Integrated Multi-modality Probe for Imaging, Profiling and Treating Intra Cavity Growths / Abnormality Sun-Woh Lye and Murukeshan VM School of Mechanical and Aerospace Engineering, NTU, Singapore E-mail:
[email protected] [email protected] 1
Health & Human body
Healthy looking outside. What about Inner health?
2
Background • Cancer of the colon and the rectum are together called colorectal cancer • Such cancer represents one of the most common malignant tumors and has become one of the leading causes of death around the world. • Begins as small pre-cancerous lesions called polyps • Polyps transform into malignant tumors in due course 3
Inside view of the colon (Images from Endoscopy center, NUH, Singapore) Normal colon
Colon with tumor
Colon wall layers : EUS image (Courtesy: Endoscopy center NUH, Singapore) 4
Brief Statistics on Colon Cancer 2000 1800
• One of the most common cancer in the world. In Singapore, it is reported to be the second most common cancer.
Number of cases
1600 1400 1200 1000 800 600 400 200 0 1975
1980
1985
1990
1995
2000
2005
Years
• Incidence rate is increasing very rapidly with about 1000 new cases diagnosed annually. Predicted to overtake lung cancer as the most common malignancy (Singapore Medical Journal, Colorectal Cancer Department, Singapore General Hospital) Good News/Motivation: • Colon cancer is highly curable if detected at a very early stage • Urgent attention for its diagnosis and follow-up treatment is needed. •R Rex, D. K., Johnson, D. A., Lieberman, D. A., Burt, R. W., and Sonnenberg, A. (2000). Colorectal Cancer Prevention 2000: Screening Recommendations of 5 the American College of Gastroenterology. The American Journal of Gastroenterology, 95 (4), pp. 868-877. •(Singapore medical journal, Colorectal cancer department, Singapore general hospital)
Polyps and Cancer Growth
Colon wall consists of 4 layers Abnormal growths in the colon are are called polyps. Polyps can be grouped into two two types Lymph nodes ¾ Protruding ¾ Non-Protruding (Flat) The growth starts from the innermost mucosal layer
Stage 0
* Rubio, C. A., Jaramillo, E., Lindblom, A., and Fogt, F. (2002). Classification of colorectal polyps: Guidelines for an endoscopist. Endoscopy, 34, pp. 226-236.
Mucosa (Epithelium) Submucosa Muscularis Mucosa Serosa Nearby tissues
6
Colon Cancer - Degree of Spreading Stage 1
Stage 0
Stage 0 - Cancer confined to innermost layer of colon Stage 1 - Cancer grown through the colon layers Stage 2 - Cancer spread to the nearby tissues Stage 3 - Cancer spread to lymph nodes Stage 4 - Cancer spread to distant organs
Mucosa (Epithelium)
Mucosa (Epithelium)
Submucosa
Submucosa Muscularis Mucosa
Muscularis Mucosa
Serosa
Serosa Nearby tissues
Nearby tissues
Lymph nodes
Lymph nodes
Stage 2
Stage 3
Mucosa (Epithelium)
Mucosa (Epithelium)
Submucosa
Submucosa
Muscularis Mucosa
Muscularis Mucosa
Serosa
Serosa
Nearby tissues
Nearby tissues
Lymph nodes
Lymph nodes
S tage 4 L ym ph nodes
Lungs
L iv e r C o lo n
7
Current Diagnostic Procedures
Invasive (via Biopsies) Non Invasive ¾ External Computer Tomography MRI Barium X-Ray
Capsules
¾ Internal Endoscopy Virtual Colonoscopy
Fecal Occult Blood Test Chair Side Endoscopes 8
Current Diagnostic Methods DIAGNOSTIC TECHNIQUES
SUBDIVISIONS
LIMITATIONS
Computed tomography (CT)
Needs x-ray examination More reliable for distant metastasis Size limitations of the polyp detection
Magnetic resonance imaging (MRI)
More reliable for distant metastasis Size limitations of the polyp detection
Barium enema
Needs x-ray examination Size limitations of the polyp detection
External imaging techniques
Internal imaging techniques
Endoscopic ultrasound (EUS)
Needs compromise between resolution and penetration depth Size limitations of the polyp detection
Colonoscopy
Electronic detection scheme is introduced in the human body Size limitations of the polyp detection
• Minimum detectable size (protruding polyps) is 5mm • All these diagnostic methods require an additional biopsy procedure for the confirmation of the presence of cancer
9
Findings ¾ Require an additional biopsy procedure for the confirmation of cancer presence ¾ Have missing polyps during capture ¾ Detection is non-confirmatory ¾ Unable to detect polyps at its very early stage (< 5mm) ¾ Image capture tends to be distorted
10
Approach – Towards an Integrated Fibre Optics Probe Diagnostic System for Intra Cavity Inspection Module 9 Locomotion 9 Optical Image Measurements
Endoscope Light rays
Capture
and
Features 9 Remote access probe distal end with accurate location and navigation control 9 9 9
Illumination source
Minimum or no tissue perforation during navigation Vision system for 2D and 3D profiling of intra cavity features capable of large area interrogation Combined imaging and affected tissue removal
11
Results of Preliminary Investigations Locomotion Design Defect Detection and Location Modelling Region of Interest
12
Endoscope Light rays
Illumination source
13
Spring loaded wheels for locomotion, clamping and steering Wheels Universal Joints Bevel gears Bellows Cameras
Body
Motors
14
Pseudo defect of size 3mm
15
Results of Preliminary Investigations Locomotion Design Defect Detection and Location Modelling Region of Interest
16
Optical Image Capture Module ¾ Shape measurement techniques explored Contact : Stylus-based probing involves line by line scanning Can cause tissue perforations or abrasions Require very high scanning resolution Time consuming and tedious Non-contact: Digital Speckle Pattern Interferometry (DSPI)
17
Optical Image Capture Module ¾ Shape measurement techniques explored Contact : Stylus-based probing involves line by line scanning Non-contact: Digital Speckle Pattern Interferometry (DSPI) Non-destructive optical technique Provides full field results with high spatial resolution. Light field is reflected from an optical rough surface under coherent illumination where grainy structures (speckle pattern) are formed. Pattern is derived from the interference of light waves of different intensities being reflected back from the surface of an object. 18
Digital Speckle Imaging System LASER DIODE LASER
DIODE F FH
CCD CAMERA
CCD CAMERA
OBJECT OBJECT
ZL
COMPUTER COMPUTER
FRAME GRABBER FRAME
GRABBER
MONITORDISPLA
MONITOR/ Y DISPLAY
(F= Single mode optical fiber, FH= Fiber holder, ZL = Zoom Lens) 19
Test Object : Phantom Colon Tissue (Simulab)
No defect
Pseudo defect
20
DSPI Technique for Defect Identifcation and Profile Mapping
Obtain images from the DSPI system Employ digital image processing techniques to extract the contour data of the speckle images. Derive relevant contour data for surface reconstruction Image Capture Methods: Correlation Method ¾ Decorrelation Method ¾
21
Optical fiber Laser
1x2 fiber coupler Normal imaging / speckle correlation analysis system
Fluorescence spectrum analysis system
Collection lens (Mitutoyo LD objective) Optical element (biprism or beam combiner) FS Endo-speckle-fluoroscope probe
FS (Finger splice)
Illumination Colon surface
Imaging
The endo-speckle-fluoroscope (E-S-F) probe system
Murukeshan VM, Sujatha N ( 2004)
22
Simulataneous operation of E-S-F system
23
Image Capture Methods Correlation Method Decorrelation Method
24
Decorrelation - Methodology ¾Object’s surface and camera target are aligned in the same plane. ¾Take reference image of specimen in initial state. ¾Subject the specimen to deformation by tilting, and take another image. ¾The speckle pattern is obtained by subtracting the second image from the initial image. ¾The scattered light is collected by a zoom lens and is imaged onto the photo sensor of the CCD camera. ¾Analog signal from the camera is digitized and saved in the host computer memory.
25
26
Acquire Image
Image Processing for Defect Identification and Location
Pre-Process (Filtering)
Eliminate Noise
Post-Process
Extract Features
Segment Image
Process Region
Resolution of each image : 512 x 512 pixels Depth of each pixel: 256 gray levels
Transform Results into World Coordinates
Visualize Results
27
Pre-Processing ¾ Decorrelated speckle pattern shows defects/abnormalities as indicated by dark or bright spots. ¾ Enhance contrast, brightness and eliminating noise using IRISTUTOR.
Post Processing ¾
Find the objects of interest using segmentation techniques
¾
Image Segmentation Segment object from background using threshold Smoothing of the object edges Noise removal Measuring the sizes of objects
28
Decorrelation – Without defect
Original image
After gauss filtering 29
Decorrelation – With defect
As filter size increases, more noise is removed
Pseudo defect of size 3mm
Original image
After gauss filtering, Size = 9 30
31
Results of Preliminary Investigations Locomotion Design Defect Detection and Location Modelling Region of Interest
32
Acquire Image
Rectify Image
Select ROI
Create ROI
Pre-Process
Feature Extraction and Modelling of
Segment Image
Process Region
Region of Interest Extract Features
Transform Results into World Coordinates
Visualize Results 33
Modelling Region of Interest (ROI) Test object
Decorrelated Image
A
Contour extraction using Hdevelop
B
34
Surface Reconstuction – Phase 1 ¾ Data on the coordinates of the contour of the object are extracted. ¾ Coordinates are imported into a modeling software (Geomagic). ¾ Surface is then constructed from the data points.
Data points with pseudo depth.
Original data points
35
Surface Reconstruction – Phase 2 Variation in heights of speckles due to reflectance of the laser light from the surface of the object.
36
Surface Reconstruction – Phase 2
Polygonal model is refined Feature holes are created before wrapping of surfaces 37
Surface Reconstruction – Phase 3
Create patches
Surface rendering – NURBS Model can be exported for other applications. Final model generated 38
Investigation Findings ¾ A model of an integrated fibre optics probe for the detection of very early stage of colon cancer has been conceptualised ¾ Results review the DSPI potential in identifying and locating small abnormality / defects below the surface ¾ Able to gauge the average defect size (
