automating scoliosis analysis by amar sahai thomas jefferson high school for science and technology...

13
Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Upload: marcia-simmons

Post on 21-Jan-2016

218 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Automating Scoliosis Analysis

By Amar Sahai

Thomas Jefferson High Schoolfor Science and Technology

2008-2009

Page 2: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Purpose

Scoliosis = curvature of the spine Current analyses are either expensive or

manual and time-consuming Try to automate this Saves time, effort and money

Page 3: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Scope of Study

Automating scoliosis detection and angle of curvature

Help pinpoint places on spine to apply pressure to most effectively deal with curve

Simpler and cheaper than other solutions

Page 4: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Similar Studies

Detection and Measurement of Hilar Region in Chest Radiograph – Australia, 2003

Automatic Computer Recognition and Analysis of Dental X-ray Film – New York, 1970

Page 5: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Procedures and Methodology – Phase 1

Coded in Java and C Requires x-ray images as input Converts inputted images to .pgm image

format Uses edge detection to get a clean outline of

the spine Darkens image to reduce noise

Page 6: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Procedures and Methodology – Phase 2

Convert .pgm file output from Phase 1 to a .gif

Input .gif into new Java program that allows user input

Program displays .gif as background Program accepts mouse as input device Left click draws point at clicked location Program draws a line between every other

point and calculates the acute angle from the vertical

Page 7: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Procedures and Methodology – Phase 2 (cont.)

Right-click toggles auto-detect mode Left click in this mode draws a point at the

“edgiest” spot within a 10 pixel radius Program draws a line between every other

point and calculates the acute angle from the vertical

Points and lines are color coded Points: red in normal, green in auto-detect Lines: blue in normal, green in auto-detect

Page 8: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Edge Detection

Previous algorithm was fairly primitive Checked for brightest point Current new method – Sobel Calculates horizontal & vertical gradients

Page 9: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Edge Detection (cont.)

Gradients are determined by “masks” Horizontal

-1 0 1-2 0 2-1 0 1

Vertical 1 2 1 0 0 0-1 -2 -1

Each set of pixels is multiplied by each mask Any pixel with a high enough G value is

recorded as an edge

Page 10: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Roadblocks

File conversion must be done in another program

Must erase ribcage in order to detect points to apply pressure

Need to add buttons to add more features without making

Page 11: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Current Results

My edge detection quality is improving but still needs work for actual use

Must eradicate extraneous parts of image (ribcage, pelvis)

User interface phase works in manual and semi-automatic modes

Page 12: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Screenshots

Page 13: Automating Scoliosis Analysis By Amar Sahai Thomas Jefferson High School for Science and Technology 2008-2009

Screenshots (cont.)