Machine learning-based LIBS spectrum analysis of human blood plasma allows ovarian cancer diagnosis

Modern Subtype Classification and Outlier Detection Using the Attention Embedder to Transform Ovarian Cancer Diagnosis

Ovarian cancer, a deadly female reproductive system disease, is a significant challenge in medical research due to its notorious lethality. Addressing ovarian cancer in the current medical landscape has become more complex than ever. This research explores the complex field of Ovarian Cancer Subtype Classification and the crucial task of Outlier Detection, driven by a progressive automated system, as the need to fight this unforgiving illness becomes critical. This study primarily uses a unique dataset painstakingly selected from 20 esteemed medical institutes. The dataset includes a wide range of images, such as tissue microarray (TMA) images at 40× magnification and whole-slide images (WSI) at 20× magnification. The research is fully committed to identifying abnormalities within this complex environment, going beyond the classification of subtypes of ovarian cancer. We proposed a new Attention Embedder, a state-of-the-art model with effective results in ovarian cancer subtype classification and outlier detection. Using images magnified WSI, the model demonstrated an astonishing 96.42% training accuracy and 95.10% validation accuracy. Similarly, with images magnified via a TMA, the model performed well, obtaining a validation accuracy of 94.90% and a training accuracy of 93.45%. Our fine-tuned hyperparameter testing resulted in exceptional performance on independent images. At 20× magnification, we achieved an accuracy of 93.56%. Even at 40× magnification, our testing accuracy remained high, at 91.37%. This study highlights how machine learning can revolutionize the medical field's ability to classify ovarian cancer subtypes and identify outliers, giving doctors a valuable tool to lessen the severe effects of the disease. Adopting this novel method is likely to improve the practice of medicine and give people living with ovarian cancer worldwide hope.

Blood metal analysis of plasmas from donors with and without SARS-CoV-2 using laser-induced breakdown spectroscopy and logistic regression

Research on the correlation between metal levels in blood and Covid-19 infection has been conducted primarily by assessing how each individual blood metal is linked to different aspects of the disease using samples from donors with various levels of severity to Covid-19 infection. Using logistics regression on LIBS spectra of plasma samples collected pre- and post- Covid-19 pandemic from donors known to have developed various levels of antibodies to the SARS-Cov-2 virus, we show that relying on the levels of Na, K, and Mg together is more efficient at differentiating the two types of plasma samples than any single blood alone.

Staging classification of omicron variant SARS-CoV-2 infection based on dual-spectrometer LIBS (DS-LIBS) combined with machine learning

Effective differentiation of the infection stages of omicron can provide significant assistance in transmission control and treatment strategies. The combination of LIBS serum detection and machine learning methods, as a novel disease auxiliary diagnostic approach, has a high potential for rapid and accurate staging classification of Omicron infection. However, conventional single-spectrometer LIBS serum detection methods focus on detecting the spectra of major elements, while trace elements are more closely related to the progression of COVID-19. Here, we proposed a rapid analytical method with dual-spectrometer LIBS (DS-LIBS) assisted with machine learning to classify different infection stages of omicron. The DS-LIBS, including a broadband spectrometer and a narrowband spectrometer, enables synchronous collection of major and trace elemental spectra in serum, respectively. By employing the RF machine learning models, the classification accuracy using the spectra data collected from DS-LIBS can reach 0.92, compared to 0.84 and 0.73 when using spectra data collected from single-spectrometer LIBS. This significant improvement in classification accuracy highlights the efficacy of the DS-LIBS approach. Then, the performance of four different models, SVM, RF, IGBT, and ETree, is compared. ETree demonstrates the best, with cross-validation and test set accuracies of 0.94 and 0.93, respectively. Additionally, it achieves classification accuracies of 1.00, 0.92, 0.92, and 0.89 for the four stages B1-acute, B1-post, B2, and B3. Overall, the results demonstrate that DS-LIBS combined with the ETree machine learning model enables effective staging classification of omicron infection.

Development of prediction model to estimate future risk of ovarian lesions: A multi-center retrospective study

Background

To develop the preoperative prediction of ovarian lesions using regression-based statistics analyses and machine learning methods based on multiple serological biomarkers in China.

Methods

1137 patients with ovarian lesions in Zhujiang Hospital and 518 patients in others hospital in China were randomly assigned to training, test and external validation cohorts. Five machine learning classifiers, including Random Forest (RF), Extreme Gradient Boosting (XGB), Support Vector Classifier (SVC), K-nearest Neighbor (KN), Multi-Layer Perceptron (MLP) and the Lasso-Logistics prediction model (LLRM) were used to derive diagnostic information from 23 predictors.

Results

The RF model had a high diagnostic value (AUC = 0.968) in predicting benign and malignant ovarian disease. Age and MLR were also potential diagnostic indicators for predicting ovarian disease except tumor indicators. The RF model well distinguished borderline ovarian tumors (AUC = 0.742). The RFM had a high predictive power to identify ovarian serous adenocarcinoma (AUC = 0.943) and ovarian endometriosis cysts (AUC = 0.914).

Conclusions

The RF models can effectively predict adnexal lesions, promising to be adjuncts to the preoperative prediction of ovarian cancer.

Visualization of laser-induced breakdown spectroscopy data of mouse organs based on the feature extraction method

At present, there is no comprehensive and systematic research on laser-induced breakdown spectroscopy (LIBS) data visualization. In particular, the LIBS spectra of biological samples have large noise and weak signals, which seriously affect the feature visualization. Here, three commonly used sample visualization methods were compared, and a new method was applied for tissue sample visualization. We used the LIBS mapping technique to obtain LIBS spectra of different organ slice samples from mice. LIBS spectral distribution was visualized after extracting the region of interest. The three spectral visualization methods are compared, and the performance of visualization algorithms is quantitatively analyzed. The potential of heat-diffusion for the affinity-based transition embedding (PHATE) method highlights the details of the LIBS spectral distribution while maintaining the overall structure of the data. The correlation coefficient between dimensionality reduction data and raw data is 0.97, and the average distance between samples of different categories is 0.64, which are superior to those of traditional principal component analysis (PCA), multidimensional scaling (MDS), and t-distributed stochastic neighbor embedding (t-SNE). The results show that the PHATE method can serve as a very promising LIBS spectral visualization tool.

Full-Stokes polarization laser-induced breakdown spectroscopy detection of infiltrative glioma boundary tissue

The glioma boundary is difficult to identify during surgery due to the infiltrative characteristics of tumor cells. In order to ensure a full resection rate and increase the postoperative survival of patients, it is often necessary to make an expansion range resection, which may have harmful effects on the quality of the patient's survival. A full-Stokes laser-induced breakdown spectroscopy (FSLIBS) theory with a corresponding system is proposed to combine the elemental composition information and polarization information for glioma boundary detection. To verify the elemental content of brain tissues and provide an analytical basis, inductively coupled plasma mass spectrometry (ICP-MS) and LIBS are also applied to analyze the healthy, boundary, and glioma tissues. Totally, 42 fresh tissue samples are analyzed, and the Ca, Na, K elemental lines and CN, C₂ molecular fragmental bands are proved to take an important role in the different tissue identification. The FSLIBS provides complete polarization information and elemental information than conventional LIBS elemental analysis. The Stokes parameter spectra can significantly reduce the under-fitting phenomenon of artificial intelligence identification models. Meanwhile, the FSLIBS spectral features within glioma samples are relatively more stable than boundary and healthy tissues. Other tissues may be affected obviously by individual differences in lesion positions and patients. In the future, the FSLIBS may be used for the precise identification of glioma boundaries based on polarization and elemental characterizing ability.

Comparison of whole blood and serum samples of breast cancer based on laser-induced breakdown spectroscopy with machine learning

To identify cancer from non-cancer is one of the most challenging issues nowadays in the early diagnosis of cancer. The primary issue of early detection is to choose a suitable type of sample collection to diagnose cancer. A comparison of whole blood and serum samples of breast cancer was studied using laser-induced breakdown spectroscopy (LIBS) with machine learning methods. For LIBS spectra measurement, blood samples were dropped on a substrate of boric acid. For the discrimination of breast cancer and non-cancer samples, eight machine learning models were applied to LIBS spectral data, including decision tree, discrimination analysis, logistic regression, naïve byes, support vector machine, k-nearest neighbor, ensemble and neural networks classifiers. Discrimination between whole blood samples showed that narrow neural networks and trilayer neural networks both provided 91.7% highest prediction accuracy and serum samples showed that all the decision tree models provided 89.7% highest prediction accuracy. However, using whole blood as sample achieved the strong emission lines of spectra, better discrimination results of PCA and maximum prediction accuracy of machine learning models as compared to using serum samples. These merits concluded that whole blood samples could be a good option for the rapid detection of breast cancer. This preliminary research may provide the complementary method for early detection of breast cancer.

Rapid quantitative analysis of rare earth elements Lu and Y in rare earth ores by laser induced breakdown spectroscopy combined with iPLS-VIP and partial least squares

Rare earth ores are complex in composition and diverse in mineral composition, requiring high technical requirements for the selection of rare earth ores. It is of great significance to explore the on-site rapid detection and analysis methods of rare earth elements in rare earth ores. Laser induced breakdown spectroscopy (LIBS) is an important tool to detect rare earth ores, which can be used for in situ analyses without complicated sample preparation. In this study, a rapid quantitative analysis method for rare earth elements Lu and Y in rare earth ores was established by LIBS combined with an iPLS-VIP hybrid variable selection strategy and partial least squares (PLS) method. First, the LIBS spectra of 25 samples were studied using laser induced breakdown spectrometry. Second, taking the spectrum processed by wavelet transform (WT) as the input variables, PLS calibration models based on interval partial least squares (iPLS), variable importance projection (VIP) and iPLS-VIP hybrid variable selection were constructed to quantitatively analyze rare earth elements Lu and Y, respectively. The results show that the WT-iPLS-VIP-PLS calibration model has better prediction performance for rare earth elements Lu and Y, and the optimal coefficient of determination (R²) of Lu and Y were 0.9897 and 0.9833, the root mean square error (RMSE) were 0.8150 μg g^-1 and 97.1047 μg g^-1, and the mean relative error (MRE) were 0.0754 and 0.0766, respectively. It shows that LIBS technology combined with the iPLS-VIP and PLS calibration model provides a new method for in situ quantitative analysis of rare earth elements in rare earth ores.

Stable sensing platform for diagnosing electrolyte disturbance using laser-induced breakdown spectroscopy

Electrolyte disturbance is very common and harmful, increasing the mortality of critical patients. Hence, rapid and accurate detection of electrolyte levels is vital in clinical practice. Laser-induced breakdown spectroscopy (LIBS) has the advantage of rapid and simultaneous detection of multiple elements, which meets the needs of clinical electrolyte detection. However, the cracking caused by serum drying and the effect of the coffee-ring led to the unstable spectral signal of LIBS and inaccurate detection results. Herein, we propose the ordered microarray silicon substrates (OMSS) obtained by laser microprocessing, to solve the disturbance caused by cracking and the coffee-ring effect in LIBS detection. Moreover, the area of OMSS is optimized to obtain the optimal LIBS detection effect; only a 10 uL serum sample is required. Compared with the silicon wafer substrates, the relative standard deviation (RSD) of the serum LIBS spectral reduces from above 80.00% to below 15.00% by the optimized OMSS, improving the spectral stability. Furthermore, the OMSS is combined with LIBS to form a sensing platform for electrolyte disturbance detection. A set of electrolyte disturbance simulation samples (80% of the ingredients are human serum) was prepared for this platform evaluation. Finally, the platform can achieve an accurate quantitative detection of Na and K elements (Na: RSD < 6.00%, R² = 0.991; K: RSD < 4.00%, R² = 0.981), and the detection time is within 5 min. The LIBS sensing platform has a good prospect in clinical electrolyte detection and other blood-related clinical diagnoses.

Model transfer method based on piecewise direct standardization in laser-induced-breakdown spectroscopy

A large number of certified samples are usually required to build models in the quantitative analysis of complicated matrices in laser-induced-breakdown spectroscopy (LIBS). Because of differences among instruments, including excitation and collection efficiencies, a quantitative model made on one instrument is difficult to apply directly to other instruments. Each instrument requires a large number of samples to model, which is very labor intensive and will hinder the rapid application of the LIBS technique. To eliminate the differences in spectral data from different instruments and reduce the cost of building new models, a piecewise direct standardization method combined with partial least squares (PLS_PDS) is studied in this work. Two portable LIBS instruments with the same configuration are used to obtain spectral data, one of which is called a master instrument because its calibration model is directly built on a large number of labeled samples, and the other of which is called a slave instrument because its model is obtained from the master instrument. The PLS_PDS method is used to build a transfer function of spectra between the master instrument and slave instrument to reduce the spectral difference between two instruments, and thus one calibration model can adapt to different instruments. Results show that for multiple elemental analyses of aluminium alloy samples, the number of samples required for slave modeling was reduced from 51 to 14 after model transferring by PLS_PDS, and the quantitative performance of the slave instrument was close to that of the master instrument. Therefore, the model transfer method can obviously reduce the sample number of building models for slave instruments, and it will be beneficial to advance the application of LIBS.

Evaluation of electrolyte element composition in human tissue by laser-induced breakdown spectroscopy (LIBS)

Laser-induced breakdown spectroscopy (LIBS) enables the direct measurement of cell electrolyte concentrations. The utility of LIBS spectra in biomarker studies is limited because these studies rarely consider basic physical principles. The aim of this study was to test the suitability of LIBS spectra as an analytical method for biomarker assays and to evaluate the composition of electrolyte elements in human biomaterial. LIBS as an analytical method was evaluated by establishing KCl calibration curves to demonstrate linearity, by the correct identification of emission lines with corresponding reference spectra, and by the feasibility to use LIBS in human biomaterial, analyzing striated muscle tissues from the oral regions of two patients. Lorentzian peak fit and peak area calculations resulted in better linearity and reduced shot-to-shot variance. Correct quantitative measurement allowed for differentiation of human biomaterial between patients, and determination of the concentration ratios of main electrolytes within human tissue. The clinical significance of LIBS spectra should be evaluated using peak area rather than peak intensity. LIBS might be a promising tool for analyzing a small group of living cells. Due to linearity, specificity and robustness of the proposed analytical method, LIBS could be a component of future biomarker studies.

Machine learning applications in gynecological cancer: A critical review

Machine Learning (ML) represents a computer science capable of generating predictive models, by exposure to raw, training data, without being rigidly programmed. Over the last few years, ML has gained attention within the field of oncology, with considerable strides in both diagnostic, predictive, and prognostic spectrum of malignancies, but also as a catalyst of cancer research. In this review, we discuss the state of ML applications on gynecologic oncology and systematically address major technical and ethical concerns, with respect to their real-world medical practice translation. Undoubtedly, advances in ML will enable the analysis of large, rather complex, datasets for improved, cost-effective, and efficient clinical decisions.

Analysis and Dynamic Monitoring of Indoor Air Quality Based on Laser-Induced Breakdown Spectroscopy and Machine Learning

The air quality of the living area influences human health to a certain extent. Therefore, it is particularly important to detect the quality of indoor air. However, traditional detection methods mainly depend on chemical analysis, which has long been criticized for its high time cost. In this research, a rapid air detection method for the indoor environment using laser-induced breakdown spectroscopy (LIBS) and machine learning was proposed. Four common scenes were simulated, including burning carbon, burning incense, spraying perfume and hot shower which often led to indoor air quality changes. Two steps of spectral measurements and algorithm analysis were used in the experiment. Moreover, the proposed method was found to be effective in distinguishing different kinds of aerosols and presenting sensitivity to the air compositions. In this paper, the signal was isolated by the forest, so the singular values were filtered out. Meanwhile, the spectra of different scenarios were analyzed via the principal component analysis (PCA), and the air environment was classified by K-Nearest Neighbor (KNN) algorithm with an accuracy of 99.2%. Moreover, based on the establishment of a high-precision quantitative detection model, a back propagation (BP) neural network was introduced to improve the robustness and accuracy of indoor environment. The results show that by taking this method, the dynamic prediction of elements concentration can be realized, and its recognition accuracy is 96.5%.

Metronidazole-loaded gold nanoparticles in natural rubber latex as a potential wound dressing

Gold nanoparticles (AuNPs) have shown interesting properties and specific biofunctions, providing benefits and new opportunities for controlled release systems. In this research, we demonstrated the use of natural rubber latex (NRL) from Hevea brasiliensis as a carrier of AuNPs and the antibiotic metronidazole (MET). We prepared AuNP-MET-NRL and characterized by physicochemical, biological and in vitro release assays. The effect of AuNPs on MET release was evaluated using UV-Vis and Laser-Induced Breakdown Spectroscopy (LIBS) techniques. AuNPs synthesized by Turkevich and Frens method resulted in a spherical shape with diameters of 34.8 ± 5.5 nm. We verified that there was no emergence or disappearance of new vibrational bands. Qualitatively and quantitatively, we showed that the MET crystals dispersed throughout the NRL. The Young's modulus and elongation values at dressing rupture were in the range appropriate for human skin application. 64.70% of the AuNP-MET complex was released within 100 h, exhibiting a second-order exponential release profile. The LIBS technique allowed monitoring of the AuNP release, indicating the Au emission peak reduction at 267.57 nm over time. Moreover, the dressing displayed an excellent hemocompatibility and fibroblast cell viability. These results demonstrated that the AuNP-MET-NRL wound dressing is a promising approach for dermal applications.

Screening ovarian cancers with Raman spectroscopy of blood plasma coupled with machine learning data processing

The mortality of ovarian cancer is closely related to its poor rate of early detection. In the search of an efficient diagnosis method, Raman spectroscopy of blood features as a promising technique allowing simple, rapid, minimally-invasive and cost-effective detection of cancers, in particular ovarian cancer. Although Raman spectroscopy has been demonstrated to be effective to detect ovarian cancers with respect to normal controls, a binary classification remains idealized with respect to the real clinical practice. This work considered a population of 95 woman patients initially suspected of an ovarian cancer and finally fixed with a cancer or a cyst. Additionally, 79 normal controls completed the ensemble of samples. Such sample collection proposed us a study case where a ternary classification should be realized with Raman spectroscopy of the collected blood samples coupled with suitable spectroscopic data treatment algorithms. In the medical as well as data points of view, the appearance of the cyst case considerably reduces the distances among the different populations and makes their distinction much more difficult, since the intermediate cyst case can share the specific features of the both cancer and normal cases. After a proper spectrum pretreatment, we first demonstrated the evidence of different behaviors among the Raman spectra of the 3 types of samples. Such difference was further visualized in a high dimensional space, where the data points of the cancer and the normal cases are separately clustered, whereas the data of the cyst case were scattered into the areas respectively occupied by the cancer and normal cases. We finally developed and tested an ensemble of models for a ternary classification with 2 consequent steps of binary classifications, based on machine learning algorithms, allowing identification with sensitivity and specificity of 81.0% and 97.3% for cancer samples, 63.6% and 91.5% for cyst samples, 100% and 90.6% for normal samples.

The use of laser-based diagnostics for the rapid identification of blood borne viruses in human plasma samples

AIMS

To demonstrate the use of a laser-based method of detection as a potential diagnostic test for the rapid identification of blood borne viruses in human plasma.

METHODS AND RESULTS

In this study, using light emissions from laser sparks on plasma samples, the successful differentiation of both human immunodeficiency virus (HIV) and hepatitis C virus (HCV) in both residual de-identified plasma samples and plasma samples spiked to clinically relevant levels with each virus were demonstrated using plasma from more than 20 individuals spanning six different blood types (O+, O-, A+, A-, B+, B-).

CONCLUSIONS

These experiments demonstrate that mathematical analysis of spectral data from laser sparks can provide accurate results within minutes. This capability was demonstrated using both spiked laboratory plasma samples and clinical plasma samples collected from infected and uninfected individuals.

SIGNIFICANCE AND IMPACT OF THE STUDY

There is an ongoing need to rapidly detect viral infections and to screen for multiple viral infections. A laser-based approach can achieve sensitive, multiplex detection with minimal sample preparation and provide results within minutes. These properties along with the flexibility to add new agent detection by adjusting the detection programming make it a promising tool for clinical diagnosis. The potential for a laser-based approach has been previously demonstrated using pathogens spiked into human blood to clinically relevant levels. This study demonstrates this same ability to detect infections in clinical and laboratory spiked plasma samples. The ability to differentiate between plasma samples from infected and uninfected donors and determine the virus type using a laser-based diagnostic has not been previously demonstrated. Furthermore, this study is the first demonstration of the capability to differentiate viral infections in clinical plasma samples whereas previously published work used laboratory samples spiked with a virus or dealt with the detection of cancer in clinical plasma samples.

From machine learning to transfer learning in laser-induced breakdown spectroscopy analysis of rocks for Mars exploration

With the ChemCam instrument, laser-induced breakdown spectroscopy (LIBS) has successively contributed to Mars exploration by determining the elemental compositions of soils, crusts, and rocks. The American Perseverance rover and the Chinese Zhurong rover respectively landed on Mars on February 18 and May 15, 2021, further increase the number of LIBS instruments on Mars. Such an unprecedented situation requires a reinforced research effort on the methods of LIBS spectral data analysis. Although the matrix effects correspond to a general issue in LIBS, they become accentuated in the case of rock analysis for Mars exploration, because of the large variation of rock compositions leading to the chemical matrix effect, and the difference in surface physical properties between laboratory standards (in pressed powder pellet, glass or ceramic) used to establish calibration models and natural rocks encountered on Mars, leading to the physical matrix effect. The chemical matrix effect has been tackled in the ChemCam project with large sets of laboratory standards offering a good representation of various compositions of Mars rocks. The present work more specifically deals with the physical matrix effect which is still lacking a satisfactory solution. The approach consists in introducing transfer learning in LIBS data treatment. For the specific application of total alkali-silica (TAS) classification of rocks (either with a polished surface or in the raw state), the results show a significant improvement in the ability to predict of pellet-based models when trained together with suitable information from rocks in a procedure of transfer learning. The correct TAS classification rate increases from 25% for polished rocks and 33.3% for raw rocks with a machine learning model, to 83.3% with a transfer learning model for both types of rock samples.

Ovarian cancer diagnosis using pretrained mask CNN-based segmentation with VGG-19 architecture

Objectives

This paper proposed the neural network-based segmentation model using Pre-trained Mask Convolutional Neural Network (CNN) with VGG-19 architecture. Since ovarian is very tiny tissue, it needs to be segmented with higher accuracy from the annotated image of ovary images collected in dataset. This model is proposed to predict and suppress the illness early and to correctly diagnose it, helping the doctor save the patient's life.

Methods

The paper uses the neural network based segmentation using Pre-trained Mask CNN integrated with VGG-19 NN architecture for CNN to enhance the ovarian cancer prediction and diagnosis.

Results

Proposed segmentation using hybrid neural network of CNN will provide higher accuracy when compared with logistic regression, Gaussian naïve Bayes, and random Forest and Support Vector Machine (SVM) classifiers.

Classification Framework for Healthy Hairs and Alopecia Areata: A Machine Learning (ML) Approach

Alopecia areata is defined as an autoimmune disorder that results in hair loss. The latest worldwide statistics have exhibited that alopecia areata has a prevalence of 1 in 1000 and has an incidence of 2%. Machine learning techniques have demonstrated potential in different areas of dermatology and may play a significant role in classifying alopecia areata for better prediction and diagnosis. We propose a framework pertaining to the classification of healthy hairs and alopecia areata. We used 200 images of healthy hairs from the Figaro1k dataset and 68 hair images of alopecia areata from the Dermnet dataset to undergo image preprocessing including enhancement and segmentation. This was followed by feature extraction including texture, shape, and color. Two classification techniques, i.e., support vector machine (SVM) and k-nearest neighbor (KNN), are then applied to train a machine learning model with 70% of the images. The remaining image set was used for the testing phase. With a 10-fold cross-validation, the reported accuracies of SVM and KNN are 91.4% and 88.9%, respectively. Paired sample T-test showed significant differences between the two accuracies with a p < 0.001. SVM generated higher accuracy (91.4%) as compared to KNN (88.9%). The findings of our study demonstrate potential for better prediction in the field of dermatology.