MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Towards Automation in Radiation Therapy: Extracting a Breathing Signal From a Patient-Specific Region Using a Depth Sensor

E Mathias*, R O'Brien, ACRF Image X institute, Sydney, NSW, AU.

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: non-contact respiratory monitoring can enhance motion management strategies in radiotherapy. Current methods are either unreliable for real-time measurements or do not automatically detect the optimal region for monitoring. We have developed a method to automatically determine the optimal region in real-time for breathing monitoring.


Methods: image obtained from the depth sensor is split into 15×15 rectangular regions and the average depth of all the pixels in a region is computed. A signal from a region is considered valid if the period (2-20 s) and amplitude (2-15 mm) of the signal is within the expected range. To ensure stable recording, the largest number of connected regions, each with a valid signal, was detected. When a region is valid for over 100 consecutive frames it is marked, and the average depth value of those regions is computed. To validate the measurements obtained from the automatically detected region, a Real-Time Position Management (RPM) sensor was used simultaneously. The Anterior-Posterior (AP) motion of the subject obtained from the RPM sensor was used as the ground truth and compared with the depth signal in the automatically detected region. Five measurements were taken from the same subject in supine position with the depth camera placed on the left side of the subject to mimic a clinical setup.


Results: location of troughs of the breathing signals obtained from the automatically detected region of the depth camera and the RPM sensor has an average correlation value of 0.99 (0.0011). The time taken for the depth sensor to detect a region was between 67 s and 239 s.


Conclusion: work demonstrates the feasibility of using the depth sensor in any motion management strategy in imaging to automatically determine the patient-specific optimal region for breathing monitoring.

Funding Support, Disclosures, and Conflict of Interest: This work was supported by grant #1161748 awarded through the Priority driven Collaborative Cancer Research Scheme and funded by Cancer Australia. RO would like to acknowledge the support of a Cancer Institute of NSW Career Development Fellowship.

Keywords

Radiation Therapy, Respiration, Computer Vision

Taxonomy

IM- Optical : Development (new technology and techniques)

Contact Email