Work Term #1 Summary of objectives and completed work

Matthew Strugari

May 2, 2014

Outline

• Background

• Workflow

• Calculations

• Results

• Future Work

Prostate Deformable Registration

• Background

• QATrack+

• Results

• Future Work

Quality Assurance

Deformable Registration

Research Project #1

Deformable Registration

A method of mapping two or more imaging data sets

Endo-rectal coil Endo-rectal coil

Prostate Prostate

Bladder

Deformable Registration

• Development and validation of an open source deformable registration technique for efficient integration of MR imaging into radiation treatment planning

Synopsis

• To develop a phantom based method to evaluate the efficacy of an established deformable image registration technique

Goal

CIRS Multi-Modality Imaging Phantom

Fixed Image Moving Image

Water Bags Water Bags

Prostate

Endo-

Rectal

Coil

Urethra

Axial Slice

CIRS Multi-Modality Imaging Phantom

Fixed Image Moving Image

Prostate

Urethra

Endo-Rectal

Coil

Seminal

Vesicles

Seminal

Vesicles

Endo-Rectal

Coil

Fiducial

Marker

Prostate

Urethra

Fiducial

Marker

Sagittal Slice

General Workflow

• Contours are created using Aria Eclipse treatment planning system

• Deformable registration technique is applied using the 3D Slicer open source software

• Measurements and calculations are performed directly within 3D Slicer

Contours - Prostate

Fixed Image Moving Image

“Eclipse”. Varian Medical Systems. 2014. 1 May 2014. <http://www.varian.com/us/oncology/radiation_oncology/eclipse/>.

Axial Slice

Import

Crop

Registration Deform

Contours

Calculations

3D Slicer Workflow

3D Slicer Workflow

Full Image

Cropped Image

B-Spline Deformable Registration

Process

1. Defining Control Points

STANESCU, Teodor et al. Investigation of a 3D system distortion correction method for MR images. Journal of Applied Clinical Medical Physics, [S.l.], v. 11, n. 1, jan. 2010.

ISSN 15269914. Available at: <http://www.jacmp.org/index.php/jacmp/article/view/2961/1799>. Date accessed: 01 May. 2014. doi:10.1120/jacmp.v11i1.2961.

2. Computing the B-Spline Functions

Spline

Basis Spline

•“Quadratic Spline”. Wikipedia. 29 Jan 2011. 1 May 2014. <http://en.wikipedia.org/wiki/File:Quadratic_spline_six_segments.svg>.

•Sharp, Gregory. “Cubic B-Splines”. Deformable Image Registration Using B-Splines. 10 Feb 2011. 1 May 2014.

<http://chapter.aapm.org/NE/DOCUMENTS/Presentations/2011Winter/sharp-neaapm-2011-02-10.pdf>.

3. Computing the Deformation Field

vx(u)=∑iPiBi(u)

Sharp, Gregory. “Cubic B-Splines”. Deformable Image Registration Using B-Splines. 10 Feb 2011. 1 May 2014.

<http://chapter.aapm.org/NE/DOCUMENTS/Presentations/2011Winter/sharp-neaapm-2011-02-10.pdf>.

4. Evaluating the Cost Function

Inflated vs. Deflated

5. Optimization: Gradient Descent

vx(u)=∑iPiBi(u)

Gradient Descent Results

Algorithm Summary

Control Point

Coefficients

B-Spline Functions

Deformation Vector Field

Cost Function

Gradient Descent

Control

Points

Deformed

Images

Results: Deformation Vector Field

Results: Deformation Vector Field

Results: Deformed Image

Fixed Image Deformed Image

Axial Slice

Results: Contour Deformation

Fixed Prostate

Deformed Prostate

Fixed Urethra Fixed Seminal Vesicle

Deformed Urethra Deformed Seminal Vesicle

Calculations: Dice Similarity Coefficient

Urethra Seminal Vesicles

Prostate

Calculations: Hausdorff Distance Metric

Axial Slice Sagittal Slice

“Hausdorff Distance Sample”. Wikipedia. 6 May 2008. 1 May 2014. <http://en.wikipedia.org/wiki/File:Hausdorff_distance_sample.svg>.

Calculation Results

Organ

Dice

Similarity

Coefficient

Hausdorff

Distance

(mm)

Reference

Volume

(cc)

Deformed

Volume

(cc)

Prostate 0.9733±0.0003 3.56±0.01 50.61 53.26±0.03

Urethra 0.917±0.001 2.40±0.10 3.265 3.529±0.009

Seminal

Vesicle 0.843±0.008 3.30±0.2 2.87 2.22±0.04

Conclusion

It is possible to develop a phantom based method to

evaluate the efficacy of an established deformable

image registration technique

The results for the main anatomical regions were

satisfactory

Conclusion

This method can be applied to a number of different

modalities such as CT, CBCT, MR, and PET

Deformable registration is useful for adaptive

radiotherapy to account for anatomical changes

throughout a patient’s treatment

Future Work

Automating the process using Plastimatch

Applying this process to in vivo data sets

Finite element analysis for prostate deformation

using Elmer FEA

Quality Assurance (QA)

QA: Outlined by CPQR

Functionality Reproducibility Accuracy

“Emergency Stop”. Dreamstime. 2014. 1 May 2014. < http://www.dreamstime.com/photos-images/security-emergency-help-point.html>.

“The Life Cycle of a Website”. Standing Dog. 12 Mar 2013. 1 May 2014. < http://www.standingdog.com/blog/the-life-cycle-of-a-website/>.

“Accuracy and Precision”. Math is Fun. 2013. 1 May 2014. < http://www.mathsisfun.com/accuracy-precision.html>.

Machine QA

Daily

Weekly

Monthly

Annually

Radiation Safety

Mechanical

Imaging

Dosimetry

Machine QA

CatPhan

“CatPhan©504 Manual”. The Phantom Laboratory. 2013. 1 May 2014. <http://www.phantomlab.com/library/pdf/catphan504manual.pdf>.

“Profiler 2”. Sun Nuclear Corporation. 2013. 1 May 2014. <http://www.sunnuclear.com/medPhys/machineqa/profiler2/profiler2.asp>.

“ArcCHECK”. Sun Nuclear Coroporation. 2013. 1 May 2014. <http://www.sunnuclear.com/medPhys/patientqa/arccheck/arccheck.asp>.

Current Condition of QA

Argus implemented in SCC

QATrack+

QATrack+

QATrack+ Features

QATrack+ Features

Trendline

Reference Value Tolerance Level

Action Level

•Oasis Daily QA, 6 MV 10x10 output

QATrack+ Features

•Oasis Daily QA: 6 MV, 10 MV, and 15 MV 10x10 output

6 MV

15 MV

10 MV

Results

Two complete months of daily QA data for all

machines

QA data and results are kept in one location for daily,

weekly, and monthly QA

Results: Oasis

6 MV Daily 6 MV Monthly

10 MV Daily

10 MV Monthly

15 MV Daily

15 MV Monthly

Conclusion

QATrack+ saves time, effort, and money

QATrack+ meets the CNSC standards and legal requirements

Records are stored to a database and are readily available

QATrack+ is fully customizable

Unpredictable run times due to the use of QATrack+ software on a desktop computer

Future Work

Modifications/elimination of weekly QA

Implementation of monthly and annual QA

Transfer QATrack+ to a server

Regular review QA data

E-mail notifications

Setup of unique user groups and user accounts

Special Thanks

Saskatchewan Cancer Agency

Medical Physics: Deluan Tu; Niranjan Venugopal; Andrew Alexander

External: Gregory Sharp, Medical Physicist (Massachusetts General Hospital); Randle Taylor, Medical Physicist (Ottawa Hospital Cancer Centre)