Illustration of the radiotherapy room and the occlusion drawback confronted by ceiling-mounted cameras on this software.
What was the subject of your PhD analysis and why was it an fascinating space?
My subject of analysis was creating an optical tactile sensor to trace head movement throughout radiotherapy. I labored on each the {hardware} and software program improvement of this sensor, although my focus was totally on the software program facet. Its significance comes from the truth that throughout radiotherapy, sufferers present process head and neck most cancers therapy are usually immobilised. That is often performed utilizing a thermoplastic masks, which may really feel very claustrophobic, or a stereotactic body. Frames are extra frequent for mind cancers, however they need to be surgically inserted into the affected person’s cranium utilizing pins. Both of those immobilisation instruments could also be used relying on the state of affairs. When sufferers are uncomfortable, they’re extra prone to transfer, which impacts the accuracy of therapy, particularly with thermoplastic masks.
One other main situation is that present techniques use ceiling-mounted cameras to file affected person movement. These cameras can’t be positioned too near the affected person due to the electromagnetic surroundings across the tools. Their view can be incessantly occluded as a result of the affected person strikes right into a tunnel to obtain the ionising beams, which makes it troublesome to seize rotational movement.
One various is an infrared digicam with a nostril marker, however this solely captures translational movement. Presently, when a nostril tracker detects motion past a sure threshold, therapy is paused, the affected person is repositioned, and therapy resumes. It’s troublesome to adapt this method to reliably measure the rotational movement of the affected person’s head within the radiotherapy surroundings.
That is the place the Movement Seize Pillow (MCP) is available in, which incorporates the optical tactile sensor I developed. The objective with this method is much like the nostril tracker, however with extra correct rotational suggestions for the radiographer. It may be positioned beneath the affected person’s head and connected to the therapy mattress. It estimates how a lot the affected person’s neck is rotating and improves affected person consolation. Radiographers can obtain real-time suggestions on each translational and rotational motion. The benefits of this method are that there aren’t any occlusions, as a result of the pillow is in direct contact with the affected person’s head, and it’s extra appropriate with radiotherapy environments as a result of the sensor is non-ferromagnetic. Its premise is to take care of affected person consolation, keep compact and straightforward to combine into the pre-existing techniques for radiotherapy, while enhancing the accuracy of the therapy by way of real-time head monitoring.
Labelled diagram of the Movement Seize Pillow – Optical tactile sensor for head monitoring throughout radiotherapy. The pneumatic pillow is a deformable rubber-like sheet with embedded white markers, held in its convex form utilizing air strain. The fibrescope represents a non-ferromagnetic fibre optic bundle used as a lens extension to an space scan digicam. The digicam is ferromagnetic and would require protected positioning and fixation.
What had been the primary contributions of your work?
There have been 4 essential contributions to my work. The first contribution targeted on making the system extra non-ferromagnetic and enhancing the imaging and monitoring strategy. Earlier work used a webcam and binary picture processing inside the optical tactile sensor to trace marker displacement. I finally determined to make use of a fibrescope, optical move monitoring algorithm, and grayscale imaging as a substitute, which improved the sensor’s monitoring capability.
The second contribution targeted on optimising marker density. The optical tactile sensor consists of an array of markers on a deformable rubber-like sheet, resembling a pillow. The deformation of those markers is captured by the digicam. I investigated how dense the marker array wanted to be by adjusting the spacing between markers to find out what labored finest for this software.
The third contribution concerned sensor fusion to enhance reliability and robustness. To do that, I built-in a gyroscope and used Kalman filtering to fuse knowledge from the gyroscope and the MCP. This was necessary for Gamma Knife techniques, that are radiosurgery platforms used for mind cancers. They have a tendency to have larger accuracy necessities than linear accelerators, that are generally used for head and neck cancers, and decrease constraints on using ferromagnetic parts.
The ultimate contribution was a participatory design research carried out in collaboration with clinicians and the social sciences division. We explored how the MCP might be built-in into hospital workflows and assessed its feasibility.
How possible is it to combine this sensor into hospital workflows?
Clinicians did appear to be very on board with it, however the research was extra qualitative than quantitative. Whereas they felt the thought had benefit, there have been reservations about adopting new know-how and the related studying curve.
They had been additionally involved about accuracy. Bettering accuracy and reliability is important for clinicians to really feel assured utilizing the system. At current, additional improvement is required earlier than it may be broadly applied.
What future work is deliberate on this space?
One space to analyze is the variations between the model and participant knowledge. The pillow form is managed by a pneumatic system with a strain sensor and air pump. When the affected person or model strikes, strain modifications happen. The system compensates to take care of a set strain, however this introduces errors within the movement readings. The model produced extra errors than the participant knowledge. It might not precisely simulate human movement on the pillow, and the testing setup might introduce discrepancies that don’t replicate real-world behaviour.
So, future work consists of stabilising and refining the strain management system to enhance reliability. If obligatory, reconsidering using gel on the sensors might be an possibility. Gel had been used beforehand however was deserted attributable to clinician issues about attenuation of ionising beams. Nonetheless, if avoiding gel considerably compromises sensor efficiency, revisiting this strategy could also be worthwhile.
As well as, extra participant knowledge assortment is required. Not all beforehand collected knowledge might be used attributable to ground-truth measurements being partially occluded within the experimental setup. Further participant research would offer a clearer understanding of efficiency throughout totally different people. One other precedence is enhancing the fibrescope’s decision and angle to raised visualise high-density marker arrays. {Hardware} upgrades would assist guarantee a clearer discipline of view and enhance total system efficiency.
About Bhoomika
 |
Bhoomika Gandhi is a latest PhD graduate from the College of Sheffield Medical Robotics group. Her undergraduate diploma was in Bioengineering – Medical Gadgets and Devices, with management engineering and robotics being the important thing themes. |
Ella Scallan
is Assistant Editor for AIhub
