Project · Medical Robotics
Robotic Assistant for MRI-Guided Low Back Pain Injections
Low back pain affects over 600 million people globally, and MRI-guided spinal injections offer a radiation-free diagnostic and therapeutic approach. This project continued development of a 5-DOF MR-compatible robot for lumbar spine needle insertion, focusing on improving mechanical rigidity to meet clinical accuracy requirements (<1 mm positioning, <1° angular error).
My primary contribution was leading the FEA-based redesign of the robot's linkage profiles. The original PLA linkages lacked sufficient stiffness for needle insertion forces, while carbon fiber—though rigid—is only MR-conditional due to potential heating effects. I systematically analyzed six beam cross-sections, optimizing for the highest second moment of inertia to weight ratio. Through iterative FEA simulation, I developed an I-beam profile in ABS that matched the 6 mm carbon fiber's stiffness (15 µm deflection under 5 N loading) while maintaining full MR compatibility.
I also contributed to CAD modeling, 3D printing the optimized profiles, validating performance through bending tests, assisting with robot debugging, and documenting the design process. The redesigned profiles are ready for integration into the robot to improve endpoint accuracy for future phantom and clinical testing.
Problem Statement
- Need: MRI-guided spinal injections require a robot with high endpoint accuracy (<1 mm positioning, <1° angular error) that is fully MR-compatible
- Prior art: The robot's original PLA linkages lacked sufficient rigidity for needle insertion forces, causing deflection that compromised endpoint accuracy. Carbon fiber linkages improved stiffness but are only MR-conditional due to potential heating in MRI fields
- Open question: Could an optimized 3D-printable profile using MR-compatible materials (ABS/PLA) match carbon fiber stiffness without changing the robot's kinematics or workspace?
Methodology
The project followed a systematic design-simulate-fabricate-test workflow for linkage profile optimization. Click a stage to jump there.
Design & Simulation
Problem Definition & Constraints: Design constraints included: fixed bolt hole dimensions, linkage length (111 mm), width (12 mm), and material limited to ABS for MR-compatibility.
Profile Selection via FEA: Six beam cross-sections were analyzed using FEA simulation under identical loading conditions (5 N cantilever load, ABS material). Profiles compared included I-beam, C-channel, hollow rectangular, solid rectangular, T-beam, and custom sections. The selection criterion was the ratio of second moment of inertia to weight—maximizing stiffness while minimizing mass. The I-beam profile emerged as optimal with the highest ratio (10.01×10⁻⁴) among all candidates.
Iterative I-Beam Optimization: With I-beam selected, iterative FEA optimization varied geometric parameters to match carbon fiber performance. Attempt 2 varied web width, confirming increased width improves rigidity. Attempt 3 fixed width at 7 mm and varied height to match 5 mm and 6 mm carbon fiber deflection targets. Attempt 4 optimized beam proportions (70% web height, 15% flange height). A second-order curvature optimization on the flanges further reduced deflection while achieving only 2% weight reduction—confirming the design was near-optimal.
Fabrication
The optimized I-beam profiles were fabricated using 3D printing with ABS material. The final I-beam parameters were: Total Height 48 mm, Web Height 33.6 mm, Web Thickness 4.2 mm, and Beam Base Height 7.2 mm. Multiple iterations were printed to validate the FEA predictions through physical testing.
Test Setup
Physical bending tests validated FEA predictions by loading 3D-printed ABS profiles and 6 mm carbon fiber members at 5, 10, and 15 N. Deflection was measured at the free end across 8 equidistant points.
Results
The 3D-printed ABS I-beam profile achieved carbon fiber-equivalent stiffness while maintaining full MR compatibility.
Key Finding Carbon fiber Young's modulus discrepancy
- The 3D-printed profile matched carbon fiber deflection as predicted by FEA
- However, measured deflections differed from FEA predictions
- Investigation revealed that the carbon fiber's true Young's modulus (~45 GPa) was significantly lower than datasheet values (240 GPa)
- Adjusting the FEA model with experimental Young's modulus reconciled the results
Achievements Key project outcomes
- Designed MR-compatible ABS I-beam profile achieving 1.6 µm deflection (outperforming 6 mm carbon fiber)
- Identified optimal cross-section from 6 candidates using second moment of inertia to weight ratio analysis
- Validated FEA predictions through physical bending tests at 5, 10, and 15 N loads
- Discovered carbon fiber Young's modulus discrepancy (45 GPa actual vs. 240 GPa datasheet)
- Redesigned profiles ready for robot integration to improve endpoint accuracy
Skills Technical skills demonstrated
- Simulation: Finite Element Analysis (FEA), ANSYS/COMSOL
- CAD: SolidWorks, profile optimization
- Fabrication: 3D Printing (ABS)
- Design: Mechanical Design, Structural Analysis, MR-Compatible Design
- Analysis: MATLAB, Experimental Validation
Team Contributions & credits
I led the finite element analysis (FEA) effort to redesign robot linkage profiles, achieving carbon fiber-equivalent stiffness using MR-compatible ABS material. Contributed to bending test validation, CAD modeling of optimized profiles, 3D printing fabrication, documentation, and robot debugging. Presented FEA work in all team meetings and project presentations.
Guided by Dr. Iordachita (Faculty Advisor) and Aabhas (Project Mentor). Technical collaboration with Tyler Lehrfeld (Software Development, Optical Tracking, Testing Procedures) and Diana Shaughnessy (CAD, Robot Debugging, Lower Body Mount, Data Analysis).
Conducted at Johns Hopkins University Laboratory for Computational Sensing and Robotics (LCSR).
Additional Information
Final project report with detailed methodology, results, and deliverables status.