A novel respiratory detection method based on automated analysis of ultrasound diaphragm video

Abstract

This paper proposes a novel respiratory detection method based on diaphragm motion measurements using a 2D ultrasound unit. The proposed method extracts a respiratory signal from an automated analysis of the internal diaphragm motion during breathing. The respiratory signal may be used for gating. Ultrasound studies of diaphragm breathing motion were performed on four volunteers. The ultrasound video stream was captured and transferred to a personal computer and decomposed into individual image frames. After straightforward image analysis, region of interest selection, and filtering, the mutual information (MI) and correlation coefficients (CCs) between a selected reference frame and all other frames were computed. The resulting MI and CC values were discovered to produce a signal corresponding to the respiratory cycle in both phase and magnitude. We also studied the diaphragm motion of two volunteers during repeated deep inspiration breath holds (DIBH) and found a slight relaxation motion of the diaphragm during the DIBH, suggesting that the residual motion may be important for treatments delivered at this breathing phase. Applying the proposed respiratory detection method to these ultrasound studies, we found that the MI and CC values demonstrate the relaxation behavior, indicatingthat our method may be used to determine the radiation triggering time for a DIBH technique.