Medtronic CARES
Medtronic CARES
Embedded Files
A fully functional prototype of a catheter based medical robot and its digital twin that can navigate through a lung model autonomously when provided the goal state. This serves as a proof of concept for an actual medical robot that can perform peripheral lung biopsy fully autonomously when the CT scan of a lung model is provided and the location of a potential cancer cell inside the lung is identified.
A fully functional prototype of a catheter based medical robot and its digital twin that can navigate through a lung model autonomously when provided the goal state. This serves as a proof of concept for an actual medical robot that can perform peripheral lung biopsy fully autonomously when the CT scan of a lung model is provided and the location of a potential cancer cell inside the lung is identified.
Skills developed:
Hardware: Rapid prototyping, Surface modelling, Wiring and soldering, Design for assembly
Software: Object oriented programming, Multithreading, Model View Controller
Algorithms: RRT star, Voxelization
Supervised Machine Learning: Convolutional Neural Network, Multi Layer perceptron, Recurrent Neural Network
Control: Forward and Inverse Kinematics with Feed forward neural network
Results - Hardware
Results - Hardware
Phases of hardware development
Results - Software
Results - Software
Graphical User Interface
Voxelization: Converting STL files to arrays for working in the backend
Path planning: Iteration - RRT, RRT star, bidirectional RRT
Pose prediction : Forward kinematics with feed forward neural network
(Val accuracy - >99%)
Catheter tip controls: Inverse kinematics with feed forward neural network (Val accuracy 89%)
Wall Detection: Convolutional Neural Network (Val accuracy 98%)
Page updated
Google Sites
Report abuse