Senior Design Project Megan Luh Hao Luo Febrary

Analysis Problem Statement Current methods of limb alignment are costly and time consuming Dependent on individual surgeon skill for accurate calibration Performance Criteria Constrained by surgical space, time, and resources Limited by lens quality, camera resolution and frame rate, and noise level

Primary Objective Proof of Concept that visual recognition software can be applied to the field of limb alignment in real-time for surgical procedures Improve the method of limb alignment used during surgical procedures Create a new method that is more efficient, can be used in real-time, more economically profitable for hospitals.

Hypothesis Solution: Utilize computer vision software in real time and implement it for limb alignment Goals: Create a computer vision system using OpenCV and design necessary components for surgery

Factors Parameters Quality is determined by the speed, accuracy, and precision of the computer algorithm Overall operating costs are reduced with a faster system Patient and surgeon both benefit from a faster, more accurate system Average operating room costs = $ per min Surgical costs Doctor visits; pre surgery and exams (total 3) $512 MRI $ Hospital $4,909 Anesthesia Doctor Charge: $3591 (surgery) total amounts =10,722.20

Interview with Dr. Christie Founder of the Vanderbilt Arthritis and Joint Replacement Center. Co-founder of the Southern Joint Replacement Institute Topics: Surgical spatial constraints Initial incision = 6 inches Initial tibia leveling = approximately 10 mm

Marker Designing a cross shape marker with some spheres on it to mark the x-ray It consists of four spheres connected in a cross configuration The two pairs of spheres vary in size and in color Use a biocompatible, disposable plastic with an x- ray contrast medium: polyethylene, polycarbonate

Flow Chart (Stage1)

Flow Chart (Stage2)

Progress Circle Detection Line Detection Contour Detection Camera Calibration

Next Step Length calculation Ratio Perception User Interface

Performance Accuracy on Circle Detection Effect of Noise 90% accurate

Testing Strategy Need an experimental procedure to quantify the success of our program Want to calculate how accurately the camera detects the location of the spheres in 3D space and their spatial orientation Do this with a simplified experimental model Tibia: modeled with a cylindrical PVC pipe Test camera at different distances and different angles

Conclusion The goal of this project is to accomplish a proof of concept that visual recognition software can be applied to the field of orthopedic limb alignment in a real-time surgical procedure. So far, we have solidified the goal and mapped out the details of software implementation. Futures works include creating the software, troubleshooting, and testing the result.

References Duda, R. O. and P. E. Hart, "Use of the Hough Transformation to Detect Lines and Curves in Pictures," Comm. ACM, Vol. 15, pp. 11–15 (January, 1972). Bradski, Gary, and Adrian Kaehler. "Image Transforms, Contours, Project and 3D vision." In Learning OpenCV: Computer Vision with the OpenCV Library. 1st ed. Sebastopol: O'Reilly Media, Inc., , , , Chleborad, Aaron. "OpenCV's cvReprojectImageTo3D." Graduate Student Robotics Blog. (accessed December 18, 2009). Levent Kosumdok. “Plastic with special built-in function.”