Middlesex Medical Image Repository Dr. Yu Qian y.qian@mdx.ac.uk
Content Introduction of MIRAGE project Introduction of Content-based Image Retrieval(CBIR) Proposed framework for MIRAGE CBIR for 2D medical images CBIR for 3D medical images Image labelling what we have done and future work
PART I Introduction of MIRAGE Project
MIRAGE (Middlesex medical Image Repository with a CBIR ArchivinG Environment) Aim: To develop a repository of medical images benefiting MSc and research students in the immediate term and serve a wider community in the long term in providing a rich supply of medical images for data mining, to complement MU current online e-learning system, OASIS+. Collaboration between three parties at MU, including EIS,CLQE and CIE. http://image.mdx.ac.uk/ JISC Innovation in the use of ICT for education and research. http://www.jisc.ac.uk/
PART II Content-Based Image Retrieval (CBIR)
Content-Based Image Retrieval (CBIR) CBIR can index an image using visual contents that an image is carrying, such as colour, texture, shape and location. Query by Example (QBE) Query by Feature (QBF) Query by Sketch(QBS) For example:
Colour-Based Retrieval
Texture-Based Retrieval
Shape-Based Retrieval
Query by Feature
Query by Sketch
Framework of Content-Based Image Retrieval
Compared with Text-Based Image Retrieval(TBIR) Advantage Disadvantage TBIR Semantic information Heavy labour and time consumption Visual information scarcity Subjectivity Language problem CBIR Less time and labour intensity Objective retrieval results Semantic gap
CBIR for Medical Images The need for CBIR For clinical diagnoses For teaching and research CBIRS ASSERT(HRCT Lung) FICBDS(PET) CBIRS(Spine X-Ray) BASS (Breast Cancer)
PART III Framework for MIRAGE
Proposed Framework for MRIAGE
1) CBIR for 2D Medical Image ----GIFT
GIFT(GNU Image Finding Tool) GIFT is open framework for content-based image retrieval and is developed by University of Geneva. Query by example and multiple query Relevance Feedback Distributed architecture (Client - Server) MRML---C-S communication protocol Demo:
GIFT Framework
2) CBIR for 3D Medical Images
Proposed Framework for 3D Image Retrieval
3D Texture Feature Extraction 3D Grey Level Co-occurrence Matrices (3D GLCM) 3D Wavelet Transform (3D WT) 3D Gabor Transform (3D GT) 3D Local Binary Pattern (3D LBP)
Similarity Measurement Histogram Intersection(3D LBP) Normalized Euclidean distance (3D GLCM,3D WT,3D GT)
Experiment Results
Processing and Query time Methods Processing time Query time 3D GLCM 10.65s 0.83s 3D WT 2.03s 0.11s 3D GT 14.3m 0.31s 3D LBP 0.78s 0.29s
3) Image Labelling
Image labelling
PART IV What We Have Done and Future Work
What We Have Done GIFT framework Uploaded and processed 73000 images 3D image retrieval Created 3D feature database using four 3D feature descriptors (One paper had been published in IADIS e-health 2010). Link MIRAGE to OASIS+
Future Work Continue working on image labelling Plug 3D image retrieval into GIFT framework System evaluation Final report
Question? Thanks