1. 1. NASS (Network-Aware Supervisory System) Framework Demonstration Airway-laser Surgery Charlie Kim Maryam Rahmaniheris Lui Sha University of Illinois at Urbana-Champaign October 2010
2. 2. Outline Demonstration Scenario Surgery Scenario Potential Hazards Motivation NASS Framework Fundamental Goals Flow of Airway Laser surgery Airway Laser Surgery Components Airway Laser Surgery Demonstration Closed-loop Safety Open-loop Safety Nass Framework- Behind the Scene
3. 3. Surgery Scenario Laser surgery on vocal cords. An endotracheal tube is used to aid the patient with breathing during the surgery.
4. 4. Potential Safety Hazard 1: Oxygen + Laser -> Fire! Relative flammability of various ETTs Laser heat as the ignition source Oxygen concentration > 20%
5. 5. http://www.mdpnp.org/uploads/Capitol_Hill_NSF_CPS_MD_PnP_9July09.pdf
6. 6. Potential Hazard2: Brain Damage Oxygen flow is closed when laser is activated Brain injury can happen due to the lack of Oxygen
7. 7. Motivation
8. 8. Motivation Seems simple to prevent! Why 600 fires every year?! The Answer is simple: To Err is Human!
9. 9. MDPnP : The Medical Device Plug-and-Play Chair: Dr. Julian Goldman Inter-disciplinary, multi-institutional medical device informatics research program. Leading patient-centric Integrated Clinical Environment (ICE) standard MDPnP group is based at CIMIT and Massachusetts General Hospital. The MDPnP Program
10. 10. ICE Architecture
11. 11. NASS Framework Replace human with machines! But machines fail, too. What if there is a network failure or the computer suddenly fails in the middle of the procedure? NASS framework An implementation of ICE Maintains safety in spite of the failure of network or any of the computers
12. 12. NASS Framework: Fundamental Goals Certifiably safe automation of inter-device coordination Replacing human implemented safety interlock with computer implemented safety interlock Maximal use of common computers and equipment Low cost Convenience Robustness
13. 13. ICE Architecture NASS framework is a fail- safe implementation of ICE architecture. NASS Supervisor Wi-Fi Medical Device Customized HW Application
14. 14. Airway laser Surgery : Components
15. 15. Flow of the Airway Laser Surgery Anesthesia Monitoring the patient Block Oxygen flow & Activate Laser Configuration check Do not block Oxygen flow Request Laser activation Is SpO2 above the safe threshold? Move the patient to OR Yes No
16. 16. NASS Framework: Behind the Scene
17. 17. NASS Framework NASS Supervisor WLAN NASS Device Medical Device FPGA NASS Device Medical Device FPGA NASS Device Medical Device FPGA NASS Device Medical Device FPGA ApplicationApplication Application Application
18. 18. NASS Supervisory Plan Time Mode 1-5s Enforced 5-15s Enforced 16-29s Prohibited 30-50s Prohibited 51-59s Prohibited 60s-~ Enforced NASS Device NASS Supervisor Records the Plan (Generates the Plan) Plays the Plan (Follows the Plan) Plan Medical Device Application Only completed plan is executed (Atomic Transaction) FPGA Execution of contingency safety plan without communication with supervisor.
19. 19. NASS Device Architecture FPGA and Medical device do not fail Any component above the FPGA can fail FPGA stores and executes the plans sent by the supervisor When the Device-Supervisor loop is open, FPGA keeps executing the stored plan NASS Device Medical Device FPGA Application
20. 20. Hardware/Software Co-design Full SW Design HW/SW Co-Design Software Application Library Operating System Microprocessor Need to Have Verified Robustness Software Application Library Operating System Microprocessor HW COTS FPGA Needs to Have Verified Robustness
21. 21. Hardware/Software Co-design Versatility FDA Certification Framework Safety layer Effectiveness layer
22. 22. Limitations of NASS Framework Currently the actual amount of medicine injected by the medical staff is not monitored by the system The NASS framework can be easily extended to cover this case IV tube sensors and scanners can be used for this purpose Safety rules are pre-loaded at the configuration time. They are verified by the physician for each specific procedure and cannot be changed on the fly. Safety rules should be simple rules that can be correctly and unambiguously interpreted by machine