model based radiographic image classification this method implements an object oriented knowledge...

Post on 15-Jan-2016

215 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Model Based Radiographic Image Classification

• This method implements an object oriented knowledge representation to automatically determines the body part class of a radiographic image. The reasoning unit estimates the most probable body part class based on class knowledge model. The model includes both objects and their spatial relationships.

Object Oriented Knowledge Model

The main object

Component objects

Description attributesComponent objectsSemantic relations

Pelvis Radiograph and Model

Key Components of Pelvis Model

Pelvis Class ModelA t t r i b u t e s : D e s c r i p t i o n : S i z e ,s h a p e ,o r i e n ta t io n , g r a y l e v e l

A t t r i b u t e s : C o m p o n e n t:

1 ) t h e h ip m o d e l2 ) t h e i l i a c m o d e3 ) t h e p u b ic m o d e l4 ) t h e f e m u r m o d e

S e m a n t i c G r a p h

P e l v i s M o d e lC l a s s

M e t h o d : C o n tr o l S t r a t e g y ( ) ;

M a t c h S e m a n t i c G r a p h ( ); F u z z y I n fe r e n c e ( ) ;

H i p M o d e l C l a s s

M e t h o d : M a t c h H i p S h a p e ( )

A t t r i b u t e s : D e s c r i p t i o n : S i z e , s h a p e , o r i e n ta t io n C o m p o n e n t

1 ) t h e l e f t i l i a c b o n e2 ) t h e r ig h t i l i a c b o n e

S e m a n t i c G r a p h

I l i a c M o d e lC l a s s

M e t h o d : M a t c h I l ia c S h a p e ( )

A t t r i b u t e s : D e s c r i p t i o n : S i z e , s h a p e ,o r ie n ta t io n g r a y l e v e l

P u b i c M o d e lC l a s s

M e t h o d : M a t c h P u b i c S h a p e ( )

A t t r i b u t e s : D e s c r i p t i o n : S i z e , s h a p e ,o r ie n ta t io n C o m p o n e n t 1 ) t h e l e f t i l i a c b o n e 2 ) th e r ig h t i l i a c b o n e S e m a n t i c G r a p h

F e m u r M o d e l C la s s

M e t h o d : M a t c h F e m u r S h a p e ( )

Experimental Results for Pelvis Model

Green Bones Best Match to Model

Experimental Results for Pelvis Model

Green Bones Best Match to Model

Chest Radiograph and Typical Edge Map

Semantic Network and Chest Model

Chest body

Spine

Left lung Right lung

insideinside

right ofleft of

Medialpart of

The semantic networks

Diagram of All Model Objects

Attributes: Component:

1) the body model2) the spine

model3) the lung model

Semantic GraphMethod:

Schedule()

Chest Model Class

Attributes: Description:Size, Shape,Graylevel

Methods:BinaryImage ()MatchBodyShape ()

Body Model Class

Attributes: Description:Size, Shape,Graylevel

Methods:SpineDetection ()RefineSpineCour ()

Spine Model Class

Attributes: Description:Size, Shape,Graylevel

Methods:GenerateInitCour ()DeformCour ()

Lung Model Class

1 1

2 3 2 3

Results -Initial Model

A

1

2

xy

Lung Location Refined

Spine Detection Based on Model

Body Part Classification Based on Knee Model

Match: Red Bone modelsReject Green model

Body Part Classification Based on Elbow Model

Match: Red Bone modelsReject Green model

Patents and Publications for Model Based Classification

• Gaborski, R., US 5,943,435: Body part recognition in radiographic images.

• Gaborski, R., US6,018,590: Technique for finding the histogram region of interest based on landmark detection for improved tonescale reproduction of digital radiographic images.

• Jang, B. and Gaborski, R., US 5,862,249: Automated method and system for determination of positional orientation of digital radiographic images.

• Gaborski, R., et al, US 5,696,805:Apparatus and method for identifying specific bone regions in digital X-ray images

Patents and Publications for Model Based Classification, continued

• Luo, H., Gaborski, R. and Acharya, R., "Knowledge Representation for Image Content Analysis in Medical Image Databases", SPIE International Symposium on Medical Imaging 2001.

• Luo,H., Gaborski,R. and Acharya, R., "Automatic Segmentation of Lung Regions in Chest Radiographs: A Model Guided Approach", IEEE International Conference on Image Processing (ICIP2000), Vancouver, Canada, 2000.

• Luo, H., Gaborski, R. and Acharya, R., "Robust Snake Model", Computer Vision and Pattern Recognition 2000, CVPR2000, Hilton Head Island, SC., 2000.

Patents and Publications for Model Based Classification, continued

• Sun, Y., Gaborski, R. and Acharya, R., "A Practical Approach for Locating Bone Structures in Radiographic Pelvis Images", 1999 Western New York Image Processing Conference Workshop, Rochester, New York, 1999.

• Luo, H., Acharya, R. and Gaborski,R. , " Fully Automatic Detection of Spine in Chest Radiographs using Fuzzy Logic Approach, " Soft computing in Biomedicine, Rochester, New York, 1999.

• Luo, H., Acharya, R., Gaborski, R., " A New Fully Automatic Approach to Detect the Spine from X-ray Image", IEEE Western New York Image Processing Workshop, Rochester, New York, 1998.

• Luo, H., Acharya, R. and Gaborski,R. , " A Knowledge-based Method for Automatic Segmentation of Lung Regions in Digital Chest Radiographs", 1999 Western New York Image Processing Conference Workshop, Rochester, New York, 1999.

top related