Monitoring Respiratory Motion during VMAT Treatment Delivery Using Ultra-Wideband Radar

The goal of this paper is to evaluate the potential of a low-cost, ultra-wideband radar system for detecting and monitoring respiratory motion during radiation therapy treatment delivery. Radar signals from breathing motion patterns simulated using a respiratory motion phantom were captured during volumetric modulated arc therapy (VMAT) delivery. Gantry motion causes strong interference affecting the quality of the extracted respiration motion signal. We developed an artificial neural network (ANN) model for recovering the breathing motion patterns. Next, automated classification into four classes of breathing amplitudes is performed, including no breathing, breath hold, free breathing and deep inspiration. Breathing motion patterns extracted from the radar signal are in excellent agreement with the reference data recorded by the respiratory motion phantom. The classification accuracy of simulated deep inspiration breath hold breathing was 94% under the worst case interference from gantry motion and linac operation. Ultra-wideband radar systems can achieve accurate breathing rate estimation in real-time during dynamic radiation delivery. This technology serves as a viable alternative to motion detection and respiratory gating systems based on surface detection, and is well-suited to dynamic radiation treatment techniques. Novelties of this work include detection of the breathing signal using radar during strong interference from simultaneous gantry motion, and using ANN to perform adaptive signal processing to recover breathing signal from large interference signals in real time.

Classification and Detection of Breathing Patterns with Wearable Sensors and Deep Learning

Rapid assessment of breathing patterns is important for several emergency medical situations. In this research, we developed a non-invasive breathing analysis system that automatically detects different types of breathing patterns of clinical significance. Accelerometer and gyroscopic data were collected from light-weight wireless sensors placed on the chest and abdomen of 100 normal volunteers who simulated various breathing events (central sleep apnea, coughing, obstructive sleep apnea, sighing, and yawning). We then constructed synthetic datasets by injecting annotated examples of the various patterns into segments of normal breathing. A one-dimensional convolutional neural network was implemented to detect the location of each event in each synthetic dataset and to classify it as belonging to one of the above event types. We achieved a mean F1 score of 92% for normal breathing, 87% for central sleep apnea, 72% for coughing, 51% for obstructive sleep apnea, 57% for sighing, and 63% for yawning. These results demonstrate that using deep learning to analyze chest and abdomen movement data from wearable sensors provides an unobtrusive means of monitoring the breathing pattern. This could have application in a number of critical medical situations such as detecting apneas during sleep at home and monitoring breathing events in mechanically ventilated patients in the intensive care unit.