1.  Using High Performance Computing to enhance the clinical decision making process in low back pain Prof. Damien Lacroix INSIGNEO Institute for in silico Medicine Department of Mechanical Engineering University of Sheffield

2. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Low back pain impact

3. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Treatment decision

4. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights IT Engineering into clinics Lumbar spine surgery leads to satisfactory short-term pain relief, but long/midterm outcomes are less positive in terms of incremental degeneration problems and pain resurgence We hypothesize that the lack of long term success is due to: -Incomplete understanding of the biomechanics of the lumbar spine -Required specificity of the patient condition (geometry, disc properties, loading, psychological profile, etc…) -A non-rationale engineering approach

5. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights MySpine consortium

6. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights MySpine objectives MySpine aims to create a clinical predictive tool to provide clinicians with patient-specific biomechanical and mechanobiological analysis. This tool will help to determine the best patient specific treatment for low back pain. The project will focus on disc degeneration pathology although the developed prototype system may be able to analyze other spinal pathologies as well.

7. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Patient-specific framework Clinical software Multiscale biomechanical analysis

8. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Subject-specific finite element model Morphing FE models onto disc / vertebrae shape model Generic FE model Subject-specific shape model Subject-specific FE model MRICT +

9. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Patient-specific mesh

10. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Patient-specific 3D L1-S1 lumbar spine Post-processing through ABAQUS D7.3

11. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Patient-specific 3D L1-S1 lumbar spine Post-processing through ABAQUS

12. VPH-Share: A secure web portal for the construction of personalised clinical workflows VPH-Share Scientific Coordinator: Prof. Rod Hose, University of Sheffield

13. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights MySpine Major Platform Components VPH-Share XNAT Database MySpine Image-based spine reconstruction HPC Server Abaqus simulation: Short terms effects (+Matlab, Python) Abaqus simulation: Short terms effects (+Matlab, Python) Abaqus simulation: Ageing effect (+Matlab, Python) Abaqus simulation: Ageing effect (+Matlab, Python)

14. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Python Script Pre-processingComputationPost-processing

15. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Python Script Pre-processingComputationPost-processing Generic Abaqus input file Personalized inputs

16. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Python Script Pre-processingComputationPost-processing Patient-specific Abaqus input file

17. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Client Access to Visualisation Cluster VirtualGL Client Iceberg – Campus Compute Cloud VirtualGL Server (NVIDIA Fermi M2070Q GPU)

18. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights P. 18 Time / Cost analysis per patient Patient specific model preparation (Gimias): 3h (1 core) Full lumbar simulation (Abaqus): 8h per loading & per treatment (16 cores) Bone remodelling and disc adaptation (Abaqus): 7h per year simulated Based on Amazon c3.4xlarge (16C, 30GB RAM) = $1.366/h

19. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights P. 19 Time to complete clinical prospective study 200 patients have been treated and followed for 2 years With 400 cores available (e.g. N8), need only 23 days to run

20. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Discussion With additional resources, further analyses could investigate additional loadings variability in surgery techniques different types of screws to provide an interval of confidence for the clinician.

21. The document is proprietary of the MySpine consortium members. No copy, distribution or use is allowed without the prior written agreement of the owner of the property rights Conclusions A patient-specific platform to analyse the biomechanical load transfer between conservative treatment, discectomy and fusion was developed. Using VPH-Share cloud facilities and local or cloud HPC system, any clinician can have a patient-specific biomechanical analysis in less than 11 hours. Remote desktop visualization and cloud computing can provide a transformative change to NHS without hardware investment.

22. Acknowledgements VPH-Share Prof. Rod Hose Dr. Susheel Varma Dr. Steven Wood

23. Institute for in silico Medicine Thank You!