Nonlinear Imaging Histopathology: A Pipeline to Correlate Gold-Standard Hematoxylin and Eosin Staining With Modern Nonlinear Microscopy

Deep learning-based classification of breast cancer molecular subtypes from H&E whole-slide images

Classifying breast cancer molecular subtypes is crucial for tailoring treatment strategies. While immunohistochemistry (IHC) and gene expression profiling are standard methods for molecular subtyping, IHC can be subjective, and gene profiling is costly and not widely accessible in many regions. Previous approaches have highlighted the potential application of deep learning models on hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) for molecular subtyping, but these efforts vary in their methods, datasets, and reported performance. In this work, we investigated whether H&E-stained WSIs could be solely leveraged to predict breast cancer molecular subtypes (luminal A, B, HER2-enriched, and Basal). We used 1433 WSIs of breast cancer in a two-step pipeline: first, classifying tumor and non-tumor tiles to use only the tumor regions for molecular subtyping; and second, employing a One-vs-Rest (OvR) strategy to train four binary OvR classifiers and aggregating their results using an eXtreme Gradient Boosting model. The pipeline was tested on 221 hold-out WSIs, achieving an F1 score of 0.95 for tumor vs non-tumor classification and a macro F1 score of 0.73 for molecular subtyping. Our findings suggest that, with further validation, supervised deep learning models could serve as supportive tools for molecular subtyping in breast cancer. Our codes are made available to facilitate ongoing research and development.

Large field-of-view metabolic profiling of murine brain tissue following morphine incubation using label-free multiphoton microscopy

BACKGROUND

Although the effects on neural activation and glucose consumption caused by opiates such as morphine are known, the metabolic machinery underlying opioid use and misuse is not fully explored. Multiphoton microscopy (MPM) techniques have been developed for optical imaging at high spatial resolution. Despite the increased use of MPM for neural imaging, the use of intrinsic optical contrast has seen minimal use in neuroscience.

NEW METHOD

We present a label-free, multimodal microscopy technique for metabolic profiling of murine brain tissue following incubation with morphine sulfate (MSO4). We evaluate two- and three-photon excited autofluorescence, and second and third harmonic generation to determine meaningful intrinsic contrast mechanisms in brain tissue using simultaneous label-free, autofluorescence multi-harmonic (SLAM) microscopy.

RESULTS

Regional differences quantified in the cortex, caudate, and thalamus of the brain demonstrate region-specific changes to metabolic profiles measured from FAD intensity, along with brain-wide quantification. While the overall intensity of FAD signal significantly decreased after morphine incubation, this metabolic molecule accumulated near the nucleus accumbens.

COMPARISON WITH EXISTING METHODS

Histopathology requires tissue fixation and staining to determine cell type and morphology, lacking information about cellular metabolism. Tools such as fMRI or PET imaging have been widely used, but lack cellular resolution. SLAM microscopy obviates the need for tissue preparation, permitting immediate use and imaging of tissue with subcellular resolution in its native environment.

CONCLUSIONS

This study demonstrates the utility of SLAM microscopy for label-free investigations of neural metabolism, especially the intensity changes in FAD autofluorescence and structural morphology from third-harmonic generation.

Simultaneous label-free autofluorescence multi-harmonic microscopy driven by the supercontinuum generated from a bulk nonlinear crystal

Nonlinear microscopy encompasses several imaging techniques that leverage laser technology to probe intrinsic molecules of biological specimens. These native molecules produce optical fingerprints that allow nonlinear microscopes to reveal the chemical composition and structure of cells and tissues in a label-free and non-destructive fashion, information that enables a plethora of applications, e.g., real-time digital histopathology or image-guided surgery. Because state-of-the-art lasers exhibit either a limited bandwidth or reduced wavelength tunability, nonlinear microscopes lack the spectral support to probe different biomolecules simultaneously, thus losing analytical potential. Therefore, a conventional nonlinear microscope requires multiple or tunable lasers to individually excite endogenous molecules, increasing both the cost and complexity of the system. A solution to this problem is supercontinuum generation, a nonlinear optical phenomenon that supplies broadband femtosecond radiation, granting a wide spectrum for concurrent molecular excitation. This study introduces a source for nonlinear multiphoton microscopy based on the supercontinuum generation from a yttrium aluminum garnet (YAG) crystal, an approach that allows simultaneous label-free autofluorescence multi-harmonic imaging of biological samples and offers a practical and compact alternative for the clinical translation of nonlinear microscopy. While this supercontinuum covered the visible spectrum (550-900 nm) and the near-infrared region (950-1200 nm), the pulses within 1030-1150 nm produced label-free volumetric chemical images of ex vivo chinchilla kidney, thus validating the supercontinuum from bulk crystals as a powerful source for multimodal nonlinear microscopy.

Multimodal label-free imaging of murine hepatocellular carcinoma with a subcellular resolution

We demonstrate label-free imaging of genetically induced hepatocellular carcinoma (HCC) in a murine model provided by two- and three-photon fluorescence microscopy of endogenous fluorophores excited at the central wavelengths of 790, 980 and 1250 nm and reinforced by second and third harmonic generation microscopy. We show, that autofluorescence imaging presents abundant information about cell arrangement and lipid accumulation in hepatocytes and hepatic stellate cells (HSCs), harmonics generation microscopy provides a versatile tool for fibrogenesis and steatosis study. Multimodal images may be performed by a single ultrafast laser source at 1250 nm falling in tissue transparency window. Various grades of HCC are examined revealing fibrosis, steatosis, liver cell dysplasia, activation of HSCs and hepatocyte necrosis, that shows a great ability of multimodal label-free microscopy to intravital visualization of liver pathology development.