve Data Analysis Problems Outlier & Intrusion Detection Guozhu Dong, PhD, Professor CSE
Modeling IDS using hybrid intelligent systems
ECE – Pattern Recognition Lecture 8 – Performance Evaluation
Presentation transcript:

Support Feature Machine for Classification of Abnormal Brain Activity W. Art Chaovalitwongse Rutgers University *Joint work with Y.J. Fan (Rutgers) and R.C. Sachdeo (Jersey Shore University Hospital) Poster #16, Mon Aug 13 This work is supported in part by research grants from NSF CAREER Grant CCF 05-46574 and Rutgers Research Council Grant-02028.

Agenda Support Feature Machine Empirical Results Research Objectives Research Background Epilepsy and seizures Electroencephalogram (EEG) time series Signal processing Support Feature Machine Empirical Results Concluding Remarks

Objectives: Develop a new pattern recognition and classification framework for multi-dimensional time series data – as a decision making support. Application in Epilepsy Identification of Seizure Pre-Cursor: Classification of seizure susceptibility periods Quick Screening Tool: Classification of epilepsy and non-epilepsy patients Seizure Prediction: Anomaly (Seizure Pre-Cursor) detection Generalize the framework to other applications that have to deal with multi-dimensional time series data.

How many people having epilepsy? “The Swamp”: Seating capacity ~90,000

Epilepsy and Seizures Nearly 3 million people in the U.S. (1% of population) have epilepsy. Anyone, at any age, can develop it. Epilepsy is defined as recurring seizures – sudden, brief changes in the way the brain works. Seizures cause temporary disturbances of brain functions such as motor control, responsiveness and recall which typically last from seconds to a few minutes. Seizures usually occur spontaneously, in the absence of external triggers. Seizures may be followed by a post-ictal period of confusion or impaired sensorial that can persist for several hours.

Intracranial EEG Acquisition

Electroencephalogram (EEG) …is a traditional tool for evaluating the physiological state of the brain. …offers good spatial and excellent temporal resolution to characterize rapidly changing electrical activity of brain activation …captures voltage potentials produced by brain cells while communicating. In an EEG, electrodes are implanted in deep brain or placed on the scalp over multiple areas of the brain to detect and record patterns of electrical activity and check for abnormalities.

10-second EEGs: Seizure Evolution Normal Pre-Seizure Seizure Onset Post-Seizure

Open Problems Seizure pre-cursors exist? Seizure is a state transition process? Can we discriminate normal EEGs from pre-seizure EEGs (seizure susceptibility period)?

Data Transformation Using Chaos Theory Measure the brain dynamics from time series: Stock Market Currency Exchanges (e.g., Swedish Kroner) Apply dynamical measures (based on chaos theory) to non-overlapping EEG epochs of 10.24 seconds = 2048 points. Maximum Short-Term Lyapunov Exponent measure the average uncertainty along the local eigenvectors and phase differences of an attractor in the phase space measure the stability/chaoticity of EEG signals Iasemidis, Shiau, Chaovalitwongse, Sackellares & Pardalos, IEEE Transactions on Biomed (2003)

Measure of Chaos

Classification of Physiological States

Support Vector Machine VS Support Feature Machine

Nearest Neighbor for Time Series Normal Pre-Seizure A d1: Average distance to blues. d2: Average distance to reds. d2 < d1, so new point is classified as red.

Similarity Measures Dynamic Time Warping (DTW) Distance Euclidean Distance T-Statistical Distance STLmax 1, 2, 3, ….. , 30 STLmax 1, 2, 3, ….. , 30 Electrode 1 2 3 . 26 D(X, Y) Y X

Support Feature Machine Given an unknown epoch of EEG signals A, we calculate statistical distances between the EEG epoch and the groups of Normal and Pre-Seizure EEGs in our data baseline. Euclidean distance T-statistical distance Dynamic Time Warping EEG sample A will be classified in the group of patient’s state (normal or pre-seizure) that yields the minimum statistical distance. Multiple Electrodes = Multiple Decisions Averaging Majority Voting: selects action with maximum number of votes Can we select/optimize the selection of a subset of electrodes that maximizes number of correctly classified samples. Chaovalitwongse et al., Submitted to Operations Research

Decision Rule: Basic Ideas Two different average distances for each sample at each electrode are calculated: Intra-Class: Average distances from each sample to all other samples in the same class at Electrode j Inter-Class: Average distances from each sample to all other samples in different class at Electrode j If for Sample i at Electrode j (Averaging VS Voting) Average distance to the same class < Average distance to different class Then Sample i is correctly classified.

Optimization Model I: Averaging Intra-Class Inter-Class Chaovalitwongse et al., Submitted to Operations Research

Model I: Averaging Formulation Chaovalitwongse et al., Submitted to KDD, 2007 and Operations Research

Optimization Model II: Voting Precision matrix, A contains elements of Chaovalitwongse et al., Submitted to Operations Research

Decision Rule: Basic Ideas Two different average distances for each sample at each electrode are calculated: Intra-Class: Average distances from each sample to all other samples in the same class at Electrode j Inter-Class: Average distances from each sample to all other samples in different class at Electrode j If for Sample i at Electrode j (Averaging VS Voting) Average distance to the same class < Average distance to different class Then Sample i at Electrode j is correctly classified. Chaovalitwongse et al., Submitted to Operations Research

Model II: Voting Formulation Chaovalitwongse et al., Submitted to KDD, 2007 and Operations Research

Data Selection and Sampling EEG Dataset Characteristics Patient ID Seizure types Duration of EEG(days) # of seizures 1 CP, SC 3.55 7 2 CP, GTC, SC 10.93 3 CP 8.85 22 4 ,SC 5.93 19 5 13.13 17 6 11.95 3.11 9 8 6.09 23 11.53 20 10 9.65 12 Total   84.71 153 CP: Complex Partial; SC subclinical; GTC: Generalized Tonic/Clonic Randomly and uniformly sample 3 EEG epochs per seizure from each of normal and pre-seizure states. For example, Patient 1 has 7 seizures. There are 21 normal and 21 pre-seizure EEG epochs sampled. Seizure Duration of EEG 30 minutes 8 hours Pre-seizure Normal

Sensitivity and Specificity Sensitivity measures the fraction of positive cases that are classified as positive. Specificity measures the fraction of negative cases classified as negative. Sensitivity = TP/(TP+FN) Specificity = TN/(TN+FP) Sensitivity can be considered as a detection (prediction or classification) rate that one wants to maximize. False positive rate can be considered as 1-Specificity which one wants to minimize.

5-Fold Cross Validation Result 81.29% 72.86% Optimize the number of neighbors Chaovalitwongse et al., IEEE Trans Systems, Man, and Cybernetics: Part A, 2008 Chaovalitwongse et al., Submitted to Operations Research

DTW Euclidean T-Statistics

DTW Euclidean T-Statistics

Automated Seizure Prediction Paradigm Multichannel   Com Feature Extraction/ Cluster Analysis Data Acquisition   Interface Technology Pattern Recognition VNS Initiate a variety of therapies (e.g., electrical stimulation, drug injection) User Drug

Concluding Remarks Overview of a Real Life Medical Problem in Spatio-Temporal Data Mining Applications of Data Mining and Optimization Techniques Potential Applications in Medical Diagnosis Automated seizure warning system Monitoring devices for clinical use in epilepsy monitoring units (EMUs) and intensive care units (ICUs) Other monitoring procedures in trauma and operation rooms Improvement of the Nearest Neighbor Classification in Time Series Classification - New Classification Framework

Reference W. Chaovalitwongse, Y.J. Fan, R.C. Sachdeo. Novel Optimization Models for Multidimensional Time Series Classification: Application to the Identification of Abnormal Brain Activity. Submitted to Operations Research. Y.J. Fan, W. Chaovalitwongse, L.D. Iasemidis, R.C. Sachdeo. Multi-Dimensional Time Series Classification for Identification of Epilepsy Patients. Submitted to KDD 2007. W. Chaovalitwongse, Y.J. Fan, and R.C. Sachdeo. On the K-Nearest Dynamic Time Warping Neighbor for Abnormal Brain Activity Classification. To appear in IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans, 2008. W. Chaovalitwongse and P.M. Pardalos. On the Time Series Support Vector Machine using Dynamic Time Warping Kernel for Brain Activity Classification. To appear in Cybernetics and Systems Analysis, 2007 W. Chaovalitwongse, P.M. Pardalos, and O.A. Prokopyev. Electroencephalogram (EEG) Time Series Classification: Applications in Epilepsy. Annals of Operations Research, 148: 227-250, 2006. W. Chaovalitwongse, L.D. Iasemidis, P.M. Pardalos, P.R. Carney, D.-S. Shiau, and J.C. Sackellares. A Robust Method for Studying the Dynamics of the Intracranial EEG: Application to Epilepsy. Epilepsy Research, 64, 93-133, 2005. W. Chaovalitwongse , P.M. Pardalos, L.D. Iasemidis, D.-S. Shiau, and J.C. Sackellares. Dynamical Approaches and Multi-Quadratic Integer Programming for Seizure Prediction. Optimization Methods and Software, 20 (2-3): 383-394, 2005 .

Acknowledgements Comprehensive Epilepsy Center, St. Peter’s University Hospital Rajesh C. Sachdeo, MD Rutgers Ph.D. Student Ya-Ju Fan, MS Industrial and Systems Engineering, University of Florida Panos M. Pardalos, PhD Bioengineering, Arizona State University Leonidas D. Iasemidis, PhD