Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Improving Robotic Spinal Surgery Precision and Safety with Advanced Imaging Techniques
This project aims to improve the precision and safety of spinal surgeries by developing a new method for placing screws in the vertebrae. The current method uses an optical navigation system that helps surgeons visualize the patient's spine in real-time during the surgery, but there can be errors due to patient movement. To address this, a student will use advanced computer technologies like deep learning to segment the vertebrae from imaging data, allowing the preoperative image to be registered to the segmented vertebrae and eliminating any errors caused by patient movement. The developed method will be evaluated in future experiments to ensure it can improve the safety and accuracy of pedicle screw placement in spinal surgeries.
Keywords: computer vision, stereo camera, optical camera, robotic spinal surgery
Pedicle screw placement is a common procedure used in spinal surgeries to treat conditions like herniated discs and vertebral fractures. During this surgery, doctors need to insert screws into the vertebrae without causing any harm to the surrounding nerves or blood vessels. To make the procedure safer and more precise, advanced computer technologies like the optical navigation system are already being used in clinics.
The optical navigation system helps doctors visualize the surgical tools and preoperative images of the patient in real-time, so they can place the screws correctly during the surgery. However, there is still room for error because the patient may move during the procedure, causing the preoperative imaging information to become outdated.
This project aims to overcome this challenge by segmenting the vertebrae from RGB-D spine data using deep learning or transformers. This means that the preoperative image can be registered to the segmented vertebra, eliminating any errors caused by patient movement. The project will use a dataset collected from cadaveric experiments that includes RGB-D and ground truths (segmented 3D meshes of the vertebra from CT scans) obtained from a ZED stereo camera and FusionTrack 500 optical tracking system.
Once a student has developed their method, they will incorporate it into the surgical robotic platform under the EU Horizon project FAROS. The performance of the developed method will be evaluated in future experiments to ensure it can improve the safety and precision of pedicle screw placement in spinal surgeries.
Pedicle screw placement is a common procedure used in spinal surgeries to treat conditions like herniated discs and vertebral fractures. During this surgery, doctors need to insert screws into the vertebrae without causing any harm to the surrounding nerves or blood vessels. To make the procedure safer and more precise, advanced computer technologies like the optical navigation system are already being used in clinics. The optical navigation system helps doctors visualize the surgical tools and preoperative images of the patient in real-time, so they can place the screws correctly during the surgery. However, there is still room for error because the patient may move during the procedure, causing the preoperative imaging information to become outdated. This project aims to overcome this challenge by segmenting the vertebrae from RGB-D spine data using deep learning or transformers. This means that the preoperative image can be registered to the segmented vertebra, eliminating any errors caused by patient movement. The project will use a dataset collected from cadaveric experiments that includes RGB-D and ground truths (segmented 3D meshes of the vertebra from CT scans) obtained from a ZED stereo camera and FusionTrack 500 optical tracking system. Once a student has developed their method, they will incorporate it into the surgical robotic platform under the EU Horizon project FAROS. The performance of the developed method will be evaluated in future experiments to ensure it can improve the safety and precision of pedicle screw placement in spinal surgeries.
The objective of this project is to create an algorithm that uses deep learning or transformers to segment each vertebra from RGB-D data. To achieve this goal, the project will involve a few different stages, including camera calibration to ensure accurate imaging data, data preparation to clean and prepare the data for the neural network, training the neural network to learn how to accurately segment the vertebrae, and finally evaluating the performance of the algorithm by testing the inference of the trained model on a spine phantom.
The objective of this project is to create an algorithm that uses deep learning or transformers to segment each vertebra from RGB-D data. To achieve this goal, the project will involve a few different stages, including camera calibration to ensure accurate imaging data, data preparation to clean and prepare the data for the neural network, training the neural network to learn how to accurately segment the vertebrae, and finally evaluating the performance of the algorithm by testing the inference of the trained model on a spine phantom.