New Near-Infrared Machine Learning Technique Identifies Dangerous Blood for Transfusion Safety
A team of researchers from Gannan Normal University and the Central Blood Station of Ganzhou City in Jiangxi Province, China, conducted a study using the GaiaSorter "Gaia" hyperspectral sorter to analyze plasma samples. This advanced technology extracts 254 spectral bands of near-infrared (NIR) plasma hyperspectral images, from 900 nm to 1700 nm, allowing for precise analysis of the blood's spectral characteristics (1).
Blood donation and blood donor - © kuarmungadd - stock.adobe.com
The study, titled "Detection of Chylous Plasma Based on Machine Learning and Hyperspectral Techniques," was led by Yafei Liu and his colleagues and published in the journal Applied Spectroscopy. The researchers used four different machine learning algorithms to classify the NIR hyperspectral plasma images: decision tree, Gaussian Naive Bayes (GaussianNB), perceptron, and stochastic gradient descent (1).
Transfusing chylous blood contains high levels of fat, often due to an obesity in blood donors, and poses significant health risks to transfusion patients. One of the most serious dangers to blood recipients is fat embolism, where fat droplets form clots that can block blood flow in vital areas such as the lungs, leading to respiratory distress or even death. The high fat content can also trigger severe allergic reactions, causing swelling, difficulty breathing, and a sudden drop in blood pressure, potentially resulting in life-threatening anaphylaxis. Additionally, the increased viscosity from the fat can elevate blood pressure, strain the heart, and increase the risk of heart attack or stroke (2).
Moreover, transfusing chylous blood can lead to febrile reactions, with symptoms like fever and chills, and can trigger inflammation, damaging various organs. This inflammation can reduce organ function and lead to conditions like hypoxia, where tissues receive insufficient oxygen. Given these risks, ensuring that chylous blood is not used in transfusions is critical for patient safety. The use of advanced techniques, such as hyperspectral imaging and machine learning, can help accurately identify chylous blood, reducing the risks of adverse reactions during blood transfusions. Spectroscopy has recently been recommended as a means for rapid and non-destructive analysis of blood samples in multiple applications (3)
Read More: Childhood Obesity Serum Samples Analyzed Using Fourier Transform Infrared Spectroscopy
Traditionally, identifying chylous blood has been a time-consuming process, but new research shows that hyperspectral imaging and machine learning could revolutionize the detection process.
Initially, the researchers compared the classification accuracies of these algorithms with the original data. They found that the decision tree and GaussianNB models performed best, achieving average accuracies of over 90%. The team then applied feature dimension reduction to the data, focusing on the most relevant spectral bands, which improved the accuracy of the decision tree model to 93.33% (1).
In the study's final analysis, the researchers concluded that the decision tree and GaussianNB models consistently provided the most accurate results. The feature dimension reduction method allowed them to identify seven key bands that were most effective for detecting chylous blood. These near-infrared bands were at wavelengths 962.93, 1122.37, 1338.87, 1380.21, 1558.16, 1656.56, and 1697.8 nm (1).
Overall, the study suggests that chylous plasma can be identified quickly and accurately using these methods, potentially reducing wasted blood resources and increasing the efficiency of blood donation centers. However, the researchers noted that their study used a limited number of samples, which could affect the generalizability of their results. They recommend further studies with larger datasets to optimize the model parameters and strengthen the reliability of the findings (1).
This breakthrough in hyperspectral imaging when combined with machine learning technology offers a promising solution to a critical medical challenge. With continued research and refinement, this technique could become a standard tool in blood donation centers, ensuring that chylous blood is identified and removed from the transfusion pool, thereby safeguarding patient health.
References
- Liu, Y.; Lai, J.; Hu, L. et al. Detection of Chylous Plasma Based on Machine Learning and Hyperspectral Techniques. Appl. Spectrosc. 2024, 78 (4), 365–375. doi:10.1177/00037028231214802
- Qian, Q.; Xu, W.; Li, W.; Wang, B.; Wang, L.; Zhou, Q. Accurate Detection of Chylous Blood Levels by Deep Learning. IEEE Access 2022, 10, 73988–73996. DOI: 10.1109/ACCESS.2022.3189369
- Acevedo, A. Childhood Obesity Serum Samples Analyzed Using Fourier Transform Infrared Spectroscopy. https://www.spectroscopyonline.com/view/childhood-obesity-serum-samples-analyzed-using-fourier-transform-infrared-spectrscopy (accessed 2024-05-03).
Deep Learning Advances Gas Quantification Analysis in Near-Infrared Dual-Comb Spectroscopy
May 15th 2024Researchers from Tsinghua University and Beihang University in Beijing have developed a deep-learning-based data processing framework that significantly improves the accuracy of dual-comb absorption spectroscopy (DCAS) in gas quantification analysis. By using a U-net model for etalon removal and a modified U-net combined with traditional methods for baseline extraction, their framework achieves high-fidelity absorbance spectra, even in challenging conditions with complex baselines and etalon effects.
New Probes for NIR Monitoring of Polymer Injection Molding Composition in Real-Time
May 2nd 2024Researchers from Kyoto University and Japan's National Institute of Advanced Industrial Science and Technology have developed innovative probes to monitor the chemical composition of biodegradable polymer blends during injection molding. This breakthrough could lead to improved production efficiency and reduced waste in the polymer industry.
