Computational Image Classification UMBC Department of Computer Science eBiquity Research Group February 19, 2010.

Slides:



Advertisements
Similar presentations
Computer Aided Diagnosis: Feature Extraction
Advertisements

電腦視覺 Computer and Robot Vision I
Carolina Galleguillos, Brian McFee, Serge Belongie, Gert Lanckriet Computer Science and Engineering Department Electrical and Computer Engineering Department.
July 27, 2002 Image Processing for K.R. Precision1 Image Processing Training Lecture 1 by Suthep Madarasmi, Ph.D. Assistant Professor Department of Computer.
電腦視覺 Computer and Robot Vision I Chapter2: Binary Machine Vision: Thresholding and Segmentation Instructor: Shih-Shinh Huang 1.
The Global Digital Elevation Model (GTOPO30) of Great Basin Location: latitude 38  15’ to 42  N, longitude 118  30’ to 115  30’ W Grid size: 925 m.
Robust Foreground Detection in Video Using Pixel Layers Kedar A. Patwardhan, Guillermoo Sapire, and Vassilios Morellas IEEE TRANSACTION ON PATTERN ANAYLSIS.
Image Indexing and Retrieval using Moment Invariants Imran Ahmad School of Computer Science University of Windsor – Canada.
WRSTA, 13 August, 2006 Rough Sets in Hybrid Intelligent Systems For Breast Cancer Detection By Aboul Ella Hassanien Cairo University, Faculty of Computer.
Medical Imaging Mohammad Dawood Department of Computer Science University of Münster Germany.
IEEE TCSVT 2011 Wonjun Kim Chanho Jung Changick Kim
EE 7730 Image Segmentation.
A Computer Aided Detection System For Digital Mammograms Based on Radial Basis Functions and Feature Extraction Techniques By Mohammed Jirari Shanghai,
Texture-Based Image Retrieval for Computerized Tomography Databases Winnie Tsang, Andrew Corboy, Ken Lee, Daniela Raicu and Jacob Furst.
1 Texture Texture is a description of the spatial arrangement of color or intensities in an image or a selected region of an image. Structural approach:
Three-dimensional co-occurrence matrices & Gabor filters: Current progress Gray-level co-occurrence matrices Carl Philips Gabor filters Daniel Li Supervisor:
Texture Texture is a description of the spatial arrangement of color or intensities in an image or a selected region of an image. Structural approach:
A Computer-Aided Diagnosis System For Digital Mammograms Based on Radial Basis Functions and Feature Extraction Techniques Dissertation written by Mohammed.
Mimicking human texture classification Eva M. van Rikxoort Egon L. van den Broek Theo E. Schouten.
A Wrapper-Based Approach to Image Segmentation and Classification Michael E. Farmer, Member, IEEE, and Anil K. Jain, Fellow, IEEE.
Texture Classification Based on Co-occurrence Matrices Presentation III Pattern Recognition Mohammed Jirari Spring 2003.
Texture Readings: Ch 7: all of it plus Carson paper
Feature Screening Concept: A greedy feature selection method. Rank features and discard those whose ranking criterions are below the threshold. Problem:
Texture-based Deformable Snake Segmentation of the Liver Aaron Mintz Daniela Stan Raicu, PhD Jacob Furst, PhD.
Feature extraction Feature extraction involves finding features of the segmented image. Usually performed on a binary image produced from.
Brief overview of ideas In this introductory lecture I will show short explanations of basic image processing methods In next lectures we will go into.
INTRODUCTION Problem: Damage condition of residential areas are more concerned than that of natural areas in post-hurricane damage assessment. Recognition.
Entropy and some applications in image processing Neucimar J. Leite Institute of Computing
Presented by: Kamakhaya Argulewar Guided by: Prof. Shweta V. Jain
Polyp-detection in Colonoscopy Stefan Ameling 2008
Methods in Medical Image Analysis Statistics of Pattern Recognition: Classification and Clustering Some content provided by Milos Hauskrecht, University.
CS 376b Introduction to Computer Vision 04 / 02 / 2008 Instructor: Michael Eckmann.
Video Summarization of Key Events Stage I - The Critical View Michael A. Grasso, MD, PhD University of Maryland School of Medicine UMBC Computer Science.
What’s Making That Sound ?
Machine Vision for Robots
Landsat classification © Team. SSIP © Team Delia Mitrea – Technical University of Cluj-Napoca, Romania Sándor Szolyka – Budapest Tech, Hungary Imre.
BACKGROUND LEARNING AND LETTER DETECTION USING TEXTURE WITH PRINCIPAL COMPONENT ANALYSIS (PCA) CIS 601 PROJECT SUMIT BASU FALL 2004.
Ajay Kumar, Member, IEEE, and David Zhang, Senior Member, IEEE.
B. Krishna Mohan and Shamsuddin Ladha
Texture. Texture is an innate property of all surfaces (clouds, trees, bricks, hair etc…). It refers to visual patterns of homogeneity and does not result.
CSE 185 Introduction to Computer Vision Pattern Recognition 2.
Microarrays.
1 Multiple Classifier Based on Fuzzy C-Means for a Flower Image Retrieval Keita Fukuda, Tetsuya Takiguchi, Yasuo Ariki Graduate School of Engineering,
Preprocessing Techniques for Image Analysis Applications Hong Zhang.
Course 9 Texture. Definition: Texture is repeating patterns of local variations in image intensity, which is too fine to be distinguished. Texture evokes.
Color wavelet covariance(CWC) Texture feature
Computer Graphics and Image Processing (CIS-601).
Prostate Cancer CAD Michael Feldman, MD, PhD Assistant Professor Pathology University Pennsylvania.
Region-Based Saliency Detection and Its Application in Object Recognition IEEE TRANSACTIONS ON CIRCUITS AND SYSTEM FOR VIDEO TECHNOLOGY, VOL. 24 NO. 5,
Levels of Image Data Representation 4.2. Traditional Image Data Structures 4.3. Hierarchical Data Structures Chapter 4 – Data structures for.
Ilya Gurvich 1 An Undergraduate Project under the supervision of Dr. Tammy Avraham Conducted at the ISL Lab, CS, Technion.
Chap 7 Image Segmentation. Edge-Based Segmentation The edge information is used to determine boundaries of objects Pixel-based direct classification methods.
Image Segmentation Shengnan Wang
Feature Extraction and Classification of Mammographic Masses
Slides from Dr. Shahera Hossain
A Statistical Approach to Texture Classification Nicholas Chan Heather Dunlop Project Dec. 14, 2005.
Robotics Chapter 6 – Machine Vision Dr. Amit Goradia.
1 Review and Summary We have covered a LOT of material, spending more time and more detail on 2D image segmentation and analysis, but hopefully giving.
Color Image Segmentation Mentor : Dr. Rajeev Srivastava Students: Achit Kumar Ojha Aseem Kumar Akshay Tyagi.
TUMOR BURDEN ANALYSIS ON CT BY AUTOMATED LIVER AND TUMOR SEGMENTATION RAMSHEEJA.RR Roll : No 19 Guide SREERAJ.R ( Head Of Department, CSE)
Finding Clusters within a Class to Improve Classification Accuracy Literature Survey Yong Jae Lee 3/6/08.
A. M. R. R. Bandara & L. Ranathunga
IMAGE PROCESSING RECOGNITION AND CLASSIFICATION
Palm Oil Plantation Area Clusterization for Monitoring
Texture Classification of Normal Tissues in Computed Tomography
Brain Hemorrhage Detection and Classification Steps
A New Approach to Track Multiple Vehicles With the Combination of Robust Detection and Two Classifiers Weidong Min , Mengdan Fan, Xiaoguang Guo, and Qing.
Texture Classification of Normal Tissues in Computed Tomography
Computer and Robot Vision I
Multiple Organ detection in CT Volumes - Week 3
Presentation transcript:

Computational Image Classification UMBC Department of Computer Science eBiquity Research Group February 19, 2010

Overview Introductions Image Classification Initial Results Future Efforts

Introductions Faculty Yelena Yesha, PhD Michael Grasso, MD, PhD John Dorband, PhD Tim Finin, PhD Milt Halem, PhD Anupam Joshi, PhD Graduate students Ronil Mokashi Darshana Dalvi

Computational Image Classification Categorize a raster image into a finite set of classes. Convert raster data into feature vectors. Support vector machine image classifier. Metadata to map specific classes to biological characteristics.

Image Classification Examples Computer assisted diagnosis of prostate and breast cancer biopsies. Segmentation of hysteroscopy video. Echocardiogram analysis. Skin cancer detection. Biomedical Imaging: from Nano to Macro, 2007;: IEEE TITB, 2008 May;12(3): Proceedings 27th IEEE EMBS, 2005;: Conf Proc IEEE Eng Med Biol Soc, 2006;1:4775-8

Related Efforts: Video Segmentation Laparoscopic cholecystectomy videos. 378 representative images from 5 videos. Analyzed 49 separate image features.

Related Efforts: Video Segmentation Image classification. Distance metric to identify best features. Support vector machine image classifier. Accuracy of 91%. Video Segments Frames

Future directions. Real-time analysis to assess patient safety. Time and motion analysis of surgical instruments. Classification of pathology. Hiatal hernias. Capsule endoscopy. Related Efforts: Video Segmentation

Related Efforts: Cancer Screening Skin caner screening. Handheld iPhone image classifier. Tool for primary care physicians. Identify lesions in need of dermatology referral. NIH and UMB proposals pending.

Image Classification Approach Feature Extraction Feature Models Model Features Image Classifier Images Organized by Class Unknown Image Feature Extraction

Image Features Spectral features (color/tone). Histogram (3D, color, binary, gray). Distribution, size, width, mean, stdev. Do not vary with translation and rotation. Textural features (spatial distribution). Gray-level co-occurrence matrix (GLCM). Energy, entropy, contrast, correlation. Independent of color distribution. IEEE Transaction on Systems, Man, and Cybernetics Nov; 3(6):

Advanced Features - Context Image segmentation. Regions of interest (contextual features). Threshold algorithms - Maximum Entropy, Otsu Threshold, Watershed, etc. Segmentation features of actin fibers. Density - Area, Mean Gray Level. Distribution - Centroid, Center of Mass. Orientation - Angle, Elliptical Fit (wrt cell). Order - Angle, Elliptical Fit (wrt fibers).

Advanced Features - Clustering K-means clustering of image features. Partitions images into clusters based on the nearest mean, based on a first-order Markov property. Based on the assumption that images with similar clinical features are more likely to be found in the same cluster.

Model Development Distance metrics Manhattan distance, Jeffrey divergence. Classification threshold. Support vector machines Machine learning methods. An N-dimensional hyperplane optimally separates images into categories. This mapping is performed by a set of mathematical functions, known as kernels.

Initial Results - Feature Analysis Initial image classification experiment. Evaluated 15 spectral and textural features. Total of 11 images in 4 groups. Focal adhesions images, actin stained. 1hdry, 1hwet, 24hdry, 24hwet. Analysis. Leave-one-out technique over all 11 images. Manhattan distance. Threshold of 5% (0.3% for histograms).

Initial Results - Feature Analysis Trait+Trait- Feature Feature Promising features. Gray-scale distribution Medium Mode Homogeneity Energy Entropy Inverse difference moment Sensitivity = 76.8% Specificity = 89.6% Accuracy = 86.4%

Initial Results - Segmentation Second image classification experiment. Evaluating 4 new image features. Density, Distribution, Orientation, Order Experimenting with threshold algorithms to optimize image segmentation. Total of 11 images in 4 groups. Focal adhesions images, actin stained. 1hdry, 1hwet, 24hdry, 24hwet.

Initial Results - Segmentation Orientation feature using elliptical fit. Image moment-preserving threshold. Elliptical fit. Cell angle = 132° Average (weighted) actin fiber angle = 129°.

Initial Results - Segmentation

Initial Results - Analysis Platform To automate and optimize image processing algorithms.

Future Efforts Use feature analysis to develop a support vector machine image classifier. Continue image segmentation work. Correlate actin data to other images. Incorporate successful algorithms in the Analysis Platform. Identify ontologies to map specific classes to biological characteristics.