An eXplainable Artificial Intelligence analysis of Raman spectra for thyroid cancer diagnosis

Integrating Raman spectroscopy and RT-qPCR for enhanced diagnosis of thyroid lesions: A comparative study of biochemical and molecular markers

Thyroid cancer is the most prevalent endocrine malignancy, with increasing incidence due to advancements in diagnostic techniques. Ultrasound (US) and fine needle aspiration (FNA) cytology, widely used in clinical practice, have detection accuracies ranging from 65 % to 95 %. However, these methods may yield inconclusive or difficult-to-interpret results, emphasizing the need for complementary diagnostic techniques. This study explores the integration of Raman spectroscopy and gene expression analysis via RT-qPCR to improve the diagnosis of thyroid lesions, classified into groups: follicular thyroid carcinoma (FTC), papillary thyroid carcinoma (PTC) and goiter tissues. Healthy tissue samples were used as normalizing controls in both analysis. Raman spectroscopy analyzed 35 samples, while RT-qPCR assessed 33 samples. For comparison, the same 19 samples previously analyzed by both techniques were examined. Raman spectroscopy, a non-invasive technique, has shown effectiveness in distinguishing between benign and malignant thyroid tissues by identifying key biochemical components such as DNA, RNA, proteins, and lipids. The distinguishing of FTC from goiter using Raman spectroscopy achieved an accuracy rate of 82.3 %. Gene expression analysis via RT-qPCR focused on six genes: TG, TPO, PDGFB, SERPINA1, TFF3, and LGALS3. Specifically, SERPINA1 was overexpressed in PTC, TFF3 showed elevated levels in FTC, and LGALS3 was elevated in both PTC and FTC compared to goiter and normal tissues. These findings align with existing literature, suggesting that these genes could serve as valuable diagnostic molecular markers. The expression analysis of these genes within this subset of samples demonstrated concordance with the classification derived from PCA of Raman spectroscopy data. The integration of Raman spectroscopy and RT-qPCR offers a complementary approach to traditional histological analysis, providing enhanced sensitivity and specificity in diagnosing thyroid lesions.

Toward cancer detection by label-free microscopic imaging in oncological surgery: Techniques, instrumentation and applications

This review examines the clinical application of label-free microscopy and spectroscopy, which are based on optical signals emitted by tissue components. Over the past three decades, a variety of techniques have been investigated with the aim of developing an in situ histopathology method that can rapidly and accurately identify tumor margins during surgical procedures. These techniques can be divided into two groups. One group encompasses techniques exploiting linear optical signals, and includes infrared and Raman microspectroscopy, and autofluorescence microscopy. The second group includes techniques based on nonlinear optical signals, including harmonic generation, coherent Raman scattering, and multiphoton autofluorescence microscopy. Some of these methods provide comparable information, while others are complementary. However, all of them have distinct advantages and disadvantages due to their inherent nature. The first part of the review provides an explanation of the underlying physics of the excitation mechanisms and a description of the instrumentation. It also covers endomicroscopy and data analysis, which are important for understanding the current limitations in implementing label-free techniques in clinical settings. The second part of the review describes the application of label-free microscopy imaging to improve oncological surgery with focus on brain tumors and selected gastrointestinal cancers, and provides a critical assessment of the current state of translation of these methods into clinical practice. Finally, the potential of confocal laser endomicroscopy for the acquisition of autofluorescence is discussed in the context of immediate clinical applications.

Comprehensive Raman Fingerprinting and Machine Learning-Based Classification of 14 Pesticides Using a 785 nm Custom Raman Instrument

Raman spectroscopy enables fast, label-free, qualitative, and quantitative observation of the physical and chemical properties of various substances. Here, we present a 785 nm custom-built Raman spectroscopy instrument designed for sensing applications in the 400-1700 cm-1 spectral range. We demonstrate the performance of the instrument by fingerprinting 14 pesticide reference samples with over twenty technical repeats per sample. We present molecular Raman fingerprints of the pesticides comprehensively and distinguish similarities and differences among them using multivariate analysis and machine learning techniques. The same pesticides were additionally investigated using a commercial 532 nm Raman instrument to see the potential variations in peak shifts and intensities. We developed a unique Raman fingerprint library for 14 reference pesticides, which is comprehensively documented in this study for the first time. The comparison shows the importance of selecting an appropriate excitation wavelength based on the target analyte. While 532 nm may be advantageous for certain compounds due to resonance enhancement, 785 nm is generally more effective for reducing fluorescence and achieving clearer Raman spectra. By employing machine learning techniques like the Random Forest Classifier, the study automates the classification of 14 different pesticides, streamlining data interpretation for non-experts. Applying such combined techniques to a wider range of agricultural chemicals, clinical biomarkers, or pollutants could provide an impetus to develop monitoring technologies in food safety, diagnostics, and cross-industry quality control applications.

The role of explainable artificial intelligence in disease prediction: a systematic literature review and future research directions

Explainable Artificial Intelligence (XAI) enhances transparency and interpretability in AI models, which is crucial for trust and accountability in healthcare. A potential application of XAI is disease prediction using various data modalities. This study conducts a Systematic Literature Review (SLR) following the PRISMA protocol, synthesizing findings from 30 selected studies to examine XAI's evolving role in disease prediction. It explores commonly used XAI methods, such as Shapley Additive Explanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), and their impact across medical fields in disease prediction. The review highlights key gaps, including limited dataset diversity, model complexity, and reliance on single data types, emphasizing the need for greater interpretability and data integration. Addressing these issues is crucial for advancing AI in healthcare. This study contributes by outlining current challenges and potential solutions, suggesting directions for future research to develop more reliable and robust XAI methods.

Could Raman spectroscopy investigate the changes of cell oxidative stress status in thyroid diseases? A pilot study on cytological samples

The incidence of thyroid nodules is rapidly increasing worldwide. Raman spectroscopy (RS) is a powerful label-free and non-invasive technique, successfully used for early stage diagnosis. Here, RS is proposed as a tool to investigate the thyroid disease, including neoplasms, through the study of cell oxidative stress (OS), which represents one of the main cancer risk factors. In this study, we enrolled 28 patients, submitted to a first and second thyroid fine needle aspiration (FNA) during follow up. The cytological samples were studied by RS and morphological examination. Typical Raman spectra of thyroid cytological samples are reported and the contribution of oxidized and reduced cytochrome b and c and carotenoids are discussed. On the basis of the evolution of the Raman features over the time lapse between the two FNAs, the 28 patients have been classified into 4 different categories and the most representative case for each category is reported and discussed in detail. For each category, the different Raman intensity ratio between oxidized and reduced cytochromes b and c is reported and associated to different cell OS status, along with the presence of carotenoids. Overall, our results support a correlation among changes in oxidative stress, carotenoids uptake and thyroid diseases, which could inspire new fundamental research on biomarkers and signaling pathways involved in thyroid OS.

Decoding the black box: Explainable AI (XAI) for cancer diagnosis, prognosis, and treatment planning-A state-of-the art systematic review

OBJECTIVE

Explainable Artificial Intelligence (XAI) is increasingly recognized as a crucial tool in cancer care, with significant potential to enhance diagnosis, prognosis, and treatment planning. However, the holistic integration of XAI across all stages of cancer care remains underexplored. This review addresses this gap by systematically evaluating the role of XAI in these critical areas, identifying key challenges and emerging trends.

MATERIALS AND METHODS

Following the PRISMA guidelines, a comprehensive literature search was conducted across Scopus and Web of Science, focusing on publications from January 2020 to May 2024. After rigorous screening and quality assessment, 69 studies were selected for in-depth analysis.

RESULTS

The review identified critical gaps in the application of XAI within cancer care, notably the exclusion of clinicians in 83% of studies, which raises concerns about real-world applicability and may lead to explanations that are technically sound but clinically irrelevant. Additionally, 87% of studies lacked rigorous evaluation of XAI explanations, compromising their reliability in clinical practice. The dominance of post-hoc visual methods like SHAP, LIME and Grad-CAM reflects a trend toward explanations that may be inherently flawed due to specific input perturbations and simplifying assumptions. The lack of formal evaluation metrics and standardization constrains broader XAI adoption in clinical settings, creating a disconnect between AI development and clinical integration. Moreover, translating XAI insights into actionable clinical decisions remains challenging due to the absence of clear guidelines for integrating these tools into clinical workflows.

CONCLUSION

This review highlights the need for greater clinician involvement, standardized XAI evaluation metrics, clinician-centric interfaces, context-aware XAI systems, and frameworks for integrating XAI into clinical workflows for informed clinical decision-making and improved outcomes in cancer care.

Raman spectroscopy for esophageal tumor diagnosis and delineation using machine learning and the portable Raman spectrometer

Esophageal cancer is one of the leading causes of cancer-related deaths worldwide. The identification of residual tumor tissues in the surgical margin of esophageal cancer is essential for the treatment and prognosis of cancer patients. But the current diagnostic methods, either pathological frozen section or paraffin section examination, are laborious, time-consuming, and inconvenient. Raman spectroscopy is a label-free and non-invasive analytical technique that provides molecular information with high specificity. Here, we report the use of a portable Raman system and machine learning algorithms to achieve accurate diagnosis of esophageal tumor tissue in surgically resected specimens. We tested five machine learning-based classification methods, including k-Nearest Neighbors, Adaptive Boosting, Random Forest, Principal Component Analysis-Linear Discriminant Analysis, and Support Vector Machine (SVM). Among them, SVM shows the highest accuracy (88.61 %) in classifying the esophageal tumor and normal tissues. The portable Raman system demonstrates robust measurements with an acceptable focal plane shift of up to 3 mm, which enables large-area Raman mapping on resected tissues. Based on this, we finally achieve successful Raman visualization of tumor boundaries on surgical margin specimens, and the Raman measurement time is less than 5 min. This work provides a robust, convenient, accurate, and cost-effective tool for the diagnosis of esophageal cancer tumors, advancing toward Raman-based clinical intraoperative applications.

Label-Free Surface-Enhanced Raman Spectroscopy with Machine Learning for the Diagnosis of Thyroid Cancer by Using Fine-Needle Aspiration Liquid Samples

The incidence of thyroid cancer is increasing worldwide. Fine-needle aspiration (FNA) cytology is widely applied with the use of extracted biological cell samples, but current FNA cytology is labor-intensive, time-consuming, and can lead to the risk of false-negative results. Surface-enhanced Raman spectroscopy (SERS) combined with machine learning algorithms holds promise for cancer diagnosis. In this study, we develop a label-free SERS liquid biopsy method with machine learning for the rapid and accurate diagnosis of thyroid cancer by using thyroid FNA washout fluids. These liquid supernatants are mixed with silver nanoparticle colloids, and dispersed in quartz capillary for SERS measurements to discriminate between healthy and malignant samples. We collect Raman spectra of 36 thyroid FNA samples (18 malignant and 18 benign) and compare four classification models: Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA), Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN). The results show that the CNN algorithm is the most precise, with a high accuracy of 88.1%, sensitivity of 87.8%, and the area under the receiver operating characteristic curve of 0.953. Our approach is simple, convenient, and cost-effective. This study indicates that label-free SERS liquid biopsy assisted by deep learning models holds great promise for the early detection and screening of thyroid cancer.

Combined SERS-Raman screening of HER2-overexpressing or silenced breast cancer cell lines

BACKGROUND

Breast cancer (BC) is a heterogeneous neoplasm characterized by several subtypes. One of the most aggressive with high metastasis rates presents overexpression of the human epidermal growth factor receptor 2 (HER2). A quantitative evaluation of HER2 levels is essential for a correct diagnosis, selection of the most appropriate therapeutic strategy and monitoring the response to therapy.

RESULTS

In this paper, we propose the synergistic use of SERS and Raman technologies for the identification of HER2 expressing cells and its accurate assessment. To this end, we selected SKBR3 and MDA-MB-468 breast cancer cell lines, which have the highest and lowest HER2 expression, respectively, and MCF10A, a non-tumorigenic cell line from normal breast epithelium for comparison. The combined approach provides a quantitative estimate of HER2 expression and visualization of its distribution on the membrane at single cell level, clearly identifying cancer cells. Moreover, it provides a more comprehensive picture of the investigated cells disclosing a metabolic signature represented by an elevated content of proteins and aromatic amino acids. We further support these data by silencing the HER2 gene in SKBR3 cells, using the RNA interference technology, generating stable clones further analysed with the same combined methodology. Significant changes in HER2 expression are detected at single cell level before and after HER2 silencing and the HER2 status correlates with variations of fatty acids and downstream signalling molecule contents in the context of the general metabolic rewiring occurring in cancer cells. Specifically, HER2 silencing does reduce the growth ability but not the lipid metabolism that, instead, increases, suggesting that higher fatty acids biosynthesis and metabolism can occur independently of the proliferating potential tied to HER2 overexpression.

CONCLUSIONS

Our results clearly demonstrate the efficacy of the combined SERS and Raman approach to definitely pose a correct diagnosis, further supported by the data obtained by the HER2 gene silencing. Furthermore, they pave the way to a new approach to monitor the efficacy of pharmacologic treatments with the aim to tailor personalized therapies and optimize patients' outcome.

Explainable artificial intelligence and microbiome data for food geographical origin: the Mozzarella di Bufala Campana PDO Case of Study

Identifying the origin of a food product holds paramount importance in ensuring food safety, quality, and authenticity. Knowing where a food item comes from provides crucial information about its production methods, handling practices, and potential exposure to contaminants. Machine learning techniques play a pivotal role in this process by enabling the analysis of complex data sets to uncover patterns and associations that can reveal the geographical source of a food item. This study aims to investigate the potential use of explainable artificial intelligence for identifying the food origin. The case of study of Mozzarella di Bufala Campana PDO has been considered by examining the composition of the microbiota in each samples. Three different supervised machine learning algorithms have been compared and the best classifier model is represented by Random Forest with an Area Under the Curve (AUC) value of 0.93 and the top accuracy of 0.87. Machine learning models effectively classify origin, offering innovative ways to authenticate regional products and support local economies. Further research can explore microbiota analysis and extend applicability to diverse food products and contexts for enhanced accuracy and broader impact.

Explainable artificial intelligence for microbiome data analysis in colorectal cancer biomarker identification

Background

Colorectal cancer (CRC) is a type of tumor caused by the uncontrolled growth of cells in the mucosa lining the last part of the intestine. Emerging evidence underscores an association between CRC and gut microbiome dysbiosis. The high mortality rate of this cancer has made it necessary to develop new early diagnostic methods. Machine learning (ML) techniques can represent a solution to evaluate the interaction between intestinal microbiota and host physiology. Through explained artificial intelligence (XAI) it is possible to evaluate the individual contributions of microbial taxonomic markers for each subject. Our work also implements the Shapley Method Additive Explanations (SHAP) algorithm to identify for each subject which parameters are important in the context of CRC.

Results

The proposed study aimed to implement an explainable artificial intelligence framework using both gut microbiota data and demographic information from subjects to classify a cohort of control subjects from those with CRC. Our analysis revealed an association between gut microbiota and this disease. We compared three machine learning algorithms, and the Random Forest (RF) algorithm emerged as the best classifier, with a precision of 0.729 ± 0.038 and an area under the Precision-Recall curve of 0.668 ± 0.016. Additionally, SHAP analysis highlighted the most crucial variables in the model's decision-making, facilitating the identification of specific bacteria linked to CRC. Our results confirmed the role of certain bacteria, such as Fusobacterium, Peptostreptococcus, and Parvimonas, whose abundance appears notably associated with the disease, as well as bacteria whose presence is linked to a non-diseased state.

Discussion

These findings emphasizes the potential of leveraging gut microbiota data within an explainable AI framework for CRC classification. The significant association observed aligns with existing knowledge. The precision exhibited by the RF algorithm reinforces its suitability for such classification tasks. The SHAP analysis not only enhanced interpretability but identified specific bacteria crucial in CRC determination. This approach opens avenues for targeted interventions based on microbial signatures. Further exploration is warranted to deepen our understanding of the intricate interplay between microbiota and health, providing insights for refined diagnostic and therapeutic strategies.

Unraveling the microbiome-metabolome nexus: a comprehensive study protocol for personalized management of Behçet's disease using explainable artificial intelligence

The presented study protocol outlines a comprehensive investigation into the interplay among the human microbiota, volatilome, and disease biomarkers, with a specific focus on Behçet's disease (BD) using methods based on explainable artificial intelligence. The protocol is structured in three phases. During the initial three-month clinical study, participants will be divided into control and experimental groups. The experimental groups will receive a soluble fiber-based dietary supplement alongside standard therapy. Data collection will encompass oral and fecal microbiota, breath samples, clinical characteristics, laboratory parameters, and dietary habits. The subsequent biological data analysis will involve gas chromatography, mass spectrometry, and metagenetic analysis to examine the volatilome and microbiota composition of salivary and fecal samples. Additionally, chemical characterization of breath samples will be performed. The third phase introduces Explainable Artificial Intelligence (XAI) for the analysis of the collected data. This novel approach aims to evaluate eubiosis and dysbiosis conditions, identify markers associated with BD, dietary habits, and the supplement. Primary objectives include establishing correlations between microbiota, volatilome, phenotypic BD characteristics, and identifying patient groups with shared features. The study aims to identify taxonomic units and metabolic markers predicting clinical outcomes, assess the supplement's impact, and investigate the relationship between dietary habits and patient outcomes. This protocol contributes to understanding the microbiome's role in health and disease and pioneers an XAI-driven approach for personalized BD management. With 70 recruited BD patients, XAI algorithms will analyze multi-modal clinical data, potentially revolutionizing BD management and paving the way for improved patient outcomes.