Multi-Organ Models for Personalized and Evidence Based Medicine Evidence-Based Medicine Kristy K Brock, Ph.D. Physicist, Radiation Medicine Program, Princess Margaret Hospital Assistant Professor, Department of Radiation Oncology, University of Toronto Manager, Core IV, STTARR
Acknowledgements Clinicians Collaboration MORFEUS Team • Laura Dawson • David Jaffray • David Jaffray • Adil Al-Mayah Adil Al M h • Charles Catton • Michael Sharpe • Deidre McGrath • Cynthia Ménard • Doug Moseley • Michael Milosevic • Joanne Moseley • • Anthony Fyles Anthony Fyles • Jeff Siewerdsen • Jeff Siewerdsen • Andrea McNiven • Andrea Bezjak • Robert Weersink • Carolyn Niu • Rebecca Wong • Martin Yaffe • Masoom Haider • Thao-Nguyen Nguyen Thao Nguyen Nguyen • Aaron Fenster A F t • Steve Gallinger • Michael Velec • Maha Guindi • Andrew Zasowski Funding Disclosures • NIH R01, NCI Canada – Terry Fox Foundation • Ontario Institute for Cancer Research, Cancer Care Ontario Research Chair • Elekta Oncology Systems, Philips Medical Systems, RaySearch Laboratories
Intervention Begin g Image-Guided Treatment New Tumor Position! New Tumor Position! Sh hift Set-up Initial Initial
Why integration is essential… DCE DWI • Diagnostic imaging provides the optimal soft tissue contrast and tumor detail • +++Time Ti • Diagnostic imaging at each Fx is not always h F i t l necessary MRSI T2
What if more than just geometry is changing? i h i ? • In room imaging allows identification of response during Tx • Adapt to these changes h • Offline diagnostic i imaging + dose i + d accumulation + integration integration
…greater than the sum of the parts CT MR • • Unique information from each Unique information from each image • Geometric discrepancies prevent accurate correlation of prevent accurate correlation of local information • Resolving these can lead to improved understanding of improved understanding of disease and therapeutic response – Multi-Modality Image Integration Multi Modality Image Integration and Validation – Physiological Modeling – Image Guided Therapeutics age Gu ded e apeut cs CT/PET Histology – Response Assessment
Why Deformable? Shrinking Shrinking Breathing Breathing Filling Filling
Registration: Completing the Loop Accurate Target Definition : Multi-modality Image Registration Pl Planned d Accurate Response Dose Assessment : Accurate Follow-up Image p g Motion Assessment : Motion Assessment : Dose Dose Predicted Predicted Registration and Multi-instance Image Effect Dose Correlation Registration with Tx Delivered Dose Accurate T Tumor Guidance : G id Online Image Registration
Delivered Planned = Delivered ? Planned
Target Definition Delivered Planned = Delivered ? d Planned Pl
Accurate Target Definition Prior to Deformable Registration coronal coronal sagittal Before After Deformable Registration GTV Volume CT = 13.9 cc MR = 6.7 cc Δ Vol = 7.2 cc (52%)
Removing Confounding Geometry CT-exhale CT GRV MR-exhale
In Vivo Image Validation Triphasic CT Images Multiple Sequence MR Images FDG-18 PET Images Surgical Excision of Liver Lobe S i l E i i f Li L b Fresh Specimen MR Imaging Specimen Fixation Fixed Specimen MR imaging Fixed Specimen MR imaging Specimen dissection Histological Analysis of Tumor
Pathology @ Fresh w/ in Fresh w/ in Pathology w/ Vivo Liver Fixed Aligned Aligned Specimen S i Aligned g Pathology @ Tumor Fixed w/ Fresh Comparison Aligned Specimen
Target Definition Internal Motion Delivered Planned = Delivered ? Planned
Physiological Modeling • Complex differences Complex differences between structures and patients • Understand physiologic processes – Interplay with image analysis – Influence on delivering Influence on delivering therapeutic intent
Multi-Organ Model
Sliding Interface A. Al-Mayah, K. Brock
? Planned = Delivered Targeting Targeting Internal Motion Planned Target Definition Target Definition Delivered
Image Guided Therapeutics • Speed and Accuracy is essential • Tissue loss/therapy response compromises i algorithms • Intra-intervention I t i t ti images lack detail
Precision Image-Guidance Resolve Geometric discrepancies New Tumor Position! Planning CT [w contrast] CBCT [w/o contrast]
Accurate Tumor Guidance 12 12 Liver Patients: 6 Fx Each i i 6 h Rigid Reg → Deformable Reg g g g Δ Tumor dLR dAP dSI abs(dLR) abs(dAP) abs(dSI) AVG -0.03 -0.01 -0.02 0.07 0.10 0.08 SD SD 0 10 0.10 0 16 0.16 0 14 0.14 0.07 0 07 0 12 0.12 0 12 0.12 Max 0.28 0.65 0.52 0.36 0.65 0.57 Min -0.36 -0.64 -0.57 0.00 0.00 0.00 Median Median -0 03 0.03 0 01 0.01 0 00 0.00 0 05 0.05 0 06 0.06 0 04 0.04 • 33% (4/12) Patients had at least 1 Fx with a ( ) Δ COM of > 3 mm in one direction • 10% of Fx had a Δ COM of > 3 mm in 1 dir.
? Planned = Delivered Targeting Targeting Geometric Internal Motion Planned Target Definition Target Definition Dosimetric Delivered
Target Definition Target Definition Internal Motion Delivered Targeting Planned = Delivered Geometric Dosimetric ? Magnitude? Planned
Change in Delivered Dose Predicted Dose EXH INH Planned Dose
Data 14 SBRT Li 14 SBRT Liver Patients: Avg Rx dose of 40 Gy in 6 Fx P ti t A R d f 40 G i 6 F • Liver, GTV, critical normal tissues, external surface external surface, and spleen were contoured on the exhale scan. • External surface, li liver, and spleen d l were also contoured on the inhale CT o e a e C scan.
Change in Predicted Dose 7 Gy Reduction in • Normal Tissue mean tumor dose Complication Probability: – -13 to 9% 13 t 9% 10 Gy Reduction • Change in Max dose to in Max Dose critical structures critical structures – Up to 10 Gy • Change in Min dose to g tumor – -8 to 2 Gy This is PREDICTED – not DELIVERED!
Tx Change in Delivered Dose Delivered Dose EXH INH Predicted Dose EXH INH Planned Dose
Changes in Dose: 2000 1500 Accumulated through Accumulated through 1000 1000 500 6 Fx 0 -500 500 -1000 -1500 cGy Static (exhale) Difference Accumulated
Results [Gy] D stat vs D pred D pred vs D acc GTV GTV -1 5 to 3 7 1.5 to 3.7 -4 2 to 2 3 4.2 to 2.3 MIN dose Stomach -4.0 to 0.4 -2.9 to 5.1 MAX dose Esophagus -0.4 to 5.2 -3.4 to 0.1 MAX dose MAX dose Duodenum -7.6 to 0.0 -0.3 to 5.7 MAX dose Liver -1.0 to 1.2 -1.4 to 0.9 MEAN dose Kidneys -3.8 to 0.0 -0.9 to 3.9 MEAN dose
Planned ≠ Delivered Targeting Geometric Internal Motion Planned Target Definition Target Definition Dosimetric Delivered Magnitude Clinical Results
Response Assessment • Longitudinal studies • Longitudinal studies Planning CT Follow-Up CT Pl i CT F ll U CT enable understanding of disease response • Gross volume/structure changes confound correlation between correlation between information time points • Inhibit correlation with locally delivered therapies
Accurate Response Assessment
Accurate Response Assessment
Volume Change vs. Delivered Dose 20 %] 15 Change [ 10 A B 5 5 n Volume C D 0 E F tive Mean G -5 H I -10 Relat -15 -20 -20 0 1000 2000 3000 4000 5000 6000 7000 Dose [cGy]
Summary • Advances in multi-modality, multi-temporal, and online imaging is generating large amounts of data amounts of data • To fully understand this information, integration and correlation is essential integration and correlation is essential • Geometry (physiological, therapeutic response, intervention) confounds integration, therefore limiting knowledge • Deformable modeling can remove these confounding factors confounding factors
Summary • Multi-modality integration can improve understanding of disease extent d t di f di t t – Correlation with pathology can validate this information and aid in development of new imaging p g g techniques, contrast agents, and probes • Physiological modeling can improve our understanding of these confounding factors understanding of these confounding factors – Aid in reducing them during therapeutic intervention • Improvements in image guidance through Improvements in image guidance through accurate incorporation with pre-Tx information • Longitudinal studies with response assessment g p can improve understanding of therapeutic response and normal tissue toxicities
Recommend
More recommend
