Pereira Carina B, Yu Xinchi, Blazek Vladimir, Leonhardt Steffen
Annu Int Conf IEEE Eng Med Biol Soc. 2015;2015:4250-3. doi: 10.1109/EMBC.2015.7319333.
An abnormal breathing rate (BR) is one of the strongest markers of physiological distress. Moreover, it plays an important role in early detection of sudden infant death syndrome, as well as in the diagnosis of respiratory disorders. However, the current measuring modalities can cause discomfort to the patient, since attachment to the patient's body is required. This paper proposes a new approach based on infrared thermography to remotely monitor BR. This method allows to (1) detect automatically the nose, (2) track the associate region of interest (ROI), and (3) extract BR. To evaluate the performance of this method, thermal recording of 5 healthy subjects were acquired. Results were compared with BR obtained by capnography. The introduced approach demonstrated an excellent performance. ROIs were precisely segmented and tracked. Furthermore, a Bland-Altman diagram showed a good agreement between estimated BR and gold standard. The mean correlation and mean absolute BR error are 0.92 ± 0.07 and 0.53 bpm, respectively. In summary, infrared thermography seems to be a great, clinically relevant alternative to attached sensors, due to its outstanding characteristics and performance.
异常呼吸频率（BR）是生理窘迫的最强指标之一。此外，它在婴儿猝死综合征的早期检测以及呼吸系统疾病的诊断中起着重要作用。然而，当前的测量方式会给患者带来不适，因为需要附着在患者身体上。本文提出了一种基于红外热成像的新方法来远程监测呼吸频率。该方法能够（1）自动检测鼻子，（2）跟踪相关的感兴趣区域（ROI），以及（3）提取呼吸频率。为了评估该方法的性能，采集了5名健康受试者的热记录。将结果与通过二氧化碳描记法获得的呼吸频率进行比较。所介绍的方法表现出优异的性能。感兴趣区域被精确分割和跟踪。此外，布兰德 - 奥特曼图显示估计的呼吸频率与金标准之间具有良好的一致性。平均相关性和平均绝对呼吸频率误差分别为0.92±0.07和0.53次/分钟。总之，由于其出色的特性和性能，红外热成像似乎是附着式传感器的一种非常好的、与临床相关的替代方法。
