fluorescence
-
KAIST Develops Wearable Carbon Dioxide Sensor to Enable Real-time Apnea Diagnosis
- Professor Seunghyup Yoo’s research team of the School of Electrical Engineering developed an ultralow-power carbon dioxide (CO2) sensor using a flexible and thin organic photodiode, and succeeded in real-time breathing monitoring by attaching it to a commercial mask
- Wearable devices with features such as low power, high stability, and flexibility can be utilized for early diagnosis of various diseases such as chronic obstructive pulmonary disease and sleep apnea
< Photo 1. From the left, School of Electrical Engineering, Ph.D. candidate DongHo Choi, Professor Seunghyup Yoo, and Department of Materials Science and Engineering, Bachelor’s candidate MinJae Kim >
Carbon dioxide (CO2) is a major respiratory metabolite, and continuous monitoring of CO2 concentration in exhaled breath is not only an important indicator for early detection and diagnosis of respiratory and circulatory system diseases, but can also be widely used for monitoring personal exercise status. KAIST researchers succeeded in accurately measuring CO2 concentration by attaching it to the inside of a mask.
KAIST (President Kwang-Hyung Lee) announced on February 10th that Professor Seunghyup Yoo's research team in the Department of Electrical and Electronic Engineering developed a low-power, high-speed wearable CO2 sensor capable of stable breathing monitoring in real time.
Existing non-invasive CO2 sensors had limitations in that they were large in size and consumed high power. In particular, optochemical CO2 sensors using fluorescent molecules have the advantage of being miniaturized and lightweight, but due to the photodegradation phenomenon of dye molecules, they are difficult to use stably for a long time, which limits their use as wearable healthcare sensors.
Optochemical CO2 sensors utilize the fact that the intensity of fluorescence emitted from fluorescent molecules decreases depending on the concentration of CO2, and it is important to effectively detect changes in fluorescence light.
To this end, the research team developed a low-power CO2 sensor consisting of an LED and an organic photodiode surrounding it. Based on high light collection efficiency, the sensor, which minimizes the amount of excitation light irradiated on fluorescent molecules, achieved a device power consumption of 171 μW, which is tens of times lower than existing sensors that consume several mW.
< Figure 1. Structure and operating principle of the developed optochemical carbon dioxide (CO2) sensor. Light emitted from the LED is converted into fluorescence through the fluorescent film, reflected from the light scattering layer, and incident on the organic photodiode. CO2 reacts with a small amount of water inside the fluorescent film to form carbonic acid (H2CO3), which increases the concentration of hydrogen ions (H+), and the fluorescence intensity due to 470 nm excitation light decreases. The circular organic photodiode with high light collection efficiency effectively detects changes in fluorescence intensity, lowers the power required light up the LED, and reduces light-induced deterioration. >
The research team also elucidated the photodegradation path of fluorescent molecules used in CO2 sensors, revealed the cause of the increase in error over time in photochemical sensors, and suggested an optical design method to suppress the occurrence of errors.
Based on this, the research team developed a sensor that effectively reduces errors caused by photodegradation, which was a chronic problem of existing photochemical sensors, and can be used continuously for up to 9 hours while existing technologies based on the same material can be used for less than 20 minutes, and can be used multiple times when replacing the CO2 detection fluorescent film.
< Figure 2. Wearable smart mask and real-time breathing monitoring. The fabricated sensor module consists of four elements (①: gas-permeable light-scattering layer, ②: color filter and organic photodiode, ③: light-emitting diode, ④: CO2-detecting fluorescent film). The thin and light sensor (D1: 400 nm, D2: 470 nm) is attached to the inside of the mask to monitor the wearer's breathing in real time. >
The developed sensor accurately measured CO2 concentration by being attached to the inside of a mask based on the advantages of being light (0.12 g), thin (0.7 mm), and flexible. In addition, it showed fast speed and high resolution that can monitor respiratory rate by distinguishing between inhalation and exhalation in real time.
< Photo 2. The developed sensor attached to the inside of the mask >
Professor Seunghyup Yoo said, "The developed sensor has excellent characteristics such as low power, high stability, and flexibility, so it can be widely applied to wearable devices, and can be used for the early diagnosis of various diseases such as hypercapnia, chronic obstructive pulmonary disease, and sleep apnea." He added, "In particular, it is expected to be used to improve side effects caused by rebreathing in environments where dust is generated or where masks are worn for long periods of time, such as during seasonal changes."
This study, in which KAIST's Department of Materials Science and Engineering's undergraduate student Minjae Kim and School of Electrical Engineering's doctoral student Dongho Choi participated as joint first authors, was published in the online version of Cell's sister journal, Device, on the 22nd of last month. (Paper title: Ultralow-power carbon dioxide sensor for real-time breath monitoring) DOI: https://doi.org/10.1016/j.device.2024.100681
< Photo 3. From the left, Professor Seunghyup Yoo of the School of Electrical Engineering, MinJae Kim, an undergraduate student in the Department of Materials Science and Engineering, and Dongho Choi, a doctoral student in the School of Electrical Engineering >
This study was supported by the Ministry of Trade, Industry and Energy's Materials and Components Technology Development Project, the National Research Foundation of Korea's Original Technology Development Project, and the KAIST Undergraduate Research Participation Project. This work was supported by the (URP) program.
2025.02.13 View 2482 -
Label-Free Multiplexed Microtomography of Endogenous Subcellular Dynamics Using Deep Learning
AI-based holographic microscopy allows molecular imaging without introducing exogenous labeling agents
A research team upgraded the 3D microtomography observing dynamics of label-free live cells in multiplexed fluorescence imaging. The AI-powered 3D holotomographic microscopy extracts various molecular information from live unlabeled biological cells in real time without exogenous labeling or staining agents.
Professor YongKeum Park’s team and the startup Tomocube encoded 3D refractive index tomograms using the refractive index as a means of measurement. Then they decoded the information with a deep learning-based model that infers multiple 3D fluorescence tomograms from the refractive index measurements of the corresponding subcellular targets, thereby achieving multiplexed micro tomography. This study was reported in Nature Cell Biology online on December 7, 2021.
Fluorescence microscopy is the most widely used optical microscopy technique due to its high biochemical specificity. However, it needs to genetically manipulate or to stain cells with fluorescent labels in order to express fluorescent proteins. These labeling processes inevitably affect the intrinsic physiology of cells. It also has challenges in long-term measuring due to photobleaching and phototoxicity. The overlapped spectra of multiplexed fluorescence signals also hinder the viewing of various structures at the same time. More critically, it took several hours to observe the cells after preparing them.
3D holographic microscopy, also known as holotomography, is providing new ways to quantitatively image live cells without pretreatments such as staining. Holotomography can accurately and quickly measure the morphological and structural information of cells, but only provides limited biochemical and molecular information.
The 'AI microscope' created in this process takes advantage of the features of both holographic microscopy and fluorescence microscopy. That is, a specific image from a fluorescence microscope can be obtained without a fluorescent label. Therefore, the microscope can observe many types of cellular structures in their natural state in 3D and at the same time as fast as one millisecond, and long-term measurements over several days are also possible.
The Tomocube-KAIST team showed that fluorescence images can be directly and precisely predicted from holotomographic images in various cells and conditions. Using the quantitative relationship between the spatial distribution of the refractive index found by AI and the major structures in cells, it was possible to decipher the spatial distribution of the refractive index. And surprisingly, it confirmed that this relationship is constant regardless of cell type.
Professor Park said, “We were able to develop a new concept microscope that combines the advantages of several microscopes with the multidisciplinary research of AI, optics, and biology. It will be immediately applicable for new types of cells not included in the existing data and is expected to be widely applicable for various biological and medical research.”
When comparing the molecular image information extracted by AI with the molecular image information physically obtained by fluorescence staining in 3D space, it showed a 97% or more conformity, which is a level that is difficult to distinguish with the naked eye.
“Compared to the sub-60% accuracy of the fluorescence information extracted from the model developed by the Google AI team, it showed significantly higher performance,” Professor Park added.
This work was supported by the KAIST Up program, the BK21+ program, Tomocube, the National Research Foundation of Korea, and the Ministry of Science and ICT, and the Ministry of Health & Welfare.
-Publication
Hyun-seok Min, Won-Do Heo, YongKeun Park, et al. “Label-free multiplexed microtomography of endogenous subcellular dynamics using generalizable deep learning,” Nature Cell Biology (doi.org/10.1038/s41556-021-00802-x) published online December 07 2021.
-Profile
Professor YongKeun Park
Biomedical Optics Laboratory
Department of Physics
KAIST
2022.02.09 View 10008