Research on Multiple Spectral Ranges with Deep Learning for SpO2 Measurement.

Oxyhemoglobin saturation by pulse oximetry (SpO2) has always played an important role in the diagnosis of symptoms. Considering that the traditional SpO2 measurement has a certain error due to the number of wavelengths and the algorithm and the wider application of machine learning and spectrum combination, we propose to use 12-wavelength spectral absorption measurement to improve the accuracy of SpO2 measurement. To investigate the multiple spectral regions for deep learning for SpO2 measurement, three datasets for training and verification were built, which were constructed over the spectra of first region, second region, and full region and their sub-regions, respectively. For each region under the procedures of optimization of our model, a thorough of investigation of hyperparameters is proceeded. Additionally, data augmentation is preformed to expand dataset with added noise randomly, increasing the diversity of data and improving the generalization of the neural network. After that, the established dataset is input to a one dimensional convolution neural network (1D-CNN) to obtain a measurement model of SpO2. In order to enhance the model accuracy, GridSearchCV and Bayesian optimization are applied to optimize the hyperparameters. The optimal accuracies of proposed model optimized by GridSearchCV and Bayesian Optimization is 89.3% and 99.4%, respectively, and trained with the dataset at the spectral region of six wavelengths including 650 nm, 680 nm, 730 nm, 760 nm, 810 nm, 860 nm. The total relative error of the best model is only 0.46%, optimized by Bayesian optimization. Although the spectral measurement with more features can improve the resolution ability of the neural network, the results reveal that the training with the dataset of the shorter six wavelength is redundant. This analysis shows that it is very important to construct an effective 1D-CNN model area for spectral measurement using the appropriate spectral ranges and number of wavelengths. It shows that our proposed 1D-CNN model gives a new and feasible approach to measure SpO2 based on multi-wavelength.

Enhanced Infrared Absorbance of the CMOS Compatible Thermopile by the Subwavelength Rectangular-Hole Arrays.

The enhanced infrared absorbance (IRA) of the complementary metal-oxide-semiconductor (CMOS) compatible thermopile with the subwavelength rectangular-hole arrays in active area is investigated. The finite-difference time-domain (FDTD) method considered and analyzed the matrix arrangement (MA) and staggered arrangement (SA) of subwavelength rectangular-hole arrays (SRHA). For the better cases of MA-SRHA and SA-SRHA, the geometric parameters are the same and the infrared absorption efficiency (IAE) of the SA type is better than that of the MA type by about 19.4% at target temperature of 60˚C. Three proposed thermopiles with SA-SRHA are manufactured based on the 0.35 µm 2P4M CMOS-MEMS process. The measurement results are similar to the simulation results. The IAE of the best simulation case of SA-SRHA is up to 3.3 times higher than that without structure at the target temperature of 60˚C. Obviously, the staggered rectangular-hole arrays with more appropriate geometric conditions obtained from FDTD simulation can excellently enhance the IRA of the CMOS compatible thermopile.

A Thermopile Device with Sub-Wavelength Hole Arrays by CMOS-MEMS Technology.

A thermopile device with sub-wavelength hole array (SHA) is numerically and experimentally investigated. The infrared absorbance (IRA) effect of SHAs in active area of the thermopile device is clearly analyzed by the finite-difference time-domain (FDTD) method. The prototypes are manufactured by the 0.35 µm 2P4M complementary metal-oxide-semiconductor micro-electro-mechanical-systems (CMOS-MEMS) process in Taiwan semiconductor manufacturing company (TSMC). The measurement results of those prototypes are similar to their simulation results. Based on the simulation technology, more sub-wavelength hole structural effects for IRA of such thermopile device are discussed. It is found from simulation results that the results of SHAs arranged in a hexagonal shape are significantly better than the results of SHAs arranged in a square and the infrared absorption efficiencies (IAEs) of specific asymmetric rectangle and elliptical hole structure arrays are higher than the relatively symmetric square and circular hole structure arrays. The overall best results are respectively up to 3.532 and 3.573 times higher than that without sub-wavelength structure at the target temperature of 60 °C when the minimum structure line width limit of the process is ignored. Obviously, the IRA can be enhanced when the SHAs are considered in active area of the thermopile device and the structural optimization of the SHAs is absolutely necessary.

A Novel Infrared Temperature Measurement with Dual Mode Modulation of Thermopile Sensor.

Superior to the traditional infrared temperature sensing architecture including infrared sensor and thermistor, we propose a novel sensing approach based on a single thermopile sensor with dual modes modulation. A switching and sensing circuit is proposed and realized with a chopper amplifier AD8551 and p-channel MOSFET (PMOS) for switching between detection of thermal radiation and the target and the ambient temperature for compensation. The error of target temperature after temperature compensation is estimated at less than 0.14 °C.