Nonlinear optical microscopy signal processing strategies in cancer

Imaging of colorectal adenomas with pseudoinvasion and malignant polyps using two-photon excitation microscopy

Introduction

Although the incidence and mortality rates of colorectal cancer exhibit significant variability, it remains one of the most prevalent cancers worldwide. Endeavors to prevent colorectal cancer development focus on detecting precursor lesions during colonoscopy. The diagnosis of endoscopically resected polyps relies on hematoxylin and eosin staining examination. For challenging cases like adenomatous polyps with epithelial misplacement, additional diagnostic methods could prove beneficial.

Methods

This paper aims to underscore stromal changes observed in malignant polyps and polyps with pseudoinvasion, leveraging two-photon excitation microscopy (TPEM), a technique extensively employed in the medical field in recent years.

Results and discussions

Both the subjective and quantitative analysis of TPEM images revealed distinct distributions and densities of collagen at the invasion front in malignant polyps compared to areas of pseudoinvasion. TPEM holds potential in discerning true invasion in malignant polyps from pseudoinvasion, offering enhanced visualization of local stromal changes.

A radiomics-based interpretable model to predict the pathological grade of pancreatic neuroendocrine tumors

OBJECTIVES

To develop a computed tomography (CT) radiomics-based interpretable machine learning (ML) model to predict the pathological grade of pancreatic neuroendocrine tumors (pNETs) in a non-invasive manner.

METHODS

Patients with pNETs who underwent contrast-enhanced abdominal CT between 2010 and 2022 were included in this retrospective study. Radiomics features were extracted, and five radiomics-based ML models, namely logistic regression (LR), random forest (RF), support vector machine (SVM), XGBoost, and GaussianNB, were developed. The performance of these models was evaluated using a time-independent testing set, and metrics such as sensitivity, specificity, accuracy, and the area under the receiver operating characteristic curve (AUC) were calculated. The accuracy of the radiomics model was compared to that of needle biopsy. The Shapley Additive Explanation (SHAP) tool and the correlation between radiomics and biological features were employed to explore the interpretability of the model.

RESULTS

A total of 122 patients (mean age: 50 ± 14 years; 53 male) were included in the training set, whereas 100 patients (mean age: 48 ± 13 years; 50 male) were included in the testing set. The AUCs for LR, SVM, RF, XGBoost, and GaussianNB were 0.758, 0.742, 0.779, 0.744, and 0.745, respectively, with corresponding accuracies of 73.0%, 70.0%, 77.0%, 71.9%, and 72.9%. The SHAP tool identified two features of the venous phase as the most significant, which showed significant differences among the Ki-67 index or mitotic count subgroups (p < 0.001).

CONCLUSIONS

An interpretable radiomics-based RF model can effectively differentiate between G1 and G2/3 of pNETs, demonstrating favorable interpretability.

CLINICAL RELEVANCE STATEMENT

The radiomics-based interpretable model developed in this study has significant clinical relevance as it offers a non-invasive method for assessing the pathological grade of pancreatic neuroendocrine tumors and holds promise as an important complementary tool to traditional tissue biopsy.

KEY POINTS

• A radiomics-based interpretable model was developed to predict the pathological grade of pNETs and compared with preoperative needle biopsy in terms of accuracy. • The model, based on CT radiomics, demonstrated favorable interpretability. • The radiomics model holds potential as a valuable complementary technique to preoperative needle biopsy; however, it should not be considered a replacement for biopsy.

SLAM medical imaging enabled by pre-chirp and gain jointly managed Yb-fiber laser

We demonstrate a pre-chirp and gain jointly managed Yb-fiber laser that drives simultaneous label-free autofluorescence-multiharmonic (SLAM) medical imaging. We show that a gain managed Yb-fiber amplifier produces high-quality compressed pulses when the seeding pulses exhibit proper negative pre-chirp. The resulting laser source can generate 43-MHz, 34-fs pulses centered at 1110 nm with more than 90-nJ energy. We apply this ultrafast source to SLAM imaging of cellular and extracellular components in various human tissues of intestinal adenocarcinoma, lung adenocarcinoma, and liver.

A computed tomography urography-based machine learning model for predicting preoperative pathological grade of upper urinary tract urothelial carcinoma

OBJECTIVES

Development and validation of a computed tomography urography (CTU)-based machine learning (ML) model for prediction of preoperative pathology grade of upper urinary tract urothelial carcinoma (UTUC).

METHODS

A total of 140 patients with UTUC who underwent CTU examination from January 2017 to August 2023 were retrospectively enrolled. Tumor lesions on the unenhanced, medullary, and excretory periods of CTU were used to extract Features, respectively. Feature selection was screened by the Pearson and Spearman correlation analysis, least absolute shrinkage and selection operator algorithm, random forest (RF), support vector machine (SVM), and eXtreme Gradient Boosting (XGBoost). The logistic regression (LR) was used to screen for independent influencing factors of clinical baseline characteristics. Machine learning models based on different feature datasets were constructed and validated using algorithms such as LR, RF, SVM, and XGBoost. By computing the selected features, a radiomics score was generated, and a diverse feature dataset was constructed. Based on the training set, 16 ML models were created, and their performance was evaluated using the validation set for metrics including sensitivity, specificity, accuracy, area under the receiver operating characteristic curve (AUC), and others.

RESULTS

The training set consisted of 98 patients (mean age: 64.5 ± 10.5 years; 30 males), whereas the validation set consisted of 42 patients (mean age: 65.3 ± 9.78 years; 17 males). Hydronephrosis was the best independent influence factor (p < 0.05). The RF model had the best performance in predicting high-grade UTUC, with AUC of 0.914 (95% Confidence Interval [95%CI] 0.852-0.977) and 0.903 (95%CI 0.809-0.997) in the training set and validation set, and accuracy of 0.878 and 0.857, respectively.

CONCLUSIONS

An ML model based on the RF algorithm exhibits excellent predictive performance, offering a non-invasive approach for predicting preoperative high-grade UTUC.

Gray-Level Co-occurrence Matrix Analysis for the Detection of Discrete, Ethanol-Induced, Structural Changes in Cell Nuclei: An Artificial Intelligence Approach

Gray-level co-occurrence matrix (GLCM) analysis is a contemporary and innovative computational method for the assessment of textural patterns, applicable in almost any area of microscopy. The aim of our research was to perform the GLCM analysis of cell nuclei in Saccharomyces cerevisiae yeast cells after the induction of sublethal cell damage with ethyl alcohol, and to evaluate the performance of various machine learning (ML) models regarding their ability to separate damaged from intact cells. For each cell nucleus, five GLCM parameters were calculated: angular second moment, inverse difference moment, GLCM contrast, GLCM correlation, and textural variance. Based on the obtained GLCM data, we applied three ML approaches: neural network, random trees, and binomial logistic regression. Statistically significant differences in GLCM features were observed between treated and untreated cells. The multilayer perceptron neural network had the highest classification accuracy. The model also showed a relatively high level of sensitivity and specificity, as well as an excellent discriminatory power in the separation of treated from untreated cells. To the best of our knowledge, this is the first study to demonstrate that it is possible to create a relatively sensitive GLCM-based ML model for the detection of alcohol-induced damage in Saccharomyces cerevisiae cell nuclei.

Monitoring the extracellular matrix remodeling of high-grade serous ovarian cancer with nonlinear optical microscopy

The mortality of high-grade serous ovarian cancer (HGSOC) accounts for 70% to 80% of all ovarian cancer deaths and overall mortality rate has not declined in the last decade. Recently, many studies have demonstrated that HGSOC originates from the fallopian tubes. The extracellular matrix (ECM) is present in all tissues, its remodeling and interaction with cells are crucial for regulating cell proliferation, migration, and differentiation. In this paper, we used label-free nonlinear optical microscopy to image tissues of the fallopian tube and ovary. Combining a set of image processing algorithms, we monitored the remodeling of ECM in the fallopian tube and ovary during the invasion of primary serous fallopian tube tumor into the ovary in microscopic dimension. With this approach, we can obtain physiological information of HGSOC at the early stage, which provided useful data for auxiliary clinical diagnosis.

Analysis of tumor nuclear features using artificial intelligence to predict response to neoadjuvant chemotherapy in high-risk breast cancer patients

PURPOSE

Neoadjuvant chemotherapy (NAC) is used to treat patients with high-risk breast cancer. The tumor response to NAC can be classified as either a pathological partial response (pPR) or pathological complete response (pCR), defined as complete eradication of invasive tumor cells, with a pCR conferring a significantly lower risk of recurrence. Predicting the response to NAC, however, remains a significant clinical challenge. The objective of this study was to determine if analysis of nuclear features on core biopsies using artificial intelligence (AI) can predict response to NAC.

METHODS

Fifty-eight HER2-positive or triple-negative breast cancer patients were included in this study (pCR n = 37, pPR n = 21). Multiple deep convolutional neural networks were developed to automate tumor detection and nuclear segmentation. Nuclear count, area, and circularity, as well as image-based first- and second-order features including mean pixel intensity and correlation of the gray-level co-occurrence matrix (GLCM-COR) were determined.

RESULTS

In univariate analysis, the pCR group had fewer multifocal/multicentric tumors, higher nuclear intensity, and lower GLCM-COR compared to the pPR group. In multivariate binary logistic regression, tumor multifocality/multicentricity (OR = 0.14, p = 0.012), nuclear intensity (OR = 1.23, p = 0.018), and GLCM-COR (OR = 0.96, p = 0.043) were each independently associated with likelihood of achieving a pCR, and the model was able to successful classify 79% of cases (62% for pPR and 89% for pCR).

CONCLUSION

Analysis of tumor nuclear features using digital pathology/AI can significantly improve models to predict pathological response to NAC.

Non-invasive Imaging Techniques: From Histology to In Vivo Imaging : Chapter of Imaging in Oncology

In this chapter, we will introduce and review molecular-sensitive imaging techniques, which close the gap between ex vivo and in vivo analysis. In detail, we will introduce spontaneous Raman spectral imaging, coherent anti-Stokes Raman scattering (CARS), stimulated Raman scattering (SRS), second-harmonic generation (SHG) and third-harmonic generation (THG), two-photon excited fluorescence (TPEF), and fluorescence lifetime imaging (FLIM). After reviewing these imaging techniques, we shortly introduce chemometric methods and machine learning techniques, which are needed to use these imaging techniques in diagnostic applications.

Two-photon images reveal unique texture features for label-free identification of ovarian cancer peritoneal metastases

For cancer patients, treatment selection fundamentally relies on staging, with "under-staging" considered a common problem. Imaging modalities that can complement conventional white-light laparoscopy are needed to detect more accurately small metastatic lesions in patients undergoing operative cancer care. Biopsies from healthy parietal peritoneum and ovarian peritoneal metastases obtained from 8 patients were imaged employing a two-photon laser scanning microscope to generate collagen-second harmonic generation (SHG) and fluorescence images at 755 nm and 900 nm excitation and 460 ± 20 nm and 525 ± 25 nm emission. Forty-one images were analyzed by automated image processing algorithms and statistical textural analysis techniques, namely gray level co-occurrence matrices. Two textural features (contrast and correlation) were employed to describe the spatial intensity variations within the captured images and outcomes were used for discriminant analysis. We found that healthy tissues displayed large variations in contrast and correlation features as a function of distance, corresponding to repetitive, increased local intensity fluctuations. Metastatic tissue images exhibited decreased contrast and correlation related values, representing more uniform intensity patterns and smaller fibers, indicating the destruction of the healthy stroma by the cancerous infiltration. The textural outcomes resulted in high classification accuracy as evaluated quantitatively by discriminant analysis.

Surgical preparations, labeling strategies, and optical techniques for cell-resolved, in vivo imaging in the mouse spinal cord

In vivo optical imaging has enabled detailed studies of cellular dynamics in the brain of rodents in both healthy and diseased states. Such studies were made possible by three advances: surgical preparations that give optical access to the brain; strategies for in vivo labeling of cells with structural and functional fluorescent indicators; and optical imaging techniques that are relatively insensitive to light scattering by tissue. In vivo imaging in the rodent spinal cord has lagged behind than that in the brain, largely due to the anatomy around the spinal cord that complicates the surgical preparation, and to the strong optical scattering of the dorsal white matter that limits the ability to image deep into the spinal cord. Here, we review recent advances in surgical methods, labeling strategies, and optical tools that have enabled in vivo, high-resolution imaging of the dynamic behaviors of cells in the spinal cord in mice. Surgical preparations that enable long-term optical access and robust stabilization of the spinal cord are now available. Labeling strategies that have been used in the spinal cord tend to follow those that have been used in the brain, and some recent advances in genetically-encoded labeling strategies remain to be capitalized on. The optical imaging methods used to date, including two photon excited fluorescence microscopy, are largely limited to imaging the superficial layers of the spinal cord by the optical scattering of the white matter. Finally, we show preliminary data that points to the use of higher-order nonlinear optical processes, such as three photon excited fluorescence, as a means to image deeper into the mouse spinal cord.

Classification of established atopic dermatitis in children with the in vivo imaging methods

Atopic dermatitis (AD) is a cutaneous disease resulting from a defective barrier and dysregulated immune response. The severity scoring of atopic dermatitis (SCORAD) is used to classify AD. Noninvasive imaging approaches supplementary to SCORAD were investigated. Cr:forsterite laser-based microscopy was employed to analyze endogenous third-harmonic generation (THG) and second-harmonic generation (SHG) signals from skin. Imaging parameters were compared between different AD severities. Three-dimensional reconstruction of imaged skin layers was performed. Finally, statistic models from quantitative imaging parameters were developed for predicting disease severity. Our data demonstrate that THG signal intensity of lesional skin in AD were significantly increased and was positively correlated with AD severity. Characteristic gray level co-occurrence matrix (GLCM) values were observed in more severe AD. In the 3D reconstruction video, individual dermal papilla and obvious fibrosis in the upper papillary dermis were easily identified. Our estimation models could predict the disease severity of AD patients with an accuracy of nearly 85%. The THG signal intensity and characteristic GLCM patterns are associated with AD severity and can serve as quantitative predictive parameters. Our imaging approach can be used to identify the histopathological changes of AD objectively, and to complement the SCORAD index, thus improving the accuracy of classifying AD severity.

Multimodal imaging to explore endogenous fluorescence of fresh and fixed human healthy and tumor brain tissues

To complement a project toward label-free optical biopsy and enhanced resection which the overall goal is to develop a multimodal nonlinear endomicroscope, this multimodal approach aims to enhance the accuracy in classifying brain tissue into solid tumor, infiltration and normal tissue intraoperatively. Multiple optical measurements based on one- and two-photon spectral and lifetime autofluorescence, including second harmonic generation imaging, were acquired. As a prerequisite, studying the effect of the time of measurement postexcision on tissue's spectral/lifetime fluorescence properties was warranted, so spectral and lifetime fluorescences of fresh brain tissues were measured using a point-based linear endoscope. Additionally, a comparative study on tissue's optical properties obtained by multimodal nonlinear optical imaging microscope from fresh and fixed tissue was necessary to test whether clinical validation of the nonlinear endomicroscope is feasible by extracting optical signatures from fixed tissue rather than from freshly excised samples. The former is generally chosen for convenience. Results of this study suggest that an hour is necessary postexcision to have consistent fluorescence intensities\lifetimes. The fresh (a,b,c) vs fixed (d,e,f) tissue study indicates that while all optical signals differ after fixation. The characteristic features extracted from one- and two-photon excitation still discriminate normal brain (a,d) cortical tissue, glioblastoma (GBM) (b,e) and metastases (c,f).

Quantitative second harmonic generation microscopy for the structural characterization of capsular collagen in thyroid neoplasms

Quantitative second harmonic generation microscopy was used to investigate collagen organization in the fibrillar capsules of human benign and malignant thyroid nodules. We demonstrate that the combination of texture analysis and second harmonic generation images of collagen can be used to differentiate between capsules surrounding the thyroid follicular adenoma and papillary carcinoma nodules. Our findings indicate that second harmonic generation microscopy can provide quantitative information about the collagenous capsule surrounding both the thyroid and thyroid nodules, which may complement traditional histopathological examination.

Collagen analysis by second-harmonic generation microscopy predicts outcome of luminal breast cancer

Second-harmonic generation microscopy represents an important tool to evaluate extracellular matrix collagen structure, which undergoes changes during cancer progression. Thus, it is potentially relevant to assess breast cancer development. We propose the use of second-harmonic generation images of tumor stroma selected on hematoxylin and eosin–stained slides to evaluate the prognostic value of collagen fibers analyses in peri and intratumoral areas in patients diagnosed with invasive ductal breast carcinoma. Quantitative analyses of collagen parameters were performed using ImageJ software. These parameters presented significantly higher values in peri than in intratumoral areas. Higher intratumoral collagen uniformity was associated with high pathological stages and with the presence of axillary lymph node metastasis. In patients with immunohistochemistry-based luminal subtype, higher intratumoral collagen uniformity and quantity were independently associated with poorer relapse-free and overall survival, respectively. A multivariate response recursive partitioning model determined 12.857 and 11.894 as the best cut-offs for intratumoral collagen quantity and uniformity, respectively. These values have shown high sensitivity and specificity to differentiate distinct outcomes. Values of intratumoral collagen quantity and uniformity exceeding the cut-offs were strongly associated with poorer relapse-free and overall survival. Our findings support a promising prognostic value of quantitative evaluation of intratumoral collagen by second-harmonic generation imaging mainly in the luminal subtype breast cancer.

Impact of Body Site, Age, and Gender on the Collagen/Elastin Index by Noninvasive in vivo Vertical Two-Photon Microscopy

BACKGROUND/AIMS

Extrinsic and intrinsic skin aging is subject to constant remodeling and degradation processes, primarily in components of the extracellular matrix. While collagen fibers thin out during the aging process, the amorphous elastin fibers accumulate. These are essential formative components of the dermis. So far, these processes have been detected in vertical histological sections of invasive biopsies and recently in noninvasive horizontal scans.

METHODS

In this pilot study, a modified noninvasive 2-photon microscope was applied to measure the collagen/elastin index of skin in vivo. The obtained images permit an immediate vertical survey and allow a conclusion on the dermal composition at once. The collagen/elastin index was quantified by the second harmonic to autofluorescence aging index of dermis (SAAID) depending on volunteers' age (18-66 years), gender, and body area.

RESULTS

The highest SAAID was measured at the volar forearm as compared to the abdominal SAAID, which was significantly lower (p < 0.05). The gluteal region showed the significantly lowest SAAID (p < 0.05). The SAAID in female skin was higher compared to male skin and decreased with increasing age.

CONCLUSION

These effects are to be considered in subsequent studies to be able to specifically detect and evaluate influences.

Influences of Orally Taken Carotenoid-Rich Curly Kale Extract on Collagen I/Elastin Index of the Skin

Two differently designed, spatially resolved reflectance spectroscopy-based scanners and two-photon tomography were used for noninvasive in vivo determination of cutaneous carotenoids, and collagen I/elastin aging index of dermis, respectively, in the skin of 29 healthy female volunteers between 40 and 56 years of age. The volunteers received a supplement in the form of a carotenoid-rich natural curly kale extract containing 1650 µg of carotenoids in total (three capsules of 550 µg), once a day. Measurements were taken before, after 5 months and after 10 months of daily supplementation. The results showed significantly increased values for the cutaneous carotenoids and the collagen I/elastin aging index of dermis 5 and 10 months after the beginning of the study. The obtained results show that a natural carotenoid-rich extract could prevent the aging-related collagen I degradation in the dermis and improve the extracellular matrix.

Epithelial Ovarian Cancer Diagnosis of Second-Harmonic Generation Images: A Semiautomatic Collagen Fibers Quantification Protocol

A vast number of human pathologic conditions are directly or indirectly related to tissular collagen structure remodeling. The nonlinear optical microscopy second-harmonic generation has become a powerful tool for imaging biological tissues with anisotropic hyperpolarized structures, such as collagen. During the past years, several quantification methods to analyze and evaluate these images have been developed. However, automated or semiautomated solutions are necessary to ensure objectivity and reproducibility of such analysis. This work describes automation and improvement methods for calculating the anisotropy (using fast Fourier transform analysis and the gray-level co-occurrence matrix). These were applied to analyze biopsy samples of human ovarian epithelial cancer at different stages of malignancy (mucinous, serous, mixed, and endometrial subtypes). The semiautomation procedure enabled us to design a diagnostic protocol that recognizes between healthy and pathologic tissues, as well as between different tumor types.

An Engineered Breast Cancer Model on a Chip to Replicate ECM-Activation In Vitro during Tumor Progression

In this work, a new model of breast cancer is proposed featuring both epithelial and stromal tissues arranged on a microfluidic chip. The main task of the work is the in vitro replication of the stromal activation during tumor epithelial invasion. The activation of tumor stroma and its morphological/compositional changes play a key role in tumor progression. Despite emerging evidences, to date the activation of tumor stroma in vitro has not been achieved yet. The tumor-on-chip proposed in this work is built in order to replicate the features of its native counterpart: multicellularity (tumor epithelial cell and stromal cell); 3D engineered stroma compartment composed of cell-assembled extracellular matrix (ECM); reliable 3D tumor architecture. During tumor epithelial invasion the stroma displayed an activation process at both cellular and ECM level. Similarly of what repeated in vivo, ECM remodeling is found in terms of hyaluronic acid and fibronectin overexpression in the stroma compartment. Furthermore, the cell-assembled ECM featuring the stromal tissue, allowed on-line monitoring of collagen remodeling during stroma activation process via real time multiphoton microscopy. Also, trafficking of macromolecules within the stromal compartment has been monitored in real time.

A spectral approach for the quantitative description of cardiac collagen network from nonlinear optical imaging

The assessment of collagen structure in cardiac pathology, such as atrial fibrillation (AF), is essential for a complete understanding of the disease. This paper introduces a novel methodology for the quantitative description of collagen network properties, based on the combination of nonlinear optical microscopy with a spectral approach of image processing and analysis. Second-harmonic generation (SHG) microscopy was applied to atrial tissue samples from cardiac surgery patients, providing label-free, selective visualization of the collagen structure. The spectral analysis framework, based on 2D-FFT, was applied to the SHG images, yielding a multiparametric description of collagen fiber orientation (angle and anisotropy indexes) and texture scale (dominant wavelength and peak dispersion indexes). The proof-of-concept application of the methodology showed the capability of our approach to detect and quantify differences in the structural properties of the collagen network in AF versus sinus rhythm patients. These results suggest the potential of our approach in the assessment of collagen properties in cardiac pathologies related to a fibrotic structural component.

Phase matching of backward second harmonic generation assisted by lattice structure in collagen tissues

Phase matching of backward second harmonic generation (SHG) in a periodic structure of collagen fibrils is investigated through theoretical modeling, simulation, and experiments. The lattice structure of collagen fibrils is considered to provide a virtual momentum for assisting the phase matching of backward SHG. Phase matching over a relatively wide excitation wavelength range is achieved by tilting the angle of the fundamental excitation and SHG wave vectors. The SHG intensity in the periodic structure is simulated to quantify the phase matching effect. The effect of the fundamental excitation and the SHG emission angles on the peak excitation wavelength of the SHG excitation spectrum is further validated in experiments, where the excitation and emission angles are controlled by spatial filtering. It is found that an optimized excitation wavelength exists for a certain collagen fibril structure, which shifts toward a shorter wavelength when the excitation and emission angles are increased. Our results show that the lattice structure of collagen fibrils can assist the phase matching, providing a mechanism for generating backward SHG in multiphoton microscopy.

Gray-Level Co-Occurrence Matrix Texture Analysis of Breast Tumor Images in Prognosis of Distant Metastasis Risk

Owing to exceptional heterogeneity in the outcome of invasive breast cancer it is essential to develop highly accurate prognostic tools for effective therapeutic management. Based on this pressing need, we aimed to improve breast cancer prognosis by exploring the prognostic value of tumor histology image analysis. Patient group (n=78) selection was based on invasive breast cancer diagnosis without systemic treatment with a median follow-up of 147 months. Gray-level co-occurrence matrix texture analysis was performed retrospectively on primary tumor tissue section digital images stained either nonspecifically with hematoxylin and eosin or specifically with a pan-cytokeratin antibody cocktail for epithelial malignant cells. Univariate analysis revealed stronger association with metastasis risk by texture analysis when compared with clinicopathological parameters. The combination of individual clinicopathological and texture variables into composite scores resulted in further powerful enhancement of prognostic performance, with an accuracy of up to 90%, discrimination efficiency by the area under the curve [95% confidence interval (CI)] of 0.94 (0.87-0.99) and hazard ratio (95% CI) of 20.1 (7.5-109.4). Internal validation was successfully performed by bootstrap and split-sample cross-validation, suggesting that the models are generalizable. Whereas further validation is needed on an external set of patients, this preliminary study indicates the potential use of primary breast tumor histology texture as a highly accurate, simple, and cost-effective prognostic indicator of distant metastasis risk.