Biophysical basis of skin cancer margin assessment using Raman spectroscopy

Hyperspectral imaging with machine learning for in vivo skin carcinoma margin assessment: a preliminary study

Surgical excision is the most effective treatment of skin carcinomas (basal cell carcinoma or squamous cell carcinoma). Preoperative assessment of tumoral margins plays a decisive role for a successful result. The aim of this work was to evaluate the possibility that hyperspectral imaging could become a valuable tool in solving this problem. Hyperspectral images of 11 histologically diagnosed carcinomas (six basal cell carcinomas and five squamous cell carcinomas) were acquired prior clinical evaluation and surgical excision. The hyperspectral data were then analyzed using a newly developed method for delineating skin cancer tumor margins. This proposed method is based on a segmentation process of the hyperspectral images into regions with similar spectral and spatial features, followed by a machine learning-based data classification process resulting in the generation of classification maps illustrating tumor margins. The Spectral Angle Mapper classifier was used in the data classification process using approximately 37% of the segments as the training sample, the rest being used for testing. The receiver operating characteristic was used as the method for evaluating the performance of the proposed method and the area under the curve as a metric. The results revealed that the performance of the method was very good, with median AUC values of 0.8014 for SCCs, 0.8924 for BCCs, and 0.8930 for normal skin. With AUC values above 0.89 for all types of tissue, the method was considered to have performed very well. In conclusion, hyperspectral imaging can become an objective aid in the preoperative evaluation of carcinoma margins.

Portable System for In-Clinic Differentiation of Skin Cancers from Benign Skin Lesions and Inflammatory Dermatoses

The exquisite sensitivity of Raman spectroscopy for detecting biomolecular changes in skin cancer has previously been explored; however, this mostly required analysis of excised tissue samples using bulky, immobile laboratory instrumentation. In this study, the technique was translated for clinical use with a portable Raman system and customized fiber optic probe and applied to differentiation of skin cancers from benign lesions and inflammatory dermatoses. The aim was to provide an easy-to-use, easy-to-manage assessment tool for clinicians to use in their daily patient examination routine to perform rapid Raman measurements of skin lesions in vivo. Using this system, >867 spectra were measured in vivo from 330 patients with a wide variety of different benign skin lesions (n = 603), inflammatory dermatoses (n = 140), and skin cancers (n = 124). Ethnicities represented were 70% European; 16% Asian; 6% Māori; 5% Pacific people; and 4% Middle East, Latin American, and African. Accurate differentiation of skin cancers from benign lesions and inflammatory dermatoses was achieved using partial least squares discriminant analysis, with area under curve for the receiver operator curves for external validation sets ranging from 0.916 to 0.958. This study shows evidence for robust clinical translation of Raman spectroscopy for rapid, accurate diagnosis of skin cancer.

Cutaneous Oncology: Strategies for Melanoma Prevention, Diagnosis, and Therapy

Skin cancer comprises one-third of all diagnosed cancer cases and remains a major health concern. Genetic and environmental parameters serve as the two main risk factors associated with the development of skin cancer, with ultraviolet radiation being the most common environmental risk factor. Studies have also found fair complexion, arsenic toxicity, indoor tanning, and family history among the prevailing causes of skin cancer. Prevention and early diagnosis play a crucial role in reducing the frequency and ensuring effective management of skin cancer. Recent studies have focused on exploring minimally invasive or non-invasive diagnostic technologies along with artificial intelligence to facilitate rapid and accurate diagnosis. The treatment of skin cancer ranges from traditional surgical excision to various advanced methods such as phototherapy, radiotherapy, immunotherapy, targeted therapy, and combination therapy. Recent studies have focused on immunotherapy, with the introduction of new checkpoint inhibitors and personalized immunotherapy enhancing treatment efficacy. Advancements in multi-omics, nanotechnology, and artificial intelligence have further deepened the understanding of the mechanisms underlying tumoral growth and their interaction with therapeutic effects, which has paved the way for precision oncology. This review aims to highlight the recent advancements in the understanding and management of skin cancer, and provide an overview of existing and emerging diagnostic, prognostic, and therapeutic modalities, while highlighting areas that require further research to bridge the existing knowledge gaps.

Acceleration of high-quality Raman imaging via a locality enhanced transformer network

Raman imaging (RI) is an outstanding technique that enables molecular-level medical diagnostics and therapy assessment by providing characteristic fingerprint and morphological information about molecules. However, obtaining high-quality Raman images generally requires a long acquisition time, up to hours, which is prohibitive for RI applications of timely cytopathology and histopathology analyses. To address this issue, image super-resolution (SR) based on deep learning, including convolutional neural networks and transformers, has been widely recognized as an effective solution to reduce the time required for achieving high-quality RI. In this study, a locality enhanced transformer network (LETNet) is proposed to perform Raman image SR. Specifically, the general architecture of the transformer is adopted with the replacement of self-attention by convolution to generate high-fidelity and detailed SR images. Additionally, the convolution in the LETNet is further optimized by utilizing depth-wise convolution to improve the computational efficiency of the model. Experiments on hyperspectral Raman images of breast cancer cells and Raman images of a few channels of brain tumor tissues demonstrate that the LETNet achieves superior 2×, 4×, and 8× SR with fewer parameters compared with other SR methods. Consequently, high-quality Raman images can be obtained with a significant reduction in time, ranging from 4 to 64 times. Overall, the proposed method provides a novel, efficient, and reliable solution to expedite high-quality RI and promote its application in real-time diagnosis and therapy.

Towards non-invasive monitoring of non-melanoma skin cancer using spatially offset Raman spectroscopy

BCC (basal cell carcinoma) and SCC (squamous cell carcinoma) account for the vast majority of cases of non-melanoma skin cancer (NMSC). The gold standard for the diagnosis remains biopsy, which, however, is an invasive and time-consuming procedure. In this study, we employed spatially offset Raman spectroscopy (SORS), a non-invasive approach, allowing the assessment of deeper skin tissue levels and collection of Raman photons with a bias towards the different layers of epidermis, where the non-melanoma cancers are initially formed and expand. Ex vivo Raman measurements were acquired from 22 skin biopsies using conventional back-scattering and a defocused modality (with and without a spatial offset). The spectral data were assessed against corresponding histopathological data to determine potential prognostic factors for lesion detection. The results revealed a positive correlation of protein and lipid content with the SCC and BCC types, respectively. By further correlating with patient data, multiple factor analysis (MFA) demonstrated a strong clustering of variables based on sex and age in all modalities. Specifically for the defocused modality (zero and 2 mm offset), further clustering occurred based on pathology. This study demonstrates the utility of the SORS technology in NMSC diagnosis prior to histopathological examination on the same tissue.

Molecular Fingerprint Detection Using Raman and Infrared Spectroscopy Technologies for Cancer Detection: A Progress Review

Molecular vibrations play a crucial role in physical chemistry and biochemistry, and Raman and infrared spectroscopy are the two most used techniques for vibrational spectroscopy. These techniques provide unique fingerprints of the molecules in a sample, which can be used to identify the chemical bonds, functional groups, and structures of the molecules. In this review article, recent research and development activities for molecular fingerprint detection using Raman and infrared spectroscopy are discussed, with a focus on identifying specific biomolecules and studying the chemical composition of biological samples for cancer diagnosis applications. The working principle and instrumentation of each technique are also discussed for a better understanding of the analytical versatility of vibrational spectroscopy. Raman spectroscopy is an invaluable tool for studying molecules and their interactions, and its use is likely to continue to grow in the future. Research has demonstrated that Raman spectroscopy is capable of accurately diagnosing various types of cancer, making it a valuable alternative to traditional diagnostic methods such as endoscopy. Infrared spectroscopy can provide complementary information to Raman spectroscopy and detect a wide range of biomolecules at low concentrations, even in complex biological samples. The article concludes with a comparison of the techniques and insights into future directions.

Raman spectroscopy in chronic heart failure diagnosis based on human skin analysis

This work aims at studying Raman spectroscopy in combination with chemometrics as an alternative fast noninvasive method to detect chronic heart failure (CHF) cases. Optical analysis is focused on the changes in the spectral features associated with the biochemical composition changes of skin tissues. A portable spectroscopy setup with the 785 nm excitation wavelength was used to record skin Raman features. In this in vivo study, 127 patients and 57 healthy volunteers were involved in measuring skin spectral features by Raman spectroscopy. The spectral data were analyzed with a projection on the latent structures and discriminant analysis. 202 skin spectra of patients with CHF and 90 skin spectra of healthy volunteers were classified with 0.888 ROC AUC for the 10-fold cross validated algorithm. To identify CHF cases, the performance of the proposed classifier was verified by means of a new test set that is equal to 0.917 ROC AUC.

Diagnostic Biomarkers for Gestational Diabetes Mellitus Using Spectroscopy Techniques: A Systematic Review

Gestational diabetes mellitus (GDM) is associated with adverse maternal and foetal consequences, along with the subsequent risk of type 2 diabetes mellitus (T2DM) and several other diseases. Due to early risk stratification in the prevention of progression of GDM, improvements in biomarker determination for GDM diagnosis will enhance the optimization of both maternal and foetal health. Spectroscopy techniques are being used in an increasing number of applications in medicine for investigating biochemical pathways and the identification of key biomarkers associated with the pathogenesis of GDM. The significance of spectroscopy promises the molecular information without the need for special stains and dyes; therefore, it speeds up and simplifies the necessary ex vivo and in vivo analysis for interventions in healthcare. All the selected studies showed that spectroscopy techniques were effective in the identification of biomarkers through specific biofluids. Existing GDM prediction and diagnosis through spectroscopy techniques presented invariable findings. Further studies are required in larger, ethnically diverse populations. This systematic review provides the up-to-date state of research on biomarkers in GDM, which were identified via various spectroscopy techniques, and a discussion of the clinical significance of these biomarkers in the prediction, diagnosis, and management of GDM.

Rapid, label-free histopathological diagnosis of liver cancer based on Raman spectroscopy and deep learning

Biopsy is the recommended standard for pathological diagnosis of liver carcinoma. However, this method usually requires sectioning and staining, and well-trained pathologists to interpret tissue images. Here, we utilize Raman spectroscopy to study human hepatic tissue samples, developing and validating a workflow for in vitro and intraoperative pathological diagnosis of liver cancer. We distinguish carcinoma tissues from adjacent non-tumour tissues in a rapid, non-disruptive, and label-free manner by using Raman spectroscopy combined with deep learning, which is validated by tissue metabolomics. This technique allows for detailed pathological identification of the cancer tissues, including subtype, differentiation grade, and tumour stage. 2D/3D Raman images of unprocessed human tissue slices with submicrometric resolution are also acquired based on visualization of molecular composition, which could assist in tumour boundary recognition and clinicopathologic diagnosis. Lastly, the potential for a portable handheld Raman system is illustrated during surgery for real-time intraoperative human liver cancer diagnosis.

Deep learning on reflectance confocal microscopy improves Raman spectral diagnosis of basal cell carcinoma

SIGNIFICANCE

Raman spectroscopy (RS) provides an automated approach for assisting Mohs micrographic surgery for skin cancer diagnosis; however, the specificity of RS is limited by the high spectral similarity between tumors and normal tissues structures. Reflectance confocal microscopy (RCM) provides morphological and cytological details by which many features of epidermis and hair follicles can be readily identified. Combining RS with deep-learning-aided RCM has the potential to improve the diagnostic accuracy of RS in an automated fashion, without requiring additional input from the clinician.

AIM

The aim of this study is to improve the specificity of RS for detecting basal cell carcinoma (BCC) using an artificial neural network trained on RCM images to identify false positive normal skin structures (hair follicles and epidermis).

APPROACH

Our approach was to build a two-step classification model. In the first step, a Raman biophysical model that was used in prior work classified BCC tumors from normal tissue structures with high sensitivity. In the second step, 191 RCM images were collected from the same site as the Raman data and served as inputs for two ResNet50 networks. The networks selected the hair structure and epidermis images, respectively, within all images corresponding to the positive predictions of the Raman biophysical model with high specificity. The specificity of the BCC biophysical model was improved by moving the Raman spectra corresponding to these selected images from false positive to true negative.

RESULTS

Deep-learning trained on RCM images removed 52% of false positive predictions from the Raman biophysical model result while maintaining a sensitivity of 100%. The specificity was improved from 84.2% using Raman spectra alone to 92.4% by integrating Raman spectra with RCM images.

CONCLUSIONS

Combining RS with deep-learning-aided RCM imaging is a promising tool for guiding tumor resection surgery.

Classification of skin cancer using convolutional neural networks analysis of Raman spectra

BACKGROUND AND OBJECTIVE

Skin cancer is the most common malignancy in whites accounting for about one third of all cancers diagnosed per year. Portable Raman spectroscopy setups for skin cancer "optical biopsy" are utilized to detect tumors based on their spectral features caused by the comparative presence of different chemical components. However, low signal-to-noise ratio in such systems may prevent accurate tumors classification. Thus, there is a challenge to develop methods for efficient skin tumors classification.

METHODS

We compare the performance of convolutional neural networks and the projection on latent structures with discriminant analysis for discriminating skin cancer using the analysis of Raman spectra with a high autofluorescence background stimulated by a 785 nm laser. We have registered the spectra of 617 cases of skin neoplasms (615 patients, 70 melanomas, 122 basal cell carcinomas, 12 squamous cell carcinomas and 413 benign tumors) in vivo with a portable Raman setup and created classification models both for convolutional neural networks and projection on latent structures approaches. To check the classification models stability, a 10-fold cross-validation was performed for all created models. To avoid models overfitting, the data was divided into a training set (80% of spectral dataset) and a test set (20% of spectral dataset).

RESULTS

The results for different classification tasks demonstrate that the convolutional neural networks significantly (p<0.01) outperforms the projection on latent structures. For the convolutional neural networks implementation we obtained ROC AUCs of 0.96 (0.94 - 0.97; 95% CI), 0.90 (0.85-0.94; 95% CI), and 0.92 (0.87 - 0.97; 95% CI) for classifying a) malignant vs benign tumors, b) melanomas vs pigmented tumors and c) melanomas vs seborrheic keratosis respectively.

CONCLUSIONS

The performance of the convolutional neural networks classification of skin tumors based on Raman spectra analysis is higher or comparable to the accuracy provided by trained dermatologists. The increased accuracy with the convolutional neural networks implementation is due to a more precise accounting of low intensity Raman bands in the intense autofluorescence background. The achieved high performance of skin tumors classifications with convolutional neural networks analysis opens a possibility for wide implementation of Raman setups in clinical setting.

Label-Free Differentiation of Cancer and Non-Cancer Cells Based on Machine-Learning-Algorithm-Assisted Fast Raman Imaging

This paper proposes a rapid, label-free, and non-invasive approach for identifying murine cancer cells (B16F10 melanoma cancer cells) from non-cancer cells (C2C12 muscle cells) using machine-learning-assisted Raman spectroscopic imaging. Through quick Raman spectroscopic imaging, a hyperspectral data processing approach based on machine learning methods proved capable of presenting the cell structure and distinguishing cancer cells from non-cancer muscle cells without compromising full-spectrum information. This study discovered that biomolecular information-nucleic acids, proteins, and lipids-from cells could be retrieved efficiently from low-quality hyperspectral Raman datasets and then employed for cell line differentiation.

Co-localized line-field confocal optical coherence tomography and confocal Raman microspectroscopy for three-dimensional high-resolution morphological and molecular characterization of skin tissues ex vivo

Line-field confocal optical coherence tomography (LC-OCT) is an optical modality that provides three-dimensional (3D) images of the skin at cellular resolution. Confocal Raman microspectroscopy (CRM) is a label-free optical technique that can provide point measurement of the molecular content of the skin. This work presents a method to co-localize LC-OCT and CRM acquisitions for morpho-molecular analysis of ex vivo skin tissues at cellular level. The co-localization method allows acquisition of Raman spectra at specific locations in a sample identified from a 3D LC-OCT image, with an accuracy of ± 20 µm. The method was applied to the characterization of tattooed skin biopsies with adverse tattoo reactions. LC-OCT images allowed to target specific regions in the biopsies where the presence of tattoo ink was revealed by detection of the Raman signature of ink pigments. Micrometer-sized foreign bodies of various materials as well as inflammatory cells were also identified within the biopsies. From these results, we demonstrate the value of the LC-OCT-CRM co-localization method and its potential for future ex vivo analysis of suspicious skin lesions.

Proposal for a Skin Layer-Wise Decomposition Model of Spatially-Resolved Diffuse Reflectance Spectra Based on Maximum Depth Photon Distributions: A Numerical Study

In the context of cutaneous carcinoma diagnosis based on in vivo optical biopsy, Diffuse Reflectance (DR) spectra, acquired using a Spatially Resolved (SR) sensor configuration, can be analyzed to distinguish healthy from pathological tissues. The present contribution aims at studying the depth distribution of SR-DR-detected photons in skin from the perspective of analyzing how these photons contribute to acquired spectra carrying local physiological and morphological information. Simulations based on modified Cuda Monte Carlo Modeling of Light transport were performed on a five-layer human skin optical model with epidermal thickness, phototype and dermal blood content as variable parameters using (i) wavelength-resolved scattering and absorption properties and (ii) the geometrical configuration of a multi-optical fiber probe implemented on an SR-DR spectroscopic device currently used in clinics. Through histograms of the maximum probed depth and their exploitation, we provide numerical evidence linking the characteristic penetration depth of the detected photons to their wavelengths and four source–sensor distances, which made it possible to propose a decomposition of the DR signals related to skin layer contributions.

Characterization of Ex Vivo Nonmelanoma Skin Tissue Using Raman Spectroscopy

Raman spectroscopy has shown great potential in detecting nonmelanoma skin cancer accurately and quickly; however, little direct evidence exists on the sensitivity of measurements to the underlying anatomy. Here, we aimed to correlate Raman measurements directly to the underlying tissue anatomy. We acquired Raman spectra of ex vivo skin tissue from 25 patients undergoing Mohs surgery with a fiber probe. We utilized a previously developed biophysical model to extract key biomarkers in the skin from the Raman spectra. We then examined the correlations between the biomarkers and the major skin structures (including the dermis, sebaceous glands, hair follicles, fat, and two types of nonmelanoma skin cancer—basal cell carcinoma (BCC) and squamous cell carcinoma (SCC)). SCC had a significantly different concentration of keratin, collagen, and nucleic acid than normal structures, while ceramide differentiated BCC from normal structures. Our findings identified the key proteins, lipids, and nucleic acids that discriminate nonmelanoma tumors and healthy skin using Raman spectroscopy. These markers may be promising surgical guidance tools for detecting tumors in resection margins.

Handheld macroscopic Raman spectroscopy imaging instrument for machine-learning-based molecular tissue margins characterization

SIGNIFICANCE

Raman spectroscopy has been developed for surgical guidance applications interrogating live tissue during tumor resection procedures to detect molecular contrast consistent with cancer pathophysiological changes. To date, the vibrational spectroscopy systems developed for medical applications include single-point measurement probes and intraoperative microscopes. There is a need to develop systems with larger fields of view (FOVs) for rapid intraoperative cancer margin detection during surgery.

AIM

We design a handheld macroscopic Raman imaging system for in vivo tissue margin characterization and test its performance in a model system.

APPROACH

The system is made of a sterilizable line scanner employing a coherent fiber bundle for relaying excitation light from a 785-nm laser to the tissue. A second coherent fiber bundle is used for hyperspectral detection of the fingerprint Raman signal over an area of 1  cm2. Machine learning classifiers were trained and validated on porcine adipose and muscle tissue.

RESULTS

Porcine adipose versus muscle margin detection was validated ex vivo with an accuracy of 99% over the FOV of 95  mm2 in ∼3  min using a support vector machine.

CONCLUSIONS

This system is the first large FOV Raman imaging system designed to be integrated in the workflow of surgical cancer resection. It will be further improved with the aim of discriminating brain cancer in a clinically acceptable timeframe during glioma surgery.

Diagnosis of active tuberculosis and latent tuberculosis infection based on Raman spectroscopy and surface-enhanced Raman spectroscopy

Current tools for screening LTBI are limited due to the long turnaround time required, cross-reactivity of tuberculin skin test to BCG vaccine and the high cost of interferon gamma release assay (IGRA) tests. We evaluated Raman spectroscopy (RS) for serum-protein fingerprinting from 26 active TB (ATB) cases, 20 LTBI cases, 34 early clearance (EC; TB-exposed persons with undetected infection) and 38 healthy controls (HC). RS at 532 nm using candidate peaks provided 92.31% sensitivity and 90.0% to distinguish ATB from LTBI, 84.62% sensitivity and 89.47% specificity to distinguish ATB from HC and 87.10% sensitivity and 85.0% specificity to distinguish LTBI from EC. RS at 532 nm with the random forest model provided 86.84% sensitivity and 65.0% specificity to distinguish LTBI from HC and 94.74% sensitivity and 87.10% specificity to distinguish EC from HC. Using preliminary sample sets (n = 5 for each TB-infection category), surface-enhanced Raman spectroscopy (SERS) showed high potential diagnostic performance, distinguishing very clearly among all TB-infection categories with 100% sensitivity and specificity. With lower cost, shorter turnaround time and performance comparable to that of IGRAs, our study demonstrated RS and SERS to have high potential for ATB and LTBI diagnosis.

Superpixel Raman spectroscopy for rapid skin cancer margin assessment

Spontaneous Raman micro-spectroscopy has been demonstrated great potential in delineating tumor margins; however, it is limited by slow acquisition speed. We describe a superpixel acquisition approach that can expedite acquisition between ~×100 and ×10 000, as compared to point-by-point scanning by trading off spatial resolution. We present the first demonstration of superpixel acquisition on rapid discrimination of basal cell carcinoma tumor from eight patients undergoing Mohs micrographic surgery. Results have been demonstrated high discriminant power for tumor vs normal skin based on the biochemical differences between nucleus, collagen, keratin and ceramide. We further perform raster-scanned superpixel Raman imaging on positive and negative margin samples. Our results indicate superpixel acquisition can facilitate the use of Raman microspectroscopy as a rapid and specific tool for tumor margin assessment.

Towards development of a novel universal medical diagnostic method: Raman spectroscopy and machine learning

This review summarizes recent progress made using Raman spectroscopy and machine learning for potential universal medical diagnostic applications.