Photon absorption remote sensing system for histological evaluation of tissues.

Abstract

The imaging device may be used for histological and / or molecular imaging of tissue samples. The imaging device may include one or more light sources. The one or more light sources generate one or more excitation beams directed to an excitation position focused on the sample to generate signals at the sample, and one or more interrogation beams directed to a detection position, where a portion of the one or more interrogation beams returning from the sample indicates at least some of the generated signals. The imaging device may include a photodetector configured to detect emission signals from the sample. The imaging device may generate an image of the sample using only pressure (photoacoustic) signals, only temperature (photothermal) signals, and / or both photoacoustic and photothermal signals from the generated signals.

Description

[Technical Field]

[0001] The present disclosure relates to the field of optical imaging, and in particular to methods for photon absorption remote sensing (PARS) and non-contact imaging of samples, such as biological tissue, in vivo, ex vivo, or in vitro. [Background technology]

[0002] (CROSS-REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of priority to U.S. Provisional Patent Application No. 63 / 483,854, filed February 8, 2023, which is incorporated herein by reference in its entirety.

[0003] In some fields of imaging, conventional imaging techniques often require that samples be stained or colored before being imaged. For example, in histological imaging, conventional histopathology workflows require that samples be preserved, embedded, and then sectioned into thin, translucent samples before imaging. This process can take days or even weeks. Furthermore, samples prepared in this manner can be stained only once and / or with only one stain set, which often causes these samples to be suitable for only one specific purpose. However, when several evaluations must be performed, many stained samples may be required, each using one specific stained sample. This can complicate and delay the diagnostic pathway, especially in the case of small tissue specimens. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] U.S. Patent No. 11,022,540 Summary of the Invention [Problem to be solved by the invention]

[0005] Thus, there is a need for an imaging technique, architecture, system, or method that is capable of capturing sufficient detail to perform cellular and subcellular assessment of tissue while reducing the need for sample processing and / or staining and avoiding many of the sample collection and preparation challenges associated with traditional pathology workflows, e.g., those in histopathology. [Means for solving the problem]

[0006] Aspects disclosed herein may be used to detect signals from any of the optical absorption or photoacoustic remote sensing systems, methods, or methods described in the following U.S. patent applications, which are incorporated herein by reference: U.S. patent application Ser. No. 16 / 847,182, filed April 13, 2020 (entitled Photoacoustic Remote Sensing (PARS)); U.S. patent application Ser. No. 17 / 091,856, filed November 6, 2020 (entitled Non-Interferometric Photoacoustic Remote Sensing (NI-PARS)); U.S. patent application Ser. No. 16 / 814,538, filed March 10, 2020 (now U.S. Patent Application No. 16 / 814,538) (entitled Camera-Based Photoacoustic Remote Sensing (C-PARS)); U.S. patent application Ser. No. 16 / 753,887, filed April 6, 2020 (entitled Coherence-Gated Photoacoustic Remote Sensing (C-PARS)); No. 16 / 647,076, filed March 13, 2020, entitled Single Source Photoacoustic Remote Sensing (SS-PARS), U.S. Patent Application No. 16 / 629,371, filed January 8, 2020, entitled Photoacoustic Remote Sensing (PARS), and No. 17 / 394,919, filed August 5, 2021 (entitled PARS Imaging Methods), and U.S. Provisional Patent Application No. 63 / 241,170, filed September 7, 2021 (entitled Non-Linear PARS Methods).Embodiments disclosed herein may be used with any of the PARS systems described in the above-referenced applications, such as time-domain PARS or TD-PARS, total absorption PARS or TA-PARS, multi-path PARS or MP-PARS, multiphoton excitation PARS or multiphoton PARS, thermally enhanced PARS or TE-PARS, temperature-sensing PARS or TS-PARS, super-resolution PARS or SR-PARS, spectrally enhanced PARS or SE-PARS, smart detection PARS or SD-PARS, camera-based PARS or C-PARS, incoherent PARS or NI-PARS, coherence-gated PARS or CG-PARS, single-source PARS or SS-PARS, optical resolution PARS or OR-PARS, dual-modality PARS combined with optical coherence tomography (PARS-OCT), and / or endoscopic PARS combined with optical coherence tomography (EPARS-OCT).

[0007] In some embodiments, the techniques described herein relate to an imaging device for histological and / or molecular imaging of a tissue sample, the device comprising: one or more light sources configured to generate (i) one or more excitation beams configured to be directed to an excitation location focused on the tissue sample to generate signals in the tissue sample, and (ii) one or more interrogation beams configured to be directed to a detection location, where a portion of the one or more interrogation beams returning from the tissue sample is indicative of at least some of the generated signals; a photodetector configured to detect emission signals from the tissue sample; and one or more processors. The one or more processors relate to the imaging device configured to perform the following steps: generate an image of the tissue sample using only pressure (photoacoustic) signals from the generated signals; generate an image of the tissue sample using only temperature (photothermal) signals from the generated signals; and generate an image of the tissue sample using both the photoacoustic and photothermal signals from the generated signals.

[0008] In some embodiments, the techniques described herein relate to an apparatus in which the photoacoustic signals used to generate an image of a tissue sample are measured in the range of 1 picosecond to 500 milliseconds of an excitation event caused by one or more excitation beams.

[0009] In some embodiments, the techniques described herein relate to an apparatus in which the photothermal signal used to generate an image of a tissue sample is measured in the range of 1 microsecond to 500 milliseconds of an excitation event caused by one or more excitation beams.

[0010] In some embodiments, the techniques described herein relate to an apparatus, wherein the one or more light sources comprise a first excitation light source configured to emit light at a first wavelength and a second excitation light source configured to emit light at a second wavelength, different from the first wavelength.

[0011] In some embodiments, the techniques described herein relate to an apparatus, wherein the first and second wavelengths of light are configured to target unique radiative and non-radiative absorption properties of localized biomolecules in a tissue sample.

[0012] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more processors are configured to generate an image based on excitation using only a first wavelength and photoacoustic and / or photothermal signals from excitation using only a second wavelength.

[0013] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more processors are configured to generate an absorption differential image based on (1) photoacoustic and photothermal signals from excitation using only a first wavelength, and (2) relative derivatives of photoacoustic and photothermal signals from excitation using only a second wavelength.

[0014] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more processors are configured to generate transmission and reflection attenuation maps via light scattering contrast images of one or more interrogation or excitation beams.

[0015] In some aspects, the techniques described herein relate to an apparatus in which a biomolecule or target of interest appears as a spot that is relatively darker than the surrounding non-absorbing medium in a light scattering contrast image.

[0016] In some embodiments, the techniques described herein relate to an apparatus, wherein the tissue sample comprises one or more of a freshly excised tissue specimen, a preserved tissue specimen, a prepared tissue specimen, an extracted tissue specimen, or an in vivo tissue.

[0017] In some aspects, the techniques described herein relate to an apparatus, the apparatus further comprising a temperature control device configured to regulate the temperature of the tissue sample. In some embodiments, the techniques described herein relate to an apparatus, the apparatus further comprising a slide for containing a tissue sample, the slide comprising a UV transparent material configured to allow imaging through the slide.

[0018] In some aspects, the techniques described herein relate to an apparatus, wherein one or more processors are further configured to calculate an intensity of a generated signal before excitation, determine a residual modulation by subtracting the calculated intensity before excitation from an intensity of a generated signal after excitation, integrate the residual modulation to be integrated, and use the integration to estimate a total absorption level of the radiative or non-radiative signal.

[0019] In some aspects, the techniques described herein relate to an apparatus, wherein the one or more processors are configured to apply noise removal or filtering before extracting the integral.

[0020] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more processors are configured to generate an image using all of the photoacoustic signal, the photothermal signal, and the radiation signal.

[0021] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more processors are configured to generate an image using a quantum efficiency ratio (QER) ratio of (1) photoacoustic and photothermal signals, to (2) radiative signals.

[0022] In some embodiments, the techniques described herein relate to an apparatus in which one or more processors are configured to generate a combined quantum efficiency ratio QER-total absorption image using (i) a quantum efficiency ratio QER ratio to define the color of the combined quantum efficiency ratio QER-total absorption image, and (ii) all of the photoacoustic signal, the photothermal signal, and the radiative signal to define the intensity of the combined quantum efficiency ratio QER-total absorption image.

[0023] In some embodiments, the techniques described herein relate to an apparatus, wherein color provides information about the type of biomolecule in a combined quantum efficiency ratio (QER)-total absorption image, and intensity of the combined quantum efficiency ratio (QER)-total absorption image provides information about the concentration of the biomolecule.

[0024] In some aspects, the techniques described herein relate to an apparatus, wherein the one or more processors are further configured to form a visualization of the extracted time-domain features that distinguishes different biomolecules with different colors.

[0025] In some aspects, the techniques described herein relate to an apparatus that further includes a secondary imaging head, the secondary imaging head being a camera-based detector configured to perform wide-area, high-resolution imaging at a high rate.

[0026] In some embodiments, the techniques described herein relate to an apparatus, wherein the one or more light sources are one or more of: (i) a white light source; and (ii) isolated wavelengths, and the one or more light sources are configured to provide one or more of: (i) a bright field image; (ii) a measurement of light attenuation within the specimen; and (iii) a measurement of autofluorescence within the specimen.

[0027] In some embodiments, the techniques described herein relate to an apparatus, wherein one or more excitation beams and / or one or more interrogation beams underfill an objective lens used for histological and / or molecular imaging of tissue samples.

[0028] In some embodiments, the techniques described herein relate to an apparatus in which one or more excitation beams and / or one or more interrogation beams precisely fill or overfill an objective lens used for histological and / or molecular imaging of a tissue sample.

[0029] In some embodiments, the techniques described herein relate to an apparatus, wherein the one or more processors are further configured to generate an image using an emission signal detected by the photodetector, wherein the emission signal is autofluorescence.

[0030] In some aspects, the techniques described herein relate to an apparatus, wherein the photodetector is configured to detect a non-radiative signal dominated by a temperature (photothermal) signal.

[0031] In some aspects, the techniques described herein relate to an apparatus, wherein the optical detector is configured to detect a non-radiative signal dominated by a pressure (photoacoustic) signal.

[0032] In some embodiments, techniques described herein include a method for configuring a scanning system to scan a sample at different pixels spaced a determined distance from each other, the different pixels corresponding to different locations of an excitation event, the method comprising: performing a first scan at two or more pixels in a region of the sample; and determining from the first scan a determined distance between each pixel, the determined distance corresponding to a minimum distance that allows for extraction of a signal from a particular pixel, whereby the particular pixel is completely illuminated. determining a determined distance at which the sample is entirely isolated and the sample is allowed to return to thermal equilibrium before being excited again; synchronizing an oscillation frequency of the MEMS mirror with a pulse repetition frequency (PRF) of the laser source; and optically scanning a beam generated by the laser source at each pixel in the subgroup across the sample via the MEMS mirror, wherein the oscillation frequency synchronized with the pulse repetition frequency PRF allows the beam to be pulsed at the determined distance onto the sample.

[0033] In some embodiments, the techniques described herein relate to a method further comprising generating one or more additional beams generated by the laser source and / or one or more additional laser sources, each beam of the one or more additional beams corresponding to a distinct subgroup of pixels, and optically scanning each beam of the one or more additional beams across the sample via a MEMS mirror at each pixel in the corresponding subgroup, wherein a synchronized scanning frequency and pulse repetition frequency PRF allows the one or more additional beams to be pulsed at a determined distance onto the sample.

[0034] In some aspects, the techniques described herein relate to a method, wherein the beam and each of the one or more additional beams are separated from each other by a determined distance.

[0035] In some aspects, techniques described herein relate to methods for configuring a scanning system to scan a sample at different pixels spaced a determined distance from one another, the different pixels corresponding to different locations of an excitation event, the method comprising: performing a first scan at two or more pixels in a region of the sample; determining from the first scan a determined distance between each pixel, the determined distance corresponding to a minimum distance that allows for extracting a signal from a particular pixel such that the particular pixel is completely isolated and the sample is allowed to return to thermal equilibrium before being excited again; pulsing multiple beams generated by one or more laser sources at a pulse repetition frequency (PRF), each beam of the multiple beams corresponding to a distinct subgroup of pixels; and optically scanning the multiple beams across the sample via a MEMS mirror at each pixel in the corresponding distinct subgroup, whereby the multiple beams are pulsed at the determined distance onto the sample.

[0036] In some embodiments, the techniques described herein relate to a method, wherein the pulse repetition frequency PRF is greater than the oscillation frequency of the MEMS mirror. In this patent document, the terms "comprises," "comprising," "including," and "having" are inclusive and thus specify the presence of features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, and / or components. Reference to elements by the articles "a," "an," and "the" does not require that there is one and only one of the elements and may also include plural forms unless the context clearly indicates otherwise. The terms "about," "approximately," "substantially," and the like, when used in describing numerical values, indicate a variation of + / - 10% of that value, unless expressly stated otherwise.

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the disclosure and, together with the description, serve to explain the principles of the disclosure. [Brief explanation of the drawings]

[0038] [Figure 1A] 1 shows a graph of the PARS signal for methylene blue at different temperatures. [Figure 1B] 1 illustrates an exemplary last pixel spacing and skip pixel architecture. [Figure 1C] 10 shows a graph of the difference in radiated and non-radiated channels at different high-speed pitches and excitation pulse repetition rates. [Figure 1D] 1 shows an exemplary specimen slide having a barcode. [Figure 1E] 1 illustrates an exemplary PARS shuttle stage system. [Figure 1F] 1 illustrates an exemplary single point scanning architecture using single point scanning. [Figure 1G] 1 illustrates an exemplary multi-point scanning architecture using multi-point scanning. [Figure 1H] 1 illustrates an exemplary correlation process for image reconstruction. [Figure 1I]1 illustrates an exemplary iterative process for image reconstruction. [Figure 1J] 10 shows an exemplary before-and-after comparison of images subjected to image reconstruction. [Figure 1K] 1 shows an etched silicon slide. [Figure 1L] 1 illustrates an exemplary architecture for a multipath PARS detection system. [Figure 1M] 1 illustrates an exemplary multi-point scanning architecture for achieving ultra-fast single-point scanning. [Figure 1N] 1 illustrates an exemplary single detection source and a single ultrafast excitation source for providing an input beam for a sequential multi-point scanning embodiment. [Figure 1O] 1 illustrates several exemplary detection sources and exemplary excitation sources for providing input beams for sequential multi-point scanning embodiments. [Figure 1P] 1 illustrates an exemplary single detection source and an exemplary single excitation source for providing input beams for simultaneous multi-point scanning embodiments. [Figure 1Q] 1 illustrates several exemplary detection sources and an exemplary single excitation source for providing input beams for simultaneous multi-point scanning embodiments. [Figure 2A] 1 illustrates another exemplary architecture for a multipath PARS detection system. [Figure 2B] 10 illustrates another exemplary architecture for a multi-path PARS detection system with an additional measurement of scattered excitation light. [Figure 2C] 10 shows another exemplary architecture for a multi-path PARS detection system having additional measurements of transmitted and reflected excitation light. [Figure 2D] 1 illustrates another exemplary architecture for a multipath PARS detection system using a camera-based detector. [Figure 2E] 1 illustrates another exemplary architecture for a PARS detection system in reflectance mode. [Figure 2F]1 illustrates another exemplary architecture for a PARS detection system in reflectance mode. [Figure 2G] 1 illustrates another exemplary architecture for a PARS detection system in reflectance mode. [Figure 2H] 1 illustrates another exemplary architecture for a PARS detection system. [Figure 2I] 1 illustrates another exemplary architecture for a PARS detection system. [Figure 2J] 1 illustrates another exemplary architecture for a PARS detection system. [Figure 2K] 1 illustrates another exemplary architecture for a PARS detection system. [Figure 2L] 10 shows another exemplary architecture for a multi-path PARS detection system using a camera-based detector with additional measurements of transmitted and reflected excitation light. [Figure 2M] 1 shows another exemplary architecture for a PARS detection system using a camera-based detector with additional measurements of transmitted and reflected excitation light. [Figure 3] 2A-2M show exemplary excitation source architectures for use in any of the PARS detection systems. [Figure 4] 1 shows a graph of a signal energy measurement process. [Figure 5] Total non-radiative absorption PARS images captured using a 266 nm excitation source and a 532 nm excitation source are shown. [Figure 6] Total radiation absorption PARS images captured using a 266 nm excitation source and a 532 nm excitation source are shown. [Figure 7] A total absorption PARS image captured using a 266 nm excitation source is shown. [Figure 8] A total absorption PARS image captured using a 532 nm excitation source is shown. [Figure 9] Total absorption PARS images captured using a 266 nm excitation source and a 532 nm excitation source are shown. [Figure 10] Quantum efficiency ratio PARS images captured using a 266 nm excitation source are shown. [Figure 11] Quantum efficiency ratio PARS images captured using a 532 nm excitation source are shown. [Figure 12] Quantum efficiency ratio PARS images captured using a 266 nm excitation source and a 532 nm excitation source are shown. [Figure 13] Quantum efficiency ratio PARS images with total absorption-based colorization captured using a 266 nm excitation source and a 532 nm excitation source are shown. [Figure 14] 10 shows differential absorption contrast extracted from non-radiative absorption PARS signals captured using a 266 nm excitation source and a 532 nm excitation source. [Figure 15] 10 is an image showing differential absorption contrast extracted from the radiation absorption PARS signal captured using a 266 nm excitation source and a 532 nm excitation source. [Figure 16A] 10 is a collection of images illustrating the k-means feature extraction method applied to thin sections of archived human breast tissue. [Figure 16B] 10 is a collection of images illustrating the k-means feature extraction method applied to thin sections of archived human breast tissue. [Figure 16C] 10 is a collection of images illustrating the k-means feature extraction method applied to thin sections of archived human breast tissue. [Figure 16D] 10 is a set of images and accompanying graphs showing the k-means feature extraction method applied to thin sections of archived human breast tissue. [Figure 17] 16A-16C show colorized images generated from combinations of different feature intensities extracted and presented in FIGS. [Figure 18] 10 is an image showing an alternative feature colorization using two k-means features and a signal energy feature (extracted using the signal energy feature extraction method). [Figure 19] 1 shows graphs of the PARS non-radiative pressure (photoacoustic signal) and temperature (photothermal signal) induced modulations produced in the local optical properties of the sample. [Figure 20A] 10 shows another graph of the PARS non-radiative pressure (photoacoustic signal) and temperature (photothermal signal) induced modulation produced in the local optical properties of the sample. [Figure 20B] 10 shows another graph of the PARS non-radiative pressure (photoacoustic signal) and temperature (photothermal signal) induced modulation produced in the local optical properties of the sample. [Figure 20C] 10 shows another graph of the PARS non-radiative pressure (photoacoustic signal) and temperature (photothermal signal) induced modulation produced in the local optical properties of the sample. [Figure 21] Schematic of an axial depth scan over a range of + / - 2 μm with a step size of 500 nm. [Figure 22] 10 shows a focus plot graph generated by an axial depth scan. [Figure 23] Two annotated PARS images are shown, one showing tissue sample boundary detection / selection and another showing the titling of the tissue sample area into smaller sub-regions. [Figure 24] Whole-slide PARS image containing contrast and brightness variations between adjacent tiles in the scatter channel. [Figure 25] FIG. 25 is the whole-slide PARS image of FIG. 24 after the whole-slide stitching and contrast leveling methods have been applied. [Figure 26] 10 illustrates a graph of a local spatiotemporal filtering implementation. [Figure 27] Two PARS images of a thin section of human skin tissue are shown before the local statistics image smoothing method is applied to the non-radiative contrast channel of the image (pre-filtering) and after the local statistics image smoothing method is applied to the non-radiative contrast channel of the image (post-filtering). [Figure 28]Two non-radiative contrast PARS images and two radiative contrast PARS images of a thin section of human skin tissue are shown before the total absorption dispersion correction method is applied (non-radiative contrast-intrinsic image, radiative contrast-intrinsic image) and after the total absorption dispersion correction method is applied (non-radiative contrast-filtered image, radiative contrast-filtered image). [Figure 29] A non-radiative contrast PARS image captured using a 266 nm excitation source is shown. [Figure 30] A non-radiative contrast PARS image captured using a 532 nm excitation source is shown. [Figure 31] Shown is an emission contrast PARS image captured using a 266 nm excitation source. [Figure 32] Shown is an emission contrast PARS image captured using a 532 nm excitation source. [Figure 33] A light scattering contrast PARS image captured from a 405 nm source is shown. [Figure 34] Shown is an indirect absorption PARS image captured in light scattering contrast with a 405 nm source. [Figure 35] An excitation-scatter contrast PARS image captured from a 266 nm excitation source is shown. [Figure 36A] 1 illustrates different exemplary embodiments of a collection cell. [Figure 36B] 1 illustrates different exemplary embodiments of a collection cell. [Figure 36C] 1 illustrates different exemplary embodiments of a collection cell. [Figure 36D] 1 illustrates different exemplary embodiments of a collection cell. [Figure 37] 1 illustrates an exemplary optical processing unit. DETAILED DESCRIPTION OF THE INVENTION

[0039] The present disclosure relates generally to photon absorption remote sensing (PARS) and to related architectures, systems, and methods. As will be discussed in more detail below, PARS (and related architectures, systems, and methods) can capture multiple direct light absorption (radiative and non-radiative) signals, indirect light absorption signals, and light scattering signals from a biological sample, e.g., a biological (e.g., cell or tissue) specimen. An excitation event may be induced by an excitation source, and subsequent time-evolution relaxation processes may then be captured. Thereafter, multiple signal features (e.g., amplitude, frequency content, phase modulation) may be extracted through processing of at least one of the multiple generated signals. These signal features may then be used to directly form visualizations, may be processed to form visualizations, and / or may be combined into feature vector characteristics (feature vector characteristics, feature quantity vector characteristics), and the feature vector characteristics may be used in further processing. In such cases, the feature vector characteristics may be used for a variety of applications, e.g., to enhance AI diagnostic tools or to develop AI-based colorization.

[0040] <PARS mechanism> PARS is an all-optical non-contact absorption microscopy technique. PARS uses excitation and detection laser sources to generate and detect optical absorption and scattering contrast in a variety of samples.

[0041] As will be further discussed below, an excitation source, e.g., an excitation laser, is used to accumulate optical energy within the sample. When light is absorbed by a chromophore, the energy causes the sample to become excited. In some cases, e.g., in inelastic scattering events (e.g., stimulated Raman or Brillouin scattering events), only a portion of the energy of the incident photon is captured. Depending on the amount of energy captured from the absorbed light, the sample can assume any of a different energy state, e.g., different vibrational, virtual, or electronic states. This absorbed energy can then dissipate through either optical emission (radiation) or non-radiative relaxation.

[0042] During nonradiative relaxation, the absorbed light energy is converted to heat. This heating causes modulations in the local material properties of the specimen by affecting material density, stress, etc. These temperature changes and corresponding optical property modifications are the basis of photothermal microscopy. In certain circumstances, the generation of heat can also cause additional effects within the heated region. In some cases, rapid heating and subsequent thermoelastic expansion can result in the generation of pressure. This is referred to as photoacoustic pressure, and photoacoustic pressure is the basis of photoacoustic microscopy. One effect caused by the generated pressure causes corresponding modulations in the sample's local material properties by affecting material density, stress, etc. The generated pressure can also cause additional effects, such as ultrasound absorption, which can further modulate material properties. In PARS, these local (nonradiative) temperature and pressure modulations are observed using a detection source. Perturbations in the sample's optical and material properties may be captured at the detection laser as backscattered time-evolving intensity modulations.

[0043] During radiation relaxation, absorbed light energy is released via the emission of photons. Typically, the emitted photons exhibit a different energy level compared to the absorbed photons. Modalities such as fluorescence microscopy, multiphoton fluorescence, or harmonic generation rely on radiation relaxation effects for contrast. In PARS, radiation contrast may be captured by measuring the emission of non-excitation photons from the specimen.

[0044] During inelastic scattering events, photons exhibit different energy levels compared to the input photons. Modalities such as spontaneous or stimulated Brillouin scattering or spontaneous or stimulated Raman scattering rely on these events for contrast. In PARS, inelastic scattering contrast can be captured by measuring photons scattered from the sample, which exhibit different energies compared to the input photons.

[0045] Overall, it can be seen that PARS can simultaneously capture nearly all optical properties of a chromophore. This includes non-radiative and radiative direct light absorption contrast and their effects, in addition to the indirect light absorption provided by the scattering signals of the excitation and detection beams. This can enhance inherent contrast and visualization. Furthermore, PARS can provide enhanced sensitivity to any range of chromophores (e.g., 100 nm to 16 μm) compared to other independent modalities. Unlike traditional methods that capture radiative or non-radiative absorption independently, PARS contrast may not be limited by common efficiency factors, such as photothermal conversion efficiency or fluorescence quantum yield.

[0046] A detailed discussion of the PARS detection mechanism is provided here. As noted above, all optical interactions of the detection and excitation beams with the sample may be collected by PARS. This includes indirect and direct absorption effects, linear scattering, nonlinear scattering, etc. These features may then be used to generate a characteristic feature vector of a biomolecule or mixture of biomolecules located at the focus of the beam.

[0047] In general, optical interactions captured by PARS can be classified into both elastic and inelastic light scattering, as well as optical absorption contrast. Absorption effects are further classified into radiative and nonradiative events based on the radiative and nonradiative relaxation processes. During nonradiative relaxation, absorbed light energy is converted to heat. During radiative relaxation, absorbed light energy is released via the emission of photons. In most cases, optical interactions and their subsequent effects are not isolated. That is, most interactions, except for elastic scattering events, induce some radiative and some nonradiative relaxation characteristics. For example, stimulated Raman scattering, an inelastic scattering effect, can induce some nonradiative relaxation.

[0048] With respect to absorption events, there are several different absorption mechanisms that can contribute to PARS. For example, both linear (e.g., single-photon) and nonlinear (e.g., multi-photon, e.g., two- or three-photon) absorption interactions can induce and / or generate PARS signals. Other effects, such as stimulated Raman absorption, vibrational absorption, electronic absorption, and surface resonance plasmon absorption, may also be utilized. In a single excitation event, any number of these or other effects can contribute to the absorption that induces and / or generates PARS radiative and non-radiative signals.

[0049] PARS can capture non-radiative contrast using a detection source(s). All non-radiative relaxations cause perturbations in the local physical and optical properties of the analyte. These time-evolving perturbations are then captured by observing their effect on the interaction of the detection laser with the analyte.

[0050] In some cases, nonradiative relaxation processes captured by PARS may exhibit temperature-based (e.g., photothermal) effects. In these cases, the nonradiative signal may be considered to be dominated by the photothermal signal. Although the sample may generate pressure, the PARS detection mechanism may not collect the pressure signal because the detection mechanism is not fast enough. Therefore, the collected signal may be primarily photothermal (e.g., dominated by the temperature-based signal). For example, in cases where a nonradiative relaxation event occurs, the accumulated energy results in localized heating and subsequent generation of thermoelastic expansion. This is the fundamental effect exploited in conventional modalities, such as photothermal microscopy, time-domain photothermal microscopy, thermal lensing, thermal lens microscopy, and photothermal deflection microscopy. Localized heating induces changes in material properties, such as density, birefringence, refractive index, absorption coefficient, or scattering behavior.

[0051] In some cases, non-radiative relaxation processes captured by PARS may exhibit pressure-based (e.g., photoacoustic) effects. In these cases, the non-radiative signal may be considered dominated by the photoacoustic signal. This is because the pressure-based signal can be much more (e.g., stronger, larger, more dominant, etc.) than the temperature-based signal. PARS may collect a pressure-based signal that is two, three, or even more times larger than the temperature-based signal. For example, if a specimen is heated quickly enough, the material is constrained to undergo a heat-induced thermoelastic expansion. The heated region then increases its pressure as it resists this constraint. This results in the generation of photoacoustic pressure. Generally, this initial pressure is generated when energy accumulates faster than the stress confinement time, which is the time it takes for pressure to propagate from the excitation region. Therefore, depending on the PARS architecture (e.g., device or system) and specimen / sample parameters, the pressure confinement time can range on the scale of femtoseconds to microseconds.

[0052] The initial pressure generation can cause many effects. The dominant effect of the initial pressure generation can include initial pressure light modulation, stress, deformation / displacement, or cavitation. When the dominant effect of the initial pressure generation is initial pressure light modulation, the large megapascal-scale initial pressure generated during the excitation event can directly affect the local optical (e.g., refractive index, scattering cross section) and material properties (e.g., density, heat capacity) of the specimen. When the dominant effect of the initial pressure generation is stress, the initial pressure generation can cause stress and strain fields within the specimen. These can cause changes in the optical properties (e.g., polarizability, reflectivity) of the specimen. When the dominant effect of the initial pressure generation is deformation / displacement, the pressure generation can cause deformation and displacement of the specimen's scatterers, surface, or subsurface features, which results in distortion and modification of the scattering profile. When the dominant effect of the initial pressure generation is cavitation, the generation of the large initial pressure and the subsequent negative pressure can result in cavitation. Cavitation can cause rapid modulation in the local optical properties at the detection laser source location(s), which results in an observable PARS signal.

[0053] The additional effect of the initial pressure generation can include ultrasound absorption, resonance, secondary reflections, surface oscillation, or vibration. When the additional effect of the initial pressure generation is ultrasound absorption, the initial pressure induced by the excitation event propagates as ultrasound waves away from the excitation location. These high-frequency ultrasound waves are attenuated by the surrounding sample, resulting in the generation of localized heating. When the additional effect of the initial pressure generation is resonance, depending on the frequency of the generated pressure, the excitation event can induce resonant vibrations within the specimen, resulting in observable modulation. When the additional effect of the initial pressure generation is secondary reflections, after the initial pressure generation, the pressure wave can interact with an interface within the specimen, resulting in secondary reflections of the ultrasound waves. These secondary reflections can have an effect similar to the primary pressure interaction after some time delay. When the additional effect of the initial pressure generation is surface vibration, the acoustic signal propagating to the surface of the sample can generate surface vibrations, which can also induce observable changes in the PARS signal (e.g., phase modulation, intensity modulation). In cases where an additional effect of the initial pressure generation is vibration, the generated pressure may also induce effects such as scatterer surface or subsurface position modulation / vibration.

[0054] In some cases, non-radiative relaxation processes captured by PARS can exhibit temperature (i.e., photothermal) and pressure (i.e., photoacoustic) based effects. Many of the listed effects are not completely independent phenomena, and as a result, there can be interactions between mechanisms and effects on the material and optical properties of the analyte. Pressure and temperature induced from previous and current excitation events can result in measurable differences in the current PARS excitation and the PARS time-domain signal excitation. This may be manifested via variations in local material and optical properties, such as density, birefringence, refractive index, absorption coefficient, scattering behavior, etc.

[0055] At the same time, differences in the sample and scanning environment can have an effect on the measured PARS signal. For example, since material properties depend on the ambient temperature, there is a corresponding temperature dependence in the PARS signal. Another example can include the introduction of mechanical stress (e.g., bending) to the sample, which can then affect the material properties of the sample (e.g., density or local optical properties such as birefringence, refractive index, absorption coefficient, scattering behavior). This can perturb the generated PARS signal compared to one generated without introducing this mechanical stress.

[0056] Traditionally, non-radiative relaxation effects are observed as backscattering or transmission amplitude modulations at the detection source. These amplitude modulations can be caused by the effects mentioned above, such as refractive index modulation, scatterer motion, surface deformation, etc. However, there are many other aspects of the detection beam that can be used. For example, other optical features that can be exploited can include polarization, phase, or frequency.

[0057] The PARS nonradiative relaxation signal may also have a dependence on some aspects of the detection laser, such as wavelength, spectral linewidth, coherence length, beam size, or fluence, because the modulation signal is indirectly observed through detection. The observed nonradiative signal may also depend on the interaction of the detection beam with the sample. This means that the signal may have some dependence on properties such as sample temperature, scatterer size, sample morphology, conductivity, or density. For example, some of the scattering, polarization, frequency, and phase content in a PARS signal may be due to the size, shape, features, and dimensions of the region that generated the signal. These details may be exploited to recover unique information, which may be used to improve the final image fidelity, classify sample regions, determine the size of constituent chromophores, and classify constituent chromophores, to name a few.

[0058] In some cases, the detection source intensity is measured directly using a single intensity sensor, such as a photodiode, avalanche photodiode, or photomultiplier tube. Other embodiments may use multiplexed sensors to simultaneously measure detection interactions at multiple locations. This may include line array sensors (e.g., linear photodiodes), avalanche photodiode arrays, or two-dimensional sensors (e.g., CMOS, CCD, or SPAD sensors). Exemplary detection sources and methods are described in U.S. Patent No. 6,229,999 (C-PARS), which is incorporated herein by reference in its entirety.

[0059] In PARS, non-radiative absorption measurements may also be simultaneously obtained using complementary techniques. In one example, the pressure-induced effects of non-radiative relaxation may be captured using conventional acoustically coupled ultrasound transducers, such as piezoelectric sensors, capacitive micromachined ultrasound transducers (CMUTs), or Fabry-Perot sensors. Other methods, such as air-coupled ultrasound detectors or optical ultrasound detectors (e.g., speckle decorrelation measurements), may also be used. In another alternative example, thermal relaxation may be captured by visualizing the blackbody radiation emitted from a heated sample. An optical detector measures infrared radiation to determine local heating of the specimen, which may then be used to measure the thermal relaxation of the excitation event.

[0060] Radiation relaxation and inelastic scattering can be captured by measuring the emission of non-excitation photons from the specimen. There are several main effects and numerous related phenomena that can be characterized as radiative relaxation. Some examples of effects that can be induced by absorption of any of the light beams include, but are not limited to, stimulated emission at the detection wavelength, stimulated Raman scattering, spontaneous Raman scattering, coherent anti-Stokes Raman scattering, harmonic generation microscopy (e.g., second or third harmonic), (auto)fluorescence and (auto)fluorescence lifetime, or Brillion scattering.

[0061] The radiation and inelastic scattering effects may be captured as aggregate measurements encompassing all light emissions from the specimen, or as isolated measurements that specifically target individual radiation and inelastic scattering effects. As discussed below, the effects may be separated by methods such as timing or color or polarization-based filtering. The isolated contrast may be observed on separate photodetectors or may be captured on a single detector with a time delay.

[0062] Camera-based radiation and inelastic scattering signals may be measured using a variety of photodetectors, such as CMOS sensors, CCD sensors, photodiodes, avalanche photodiodes, or photomultiplier tubes. In some examples, the photodetectors may be arranged as either a single sensor or an array of sensors. Depending on the sensor array, radiation and inelastic scattering emissions may be measured simultaneously at two or more locations by using spatially multiplexed detectors. In some cases, the same radiation and inelastic scattering signals may be captured on several different or similar photodetectors to facilitate noise reduction methods that utilize measurement redundancy. In other cases, radiation and inelastic scattering emissions may be separated by color, with each color band measured on an individual photodetector or on a different region of a multiplexed photodetector. This may provide a measurement of the intensity of radiation and inelastic scattering emissions versus the wavelength of the emission.

[0063] In addition to direct absorption measurements captured via radiative and nonradiative absorption contrast, PARS can also assess optical absorption via indirect methods. For example, the scattering intensity of the excitation and detection beams has a dependence on local absorption characteristics. PARS can then capture indirect absorption contrast using the light scattering signal to estimate the level of absorption at a given wavelength. The wavelength range can vary widely. In some embodiments, the wavelength can be 100 nm to 500 nm. In other embodiments, the wavelength can be 500 nm to 1 μm. In further embodiments, the wavelength can be 1 μm to 10 μm. Some embodiments can have wavelengths in the range of 10 μm to 16 μm. Generally, the wavelength can range from 100 nm or about 100 nm to 16 μm or about 16 μm. Indirect absorption can be measured using any type of photodetector or by measuring the transmitted or reflected light intensity and calculating the indirect optical absorption. In some embodiments, this can be done using a photodiode, an avalanche photodiode, or a photomultiplier tube. In other cases, multiplexed detectors may be used, such as photodiode arrays (linear or two-dimensional), avalanche detector arrays (linear or two-dimensional), CCD sensors, CMOS sensors, or SPAD sensors. Figures 2D, 2G, 2K, 2L, and 2M show example architectures for camera-based systems.

[0064] In addition to the light absorption effects described above, PARS is also enabled to capture the light scattering contrast at each excitation and detection wavelength. In these embodiments, any number of different light scattering features may be evaluated for each wavelength. Some techniques may monitor scattering intensity over any range of angles or across scattering distributions / profiles. Additional features, such as polarization, frequency, and phase content, may also be evaluated within each scattering signal. Such evaluation may be performed in transmitted or reflected light, or at any observation angle that is reasonable for a given architecture. This means that PARS can provide equivalent contrast to traditional scattering modalities, such as differential interference contrast microscopy, optical coherence tomography, and laser interference microscopy. Scattering may be measured using any type of photodetector or by measuring the transmitted or reflected intensity of light. In some embodiments, this may be done using a photodiode, an avalanche photodiode, or a photomultiplier tube. In other cases, multiplexed detectors may be used, such as photodiode arrays (linear or two-dimensional), avalanche detector arrays (linear or two-dimensional), CCD sensors, CMOS sensors, or SPAD sensors. Figures 2D, 2G, 2K, 2L, and 2M show example architectures for camera-based systems.

[0065] <PARSアーキテクチャおよびシステム> PARS, in some embodiments, can be a point-scanning microscope. That is, when forming a PARS image, the measurement point is required to move relative to the sample. Two different methods for performing the task of moving the measurement point are considered in this disclosure: mechanical scanning and optical scanning.

[0066] In mechanical scanning, the optical excitation spot remains stationary while the sample physically moves across the optical spot. In this case, collecting the signal can be a simpler process because all signals originate from a stationary position. The detection system can be focused on this position, so that all optical signals are collected and mapped to the detector. Examples of such are shown in Figures 2A-2M.

[0067] In optical scanning, a PARS interrogation point is scanned across the sample using an optical scanning device (e.g., a microelectromechanical system ("MEMS") mirror, a galvanometer mirror, and a polygon mirror). Examples of optical scanning are shown in Figures 1F-1G. In these cases, collecting the PARS optical signal can be non-trivial.

[0068] As the PARS interrogation spot moves optically across the sample, additional consideration must be given to collecting signals from different locations. Collection cells can be added to collect optical signals from different locations across the sample and redirect these signals for analysis. Collection cells can use any combination of optical components (e.g., mirrors, lenses, etc.) to collect optical signals from different physical locations and redirect them to the appropriate location for measurement. Exemplary embodiments of collection cells are shown in Figures 36A-36D.

[0069] The collection cell 700 shown in Figure 36A includes an input objective lens 710, a sample 708, a receiving objective lens 706, a mirror 704, and an optical fiber 702 (single-mode or multimode). The input objective lens 710 optically scans the sample 708 and directs the scan to the receiving objective lens 706 to collect the optical signal. The receiving objective lens 706 then directs the signal through a series of mirrors 704. The signal is finally redirected into the optical fiber 702, which can carry the light to different locations where an optical processing unit is used to split and measure the different PARS signal components. An exemplary processing unit that can do so is shown in Figure 37.

[0070] The collection cell 712 shown in Figure 36B includes an input objective lens 722, a sample 720, a receive objective lens 718, a mirror 716, and a photodetector 714. The input objective lens 722 optically scans the sample 720 and directs the scan to the receive objective lens 718 to collect the optical signal. The receive objective lens 718 then directs the signal through a series of mirrors 716. The signal is finally redirected to the photodetector 714, which is enabled to directly measure the optical signal derived from the sample 720. In this case, different optical filters (e.g., wavelength filters, power filters, polarization filters, etc.) may be added to the beam path to separate different portions of the PARS optical signal for measurement.

[0071] The collection cell 724 shown in Figure 36C includes an input objective lens 734, a sample 732, a receive objective lens 730, mirrors 728, and a camera (or other multiplexing array) 726. The input objective lens 734 optically scans the sample 732 and directs the scan to the receive objective lens 730 to collect the optical signal. The receive objective lens 730 then directs the signal through a series of mirrors 728. The signal is finally redirected to the camera 726, which can perform measurement functionality substantially similar to the photodetector 714 in Figure 36B while providing an additional dimension of spatial resolution. Thus, the camera 726 can help reveal where on the sample 732 an interrogation event occurred.

[0072] The collection cell 736 shown in FIG. 36D includes an input objective 746, a sample 744, a receiving objective 742, mirrors 740, and an optical system 738. The input objective 746 optically scans the sample 744 and directs the scan to the receiving objective 742 to collect the optical signal. The receiving objective 742 then directs the signal through a series of mirrors 740. The signal is finally redirected to the optical system 738. In this case, the collection cell 736 may be designed to "pre-process" the light to make it compatible with multiple optical systems, such as the optical processing unit described in FIG. 37, where the optical system is used to split and measure different PARS signal components. Thus, the optical system 738 can be any type of optical system configured to process optical signals.

[0073] The optical processing unit or PARS signal detection cell 800 shown in FIG. 37 includes spectral filters 802 and 806, one or more mirrors 804, a beam input 808, a photodetector 810, and a photodetector 812. The PARS signal detection cell 800 separates and measures radiative and non-radiative signal intensities. The photodetector 810 may be configured to provide radiative detection, and the photodetector 812 may be configured to provide non-radiative detection. A collection cell (e.g., collection cell 700 or collection cell 736) may be configured to provide the beam input 808 to the PARS signal detection cell 800. Furthermore, the PARS signal detection cell 800 may be substantially similar to the transmission side of any one of the architectures shown in FIGS. 2A-2D and 2H-2M.

[0074] The collection cell may be even more important for transmission-mode PARS imaging because, in reflection-mode imaging, the light beam propagates along the same path when traveling to and returning from the sample. That is, the beam passes back through the optical scanning system upon reflecting off the sample. As the beam passes back through the optical scanning system in the opposite direction, the beam can perform essentially the same function with the collection cell: mapping interrogation points across the sample and returning to the same location for detection.

[0075] The present disclosure also provides PARS architectures and systems for imaging samples, and in some cases, for histological and / or molecular imaging of any form of sample. For example, the disclosed PARS architectures and systems may be used for label-free histological imaging of unprocessed and / or unstained / unlabeled samples. As used herein, the terms "unlabeled" and "unlabeled" refer to tissue that has not been stained with stains used in histology. For example, unlabeled tissue has not been stained with stains such as hematoxylin, eosin, acid dyes, basic dyes, periodic acid-Schiff reaction stains, Masson's stain, Alcian blue stain, Van Gieson stain, reticulin stain, Giemsa stain, toluidine blue stain, silver and gold stain, chrome alum stain, hematoxylin stain, isamine blue stain, osmium stain, PAS, T-blue, Congo red, or crystal violet. In other instances, the sample may be unstained or essentially unstained with any stain and / or may be free of any stain.

[0076] Suitable samples may include any biological specimen (liquid or solid), including any combination of histological or cytological specimens. Histological or cytological specimens may be, but are not limited to, cellular tissue specimens, freshly excised tissue specimens (i.e., tissue smears, cytological samples, endoscopic tissue biopsies, core needle biopsies, liquid tissue biopsies, gross surgical resections), preserved tissue specimens (i.e., formalin-fixed tissues or cells, ethanol-fixed tissues or cells, acetic acid-fixed tissues or cells), prepared tissue specimens (i.e., formalin-fixed paraffin-embedded tissues or cells, formalin-fixed paraffin-embedded thin tissue sections, frozen sections), and / or extracted tissue specimens (i.e., purified protein samples, cell cultures). Thin tissue specimens may be from tissue smears as thin as single cells, standard frozen sections, standard paraffin-embedded slides, standard paraffin-embedded tissue blocks, or may be as thick as bulk unprocessed freshly excised tissue of any thickness. Additionally, the penetration depth of the optical system may determine the preferred tissue thickness.

[0077] In some instances, to further facilitate imaging, the sample may be mounted on a slide. The slide may be formed from any material, and the material may be configured to allow imaging through the slide, to allow tissue to lie flat against the slide to reduce or prevent surface variations, to allow imaging of liquid samples and / or samples not fully fixed to the slide, to reduce or prevent sample dripping or spillage, and / or to be compatible with fresh tissue and fixed slides. For example, the material may include various types of glass, such as UV-transparent materials (e.g., quartz or UV-fused silica). In certain cases, the slide may be configured to allow temperature control of the slide and / or to stabilize the direct current (DC) value of the signal. In these cases, the PARS architecture and system may include a heater or thermoelectric cooler (TEC) (e.g., a Peltier device or refrigerator) to control the temperature of the slide so that the direct current (DC) value of the signal may be more stable and may return to the same direct current (DC) value, and / or to promote contrast stability over time. The temperature control device is configured to regulate the temperature of the slide. Figure 1A shows a graph of the PARS signal for methylene blue at different temperatures, with modulation plotted on the y-axis and time plotted on the x-axis.

[0078] In some examples, the PARS system may be used for in vivo applications. For example, a histological sample may be replaced by any in vivo target. Thus, the present disclosure contemplates using an in vivo target in place of a histological sample in any of the embodiments described herein. An exemplary in vivo application may include placing a patient's hand under a microscope to perform the various functions described in this disclosure. Another exemplary in vivo application may include placing a microscope on a surgical arm and then placing the surgical arm on a patient, again allowing the various functions described in this disclosure to be performed.

[0079] In some examples, a PARS system may feature a secondary imaging head that uses a camera-based detector to perform wide-area, high-resolution imaging at high speeds. The system may have multiple illumination sources and detection filter options to provide different contrast within the specimen. For example, illumination may be selected as a white light source to provide a "bright-field image," as isolated wavelengths (UV-IR) to provide measurements of light attenuation within the specimen, or as isolated wavelengths (UV-IR) to provide measurements of autofluorescence within the specimen. These images may be used to guide PARS imaging or to enhance PARS contrast. For example, when combined with methods such as automatic edge detection, these images may be utilized to select only regions of interest for PARS scanning. Alternatively, fluorescence or attenuation measurements may be registered with the PARS image and added to the PARS data vector collected at each pixel.

[0080] When the secondary imaging head is used for margin assessment, the goal is to identify whether the extracted surface is cancer-free. More specifically, the goal is to conclude whether the extracted surface is cancer-free as quickly as possible for margin assessment. In this design, the secondary imaging head indicates high-probability cancerous regions, and the PARS imaging head is then enabled to scan these regions and evaluate positive or negative regions. The secondary imaging head is enabled to indicate high-probability cancerous regions because it can quickly provide an outline of the tissue and enable the user to automatically (or manually) select the perimeter of tissue that appears cancerous. Confirmation can later be received by the PARS primary imaging head (or precision optical head). Additionally, the secondary imaging head is enabled to identify coarse features, such as nuclei-dense tumor tissue, and bulk tissue deformation, such as necrosis or inflammation. These evaluations may be performed at a lower resolution of approximately 10-15x magnification. This modality is useful in samples with large sample areas, for example, in breast cancer margin assessment. In other applications where the sample is smaller, such as Mohs, it may be more efficient to scan the entire tissue at the highest resolution of the PARS imaging head without the need for a secondary imaging head.

[0081] Based on the images and analysis from the secondary imaging head, the outcome can be to indicate which excitation wavelengths to use in specific areas of tissue for scanning in the PARS cell to identify biomolecules of interest. By targeting the required excitation wavelengths, the time scan in the PARS head can be reduced and then better focused on the required analysis.

[0082] As discussed below, the disclosed PARS architecture and system may include at least one excitation and detection beam. There may be two or more excitation spots and / or one or more detection spots. The excitation and / or detection beams may originate from different laser sources, or the excitation and / or detection beams may be derived as portions of the same laser source or as harmonics of the same laser source. Each of the excitation and detection beams may be either pulsed or continuous, with pulse widths ranging from femtoseconds to microseconds, or in some cases even longer. It is understood that the disclosed PARS architecture and system may be designed and / or optimized in light of desired imaging results. Accordingly, the PARS architecture and system possess several unique design features for providing substantially enhanced sensitivity in histological samples.

[0083] The PARS system may feature specific optimization of the non-radiative detection path, which enhances sensitivity. For example, the detection and / or excitation beam may be made to intentionally underfill the objective lens. This means that the beam is smaller than the intended input beam for the optical lens. In this case, PARS detection may provide a loose forward focus while achieving high-efficiency collection of backscattered intensity modulation. This design may make PARS detection more sensitive to non-radiative modulation compared to completely filling the objective lens. Alternatively, the length of the detection path, e.g., the distance path between the sample and the photodetector, may be intentionally stretched or compressed. When the path length is adjusted in this manner, the path length may provide enhanced sensitivity to variations in the backscattered non-radiative modulation, which may change the detection beam path. In certain cases, a longer path may provide greater sensitivity to smaller modulations in the beam. In addition, spatial filtering, e.g., a pinhole, may be added to the non-radiative detection path to provide enhanced sensitivity to small non-radiative modulations in the detection beam.

[0084] In other embodiments, the detection and / or excitation beams may be made to intentionally overfill the objective lens. In contrast, the beam is larger than the intended input beam for the optical lens. The detection and / or excitation beams may also be made to exactly fill the objective lens.

[0085] Some embodiments of the PARS system of the present disclosure may utilize a multipath non-radiative detection architecture, such as the multipath PARS system described in International Application No. PCT / IB2022 / 054433, which is incorporated herein by reference.

[0086] Generally, in a multi-path detection architecture, the captured detection beam returning from the sample is redirected back to the sample. In some embodiments, the recaptured detection beam may be aligned to be confocal with the initial detection spot and the excitation spot, although non-confocal embodiments are contemplated as being encompassed by this application. The result is that the detection beam repeatedly interacts with the excitation modulation spot multiple times, obtaining additional information from the excitation spot each time it interacts with the sample. Thus, in some cases, the non-radiative PARS signal may then be nonlinearly broadened. In other words, in a multi-path detection system, the detection may be redirected to interact with the sample any number of times, resulting in a corresponding degree of nonlinear broadening in the non-radiative absorption contrast.

[0087] A detailed discussion of a fast-scan "skip pixel" architecture that enables high-speed PARS scanning is provided herein. In a PARS architecture, the signal and resulting contrast (both emission and non-emission channels) at each pixel are related to the excitation laser event at the pixel of interest and the corresponding light absorption from the tissue. To maximize image quality, it is desirable to extract the PARS signal at each pixel based solely on the local light absorption characteristics at the spot of interest (e.g., the focal spot of the excitation and detection beams) without contamination by the surrounding thermal environment. In addition to achieving high image quality, it is also desirable to rapidly image the tissue, which can be achieved by increasing the repetition rate of the excitation laser.

[0088] For example, in an exemplary best case, each pixel is completely isolated, and the sample is allowed to fully return to thermal equilibrium before being excited again. If excitation events overlap spatially (e.g., occur spatially close together within the excitation region) and temporally (e.g., occur before a sufficient relative time has passed), the non-radiative signal will "stack" on top of the previous excitation event, providing an erroneous reading. That is, the signal is affected by the previous excitation event on which the signal was "stacked," and is not a true reading from the sample. Thus, there is essentially 0% or near 0% signal overlap.

[0089] For an exemplary worst case scenario with two overlapping excitation events, 50% of the measured signal from the second measurement is actually from the first event (e.g., two modulations in amplitude, one perfectly overlapping, providing two measured amplitudes, one from the first event and one from the second event).

[0090] The high-speed pitch outlined here is the spatial separation required to ensure there is no pixel overlap. Because a portion of the PARS signal is thermally based, dramatically increasing the laser excitation repetition rate can result in both tissue damage and interference with individual pixel signal extraction and its associated PARS amplitude and time-domain signal decay. Due to the energy flux of the excitation laser spot, heat propagates to adjacent tissue regions outside the intended target pixel area. The heat-affected zone is a function of tissue type, thickness, and paraffin embedding technique (e.g., paraffin type).

[0091] It is often necessary to "skip pixels" between successive excitation pulse events to ensure that "new pixels" are not thermally affected by the previous excitation event. Skipping pixels may be necessary from an image quality perspective and to protect the sample from irreversible thermal damage. The ideal distance for separating pixels along the fast axis is referred to as the "fast pitch." In embodiments where the fast pitch is larger than the final pitch corresponding to the desired final resolution, the system must perform multiple scans of the tissue to fill in the gaps in the fast pitch and ultimately obtain the final pixel pitch and resolution, which is nominally approximately 250 nm for PARS systems. Pixel pitch can range from a few nm (e.g., 1-4) to several microns (e.g., 3-7), depending on the desired resolution. Typical pathology systems can range from 25 nm pixels to 10 μm pixels. An example of skipped pixels and the "fast pitch" axis is shown in Figure 1B.

[0092] FIG. 1B illustrates a final pixel spacing 130 and a skip pixel architecture 132. The final pixel spacing 130 includes sample pixels 134 on a portion of a tile 136 at a spacing equal to half the resolution at the Nyquist rate. The skip pixel architecture 132 includes sample pixels 134 with isolated spacing due to skip pixels 138 between them. The spacing between sample pixels 134 may be a fast pitch. In this exemplary embodiment shown in FIG. 1B, the tile 136 has a length of 500 μm in the x-axis (i.e., the "slow axis") and a width of 200 μm in the y-axis (i.e., the "fast pitch" axis). For this exemplary embodiment, eight sweeps are required in the slow axis to completely fill the 250 nm final resolution pitch, which is shown as eight sample pixels 134 on the tile 136 at a fast pitch of 2 μm.

[0093] The ideal fast pitch is the smallest distance between two pixels (n and n+1) such that the SNR and signal waveform from pixel n+1 are not affected by an excitation event at pixel n. Figure 1C shows in section 140 the difference in the radiating and non-radiating channels at fast pitches greater than or equal to the ideal pitch rate at a slow (e.g., 50 kHz) excitation pulse repetition rate or frequency ("PRR" or "PRF," respectively), and in section 142 the difference in the radiating and non-radiating channels at fast pitches less than the ideal pitch rate at a faster (500 kHz) excitation pulse repetition rate ("PRR"). Section 144 shows the difference in radiating and non-radiating channels at high-speed pitches significantly exceeding the ideal pitch rate at a slow (50 kHz) excitation pulse repetition rate or frequency ("PRR" or "PRF," respectively), and section 146 shows the difference in radiating and non-radiating channels at high-speed pitches equal to or greater than the ideal pitch rate at a faster (500 kHz) excitation pulse repetition rate or frequency ("PRR" or "PRF," respectively). The ideal high-speed pitch is correlated to the sample itself and varies with the sample's mechanical properties. Therefore, there are different methods for determining high-speed pitch. These methods may or may not be combined and may be theoretical, empirical, or a combination of theory and empirical.

[0094] One way to determine the high-speed pitch is to scan the sample's barcode, which contains information such as the wax type (e.g., paraffin type) and sample thickness. An exemplary sample 148 having a barcode 150 is shown in FIG. 1D. Scanning the barcode 150 can provide advance information about the thermal properties of the tissue. For example, the barcode 150 can contain information about sample preparation parameters, tissue thickness, paraffin type, tissue type, etc. The mechanical properties (e.g., density, conductivity, etc.) of the tissue sample (148) can be used to calculate the ideal point spacing. In other words, the correct high-speed pitch in the scanning direction can be calculated so as not to damage the sample 148 or negatively affect signal extraction.

[0095] Another method for determining the optical high-speed pitch is to experimentally measure the time-domain signature of the sample being imaged. Thermal decay characteristics may be measured at a single pixel, or the ideal high-speed pitch between pixels may be determined by correlating the time decay to a pre-calculated thermal model (e.g., a thermal FEA model). A different way to experimentally determine the optical high-speed pitch is to perform full scans at various pitches in a small area and experimentally determine a pitch that does not cause thermal interference. This is done in areas of the sample that lie on the edge of the sample, or in clinically low-value areas of the sample as determined by an overview camera, so as not to damage critical aspects of the sample under test.

[0096] FIG. 1E illustrates a PARS shuttle stage system 152 in accordance with one or more embodiments of the present disclosure. More specifically, FIG. 1E illustrates the PARS shuttle stage system 152 as a sample tile 170 is positioned in an overview camera cell 154, a lamp-based camera scanner or secondary imaging head (as described herein) cell 156, and a PARS cell 158. FIG. 1E additionally illustrates a shuttle stage 160 in an isometric view. The PARS shuttle stage system 152 includes a sample load area 162, an overview camera 164, a secondary imaging head 166, and a PARS system 168. The shuttle stage 160 may be configured to enable movement of the sample tile 170 between three different cells. As shown, the sample loading area 162 and overview camera 164 may be at one end of the shuttle stage 160, the PARS system 168 may be at the other end of the shuttle stage 160, and the secondary imaging head 166 may reside between the sample loading area 162 (and overview camera 164) and the PARS system 168. In the event that the sample thickness of the sample is unknown or not recorded in the barcode information, the overview camera 164 may determine the sample thickness by focusing on the top of the tissue and the bottom of the slide. The overview camera 164 may be fixed such that the sample tile 170 is movably engaged with the overview camera 164 via a mechanical stage (e.g., the shuttle stage 160). If the depth of focus of the overview camera 164 is too large to distinguish tissue thicknesses, different non-contact optical-based thickness measurements may be implemented (e.g., laser triangulation, ultrasound thickness measurement, time-of-flight camera, interferometry, etc.).

[0097] One method for enabling a fast-scanning architecture uses a high-speed mechanical stage in one axis that is time-matched with the excitation repetition rate. An example of a fast-scanning mechanical stage is a voice coil motor (VCM) or actuator. In this method, the stage velocity must be time-matched with the excitation repetition rate to enable the desired fast-axis pitch. Another group of fast-scanning methods creates multiple points simultaneously (i.e., at one time) in either a one-dimensional or two-dimensional array, as opposed to using only one point at a time as detailed in the methods above and further described below. The simultaneous points must be spaced apart at a minimum in the fast axis due to signal contamination and sample damage thresholds, as described above.

[0098] Additionally, a diffraction grating can be used to generate a point array with a pitch that meets the high-speed pixel requirements. A one-dimensional or two-dimensional diffractive optical element (DOE) can be used to split a first-order collimated beam into a multi-point array. The rotary motor can include an assortment of diffraction gratings, and the diffraction element and corresponding high-speed pixel pitch can be selected based on the current sample being investigated. Alternatively, a focusing lens on a linear stage can be actuated and varied to achieve the ideal high-speed pitch on the sample.

[0099] Generally, skipping pixels in the x and y axes also allows the complete image (at a lower resolution) to be scanned, processed, and displayed on the screen as the PARS head makes multiple returns passes. The image on the screen is allowed to update its resolution as more data and pixels are fully filled. Different architectures can be used to allow pixel skipping and / or fast scanning.

[0100] One such architecture uses single-point optical scanning. An exemplary front view (400), top view (402), and isometric detail view (404) of this scanning architecture are shown in FIG. 1F. A resonant microelectromechanical systems (“MEMS”) mirror oscillation frequency (e.g., its scanning frequency) is timed together (e.g., synchronously) with an excitation pulse repetition rate or frequency (respectively, “PRR” or “PRF”) to rapidly optically scan the beam across the sample tile 174 in either one or two dimensions in the x-y plane of the sample tile 174. As shown in FIG. 1F, the entire width of the sample tile 174 in the y-axis may be reached by the scanning beam with the MEMS mirror, and the x-axis is reached by mechanically moving a slow stage in the x-axis. The beam path may be divided into a linear region extending the width of the sample tile 174 that is substantially parallel to the y-axis, and a turning region at the side of the sample tile 174, which indicates a change in beam path direction when the sample tile 174 is mechanically moved in the x-axis. The high-speed pitch may be determined by the scanning frequency of the pump laser PRR and the MEMS mirror.

[0101] Scanning architecture 172 includes a sample tile 174, an objective lens 176, a telescope 178, a single incident collimated detection and excitation beam 180, and a resonant MEMS mirror 182. In scanning architecture 172, the collimated and collinear detection and excitation beams 180 are optically scanned across the field of view (“FOV”) of objective lens 176 by using the resonant MEMS mirror 182. The resonant MEMS mirror 182 may optionally be replaced with a resonant galvo or a rotating polygon, while the downstream architecture (e.g., telescope 178 and objective lens 176) remains the same. In slower systems, the MEMS mirror or galvo may operate in a linear mode (e.g., not resonant), with other design aspects remaining the same. To map the optical scan angle of the resonant MEMS mirror 182 to the entrance pupil of the objective lens 176, a Keplerian telescope is used to expand the beam diameter on the resonant MEMS mirror 182 so that it fills the entrance pupil diameter of the objective lens 176. Simultaneously, the telescope 178 changes the pivot point of the optical angle range swiveling at the entrance pupil, linearly reducing the optical scan angle with the magnification of the telescope 178. For example, in a typical system, the beam diameter at the MEMS mirror is 1 mm, and the MEMS mirror is mechanically moved + / - 2.5 degrees to create an optical scan angle of + / - 5 degrees. A typical magnification of the telescope may be 3, which results in a 3 mm beam diameter at the entrance pupil of the objective lens, together with an optical scan angle of + / - 1.67 degrees swiveling about the entrance pupil diameter (EPD).

[0102] Another architecture uses multi-point optical scanning. An exemplary front view (406), top view (408), and isometric detail view (410) of this scanning architecture are shown in FIG. 1G. Scanning architecture 184 is similar in design to scanning architecture 172, but instead of sending a single collimated detection and excitation beam 180 to the scanning mirror (e.g., a resonant MEMS mirror 182, a resonant galvo, a linear drive mirror, a polygon, etc.), scanning architecture 184 sends a multi-point spot array to the scanning mirror. As shown in FIG. 1G, multiple incident collimated detection and excitation beams 186 may be scanned using a resonant MEMS mirror 182. Scanning architecture 184 reduces overall scan time by a large number of points in the non-scanning direction. Multi-point optical scanning architectures can be useful for systems requiring a high pulse repetition frequency (PRF). That is, because multiple beams can be pulsed alternately, the pulse repetition frequency (PRF) can be much faster and, in some cases, higher than the maximum oscillation frequency of the MEMS mirror. For example, if two beams each have a pulse repetition frequency (PRF) of 5 MHz, the beams can be alternately pulsed at an actual or equivalent pulse repetition frequency (PRF) of 10 MHz to achieve faster imaging times. Because the oscillation frequency of the MEMS mirror is often the upper limit for the pulse repetition frequency (PRF) in an imaging system, alternating beams allows for larger pulse repetition frequency (PRF) values ​​without requiring the MEMS mirror to have an increase in oscillation frequency. In other words, the MEMS mirrors do not need to be any faster to maintain the minimum ideal (fast pitch) spacing.

[0103] Instead of one confocal beam spot at a given time, multiple confocal beams exist simultaneously, and the multiple confocal beams can be scanned across the sample tile 174 in the y-axis using a resonant scanning mirror. As shown in FIG. 1G, each beam can scan a separate subgroup of pixels, thus providing faster scanning than single-point scanning architectures. One light source can provide each beam, each beam can be provided by a different light source, or any other quantity of light sources for providing multiple beams is contemplated in this disclosure. As shown in FIG. 1G, the entire width of the sample tile 174 in the y-axis can be reached with MEMS mirrors by multiple scanning beams, and the x-axis is reached by mechanically moving a slow stage in the x-axis. The beam path can be divided into a linear region extending the width of the sample tile 174 substantially parallel to the y-axis and a turning region at the side of the sample tile 174, where the turning region indicates a change in beam path direction when the sample tile 174 is mechanically moved in the x-axis. The scanning beams may be configured so that the individual beams are at the same y-value relative to each other as they move in the y-axis from one side of the sample tile 174 to the other (e.g., the beams may be offset only in the x-axis, whereby each beam scans the same y-value at different x-values). The most time-efficient configuration of the scanning beams may be when the pitch between simultaneous scanning spots (e.g., the distance between scanning beams in the x-axis at an instant) is the height (i.e., x-dimension) of the sample tile 174 divided by the number of scanning beams. The fast pitch may be determined by the scanning frequency of the pump laser PRR and the MEMS mirror. In the rate-limited case, the number of scanning beams divided by the height of the sample tile 174 equals the ideal fast pitch.

[0104] The embodiment of the present disclosure shown in Figure 1P relates to the delivery of an excitation (and detection) spot 640 to a sample. The excitation and detection beams can be generated from a single detection (and excitation) source 634, in which case the beams are delivered simultaneously and detected on an array detector, e.g., a line array or a camera. The beams are combined using a dichroic mirror (DM) 636 and then split into several different paths using a cascade of beam splitters (BS) 638. The independent beams can be aligned to the appropriate angles using standard turning mirrors in the MEMS mirrors.

[0105] 1Q shows another embodiment of the present disclosure where the detection beams are delivered simultaneously and spectrally separated. A single excitation source 642 may be used to provide several detection sources (DS) 650, 652, 654, 656, and 658 to an excitation spot 644. The excitation source (642) may be split using a series of cascaded beam splitters (BS) 646 and then combined with any number of different detection sources (DS) 650, 652, 654, 656, and 658 having different wavelengths along independent beam paths using dichroic mirrors (DM) 648. The independent beams may be aligned to the appropriate angles using standard turning mirrors in the MEMS mirrors.

[0106] Another variation of the multi-point optical scanning architecture allows for the effective construction of "ultrafast" single-point scans using very high pulse repetition rate (PRR) lasers. Even in current hybrid scanning embodiments, finite limits on the maximum usable pulse repetition rate (PRR) exist due to physical limitations associated with the MEMS scanning mirror. To circumvent these challenges, a multi-point optical scanning design can be used to provide a virtual scan axis.

[0107] An example of this is shown in FIG. 1M. An exemplary front view (406), top view (408), and isometric detail view (410) of this scanning architecture are shown. The scanning architecture is similar in design to scanning architecture 172, but instead of sending a single collimated detection and excitation beam 180 to the scanning mirror (e.g., a resonant MEMS mirror 182, a resonant galvo, a linear drive mirror, a polygon, etc.), scanning architecture 184 sends a multi-point spot array to the scanning mirror. As shown in FIG. 1M, multiple incident collimated detection and excitation beams 186 may be scanned using the resonant MEMS mirror 182. However, unlike scanning architecture 184, the excitation spots are not delivered simultaneously; instead, the events are separated in time.

[0108] Only one excitation and detection beam spot is active at a given time. However, because these spots are essentially separated by minimum point (i.e., fast pitch) spacing due to their alignment, there is no requirement to wait until the system has achieved safe spacing before introducing another excitation event. Instead, alternating points (alternate spots) of the multipoint array are activated, which provides the required spatial separation for the excitation and detection spots. This provides a "virtual" scan axis that coincides with the mechanical scan axis (i.e., the x-axis) due to the multipoint excitation. The virtual scan axis, consisting of multiple points, can be scanned in the y-axis across the sample tile 174 using a resonant scanning mirror.

[0109] As shown in FIG. 1G, the entire width of the sample tile 174 in the y-axis may be reached by multiple scanning beams using MEMS mirrors, and the x-axis may be reached by mechanically moving a slow stage in the x-axis. The beam path may be divided into a linear region extending the width of the sample tile 174 that is substantially parallel to the y-axis and a turning region at the side of the sample tile 174, where the turning region indicates a change in beam path direction when the sample tile 174 is mechanically moved in the x-axis. The scanning beams may be configured so that the individual beams are at the same y-value relative to each other as they move in the y-axis from one side of the sample tile 174 to the other (e.g., the beams may be offset only in the x-axis, such that each beam scans the same y-value at different x-values). The most time-efficient configuration of scanning beams may be when the pitch between simultaneous scanning spots (e.g., the distance between scanning beams in the x-axis at an instant) is the height of the sample tile 174 (i.e., the x-dimension) divided by the number of scanning beams. The fast pitch may be determined by the scanning frequency of the pump laser PRR and the MEMS mirror. In the rate-limited case, the number of scanning beams divided by the sample tile 174 height equals the ideal fast pitch.

[0110] As shown in FIG. 1N, an embodiment of the present invention relates to the delivery of excitation and detection spots 610 to a sample. A different approach is required when multiple excitation and detection beams are not delivered simultaneously but instead introduced sequentially to provide "ultrafast" single-point scanning. In this embodiment, multiple beams can be provided from a single excitation source 608 and detection source 600, but this requires a different design, as reflected in FIG. 1N. In this design, similar to FIG. 1P, the beams are combined using a dichroic mirror (DM) 602 and then split into several different paths using a cascade of beam splitters (BS) 604. However, in this case, each path features a pulse picker (PP) 606, which is used to modulate the combined excitation and detection beams on and off, resulting in a series of sequential excitation and detection "packages" traveling along independent beam paths. The independent beams can be aligned to the appropriate angles using standard rotating mirrors in the MEMS mirrors. This design allows a single high repetition rate laser to operate as if it were an array of lower repetition rate lasers.

[0111] Alternatively, as shown in FIG. 10, multiple beams can be provided by multiple independent excitation sources 616, 620, 624, 628, and 632, and multiple time-synchronized detection sources 612, 618, 622, 626, and 630. In this design, several lower repetition rate excitation lasers are used along with several detection lasers. In each path, the excitation and detection beams are combined using a dichroic mirror (DM) 614. The sources are then electrically modulated, resulting in a series of sequential excitation and detection "packages" traveling along independent beam paths. The independent beams can be aligned to the appropriate angles using standard rotating mirrors in the MEMS mirrors.

[0112] Aspects of the present invention relate to the ability to reconstruct high-resolution images, where the high-resolution images are free of mechanical or optical jitter that results in a loss of resolution in the reconstructed image. Each of the high-speed scanning architectures listed above has different sources of image artifacts that make image reconstruction difficult. The following design solutions (both software and hardware) to aid in reconstruction relate specifically to one-dimensional hybrid MEMS scanning, although aspects may also be applied to the other architectures listed above.

[0113] <1. Image reconstruction assistance method 1> In current embodiments of the hybrid scanning architecture, the PARS system lacks position feedback from the MEMS positioning mirrors used for optical scanning. A series of processing steps is then applied to reconstruct the image without the need for position feedback. These methods rely heavily on structural correlation and data redundancy to iterate through the scattered data until a solution is produced that optimizes image quality.

[0114] One step in this process is to correlate each mirror sweep. Each sweep of the mirror is assumed to follow a sinusoidal path across the sample. The scattering data is then fit to the assumed sinusoidal position. This is observed in FIG. 1H. In many cases, a direct fit to the sinusoidal position will not perform well because some jitter is present in the trigger. To correct for this jitter, the phase of the sinusoidal position is adjusted until the correlation between the scattering signals observed in the forward and backward sweeps of the mirror is maximized, resulting in an optimized mirror phase offset. This is observed in FIG. 1H. Graph 188 shows the raw mirror sweep scattering data fitted to the assumed sinusoidal position, graph 190 shows the sinusoidal phase offset selected to maximize correlation, and graph 192 shows an example of the calculated phase error in an actual image.

[0115] Another step in this process is the iterative process shown in FIG. 1I. First, all stage paths are aligned by structural correlation (e.g., step 1 in "Iteration 1" 412). Second, the aligned images are averaged together to form a merged image (e.g., step 2 in "Iteration 1" 412). Third, each line (i.e., mirror sweep) from each stage path is independently repositioned to maximize correlation with the merged image (e.g., step 3 in "Iteration 1" 412). Fourth, the process is repeated until no further positional shifts remain to be applied (e.g., generally, "Iteration 2" 414). A diagram showing a pre-correction image 416 and a post-correction image 418 using the image reconstruction process is presented in FIG. 1J.

[0116] <2. Image reconstruction assistance method 2> When driving a MEMS mirror at resonance, the voltage waveform is typically sinusoidal. Due to electrical instabilities in the waveform and mechanical instabilities in the mirror itself, pixels do not position exactly as expected, resulting in edge blurring in the resulting reconstructed image. One solution is to add a position-sensitive detector ("PSD") to the system, which captures the entire optical stroke on the sensor's surface. The PSD then outputs an analog voltage waveform that is linearly scaled to position as the signal is amplified and conditioned. Because the output of the PSD module is analog, the signal can be converted to digital via an A / D sampling card, allowing it to be synchronized with the detector channel using the same system clock or trigger. The detection beam itself can be taken between the MEMS mirror and the telescope (e.g., a portion of the beam's energy can be sampled using a beam sampler to capture 1% or about 1% of the beam intensity without significantly modifying or affecting the initial beam) and collected on a position-sensitive detector PSD or a secondary (non-PARS-related) collimated laser. Alternatively, the LED can be reflected off the MEMS and collected on a position-sensitive detector PSD that is independent of the PARS excitation and detection optical paths.

[0117] <3. Image reconstruction assistance method 3> In 1D hybrid scanning (and 2D stage scanning), stage position feedback via optical encoders, Hall sensors, or laser proximity sensors can be used, synchronized with system triggers and clocks to help aid in reconstruction. The stage position is recorded directly as a number.

[0118] <4. Image reconstruction assistance method 4> Another method to aid image reconstruction is to image a known spatial calibration target using the scanning element in the instrument. In this manner, the known target is enabled to act as a master reference image for reconstructing unknown geometry, similar to the case of a histological sample. An example of a known spatial calibration target is the etched silicon slide shown in FIG. 1K. The etch depth (into the page) of the silicon feature is designed to be the outer focal depth of the primary objective lens, thereby measuring contrast from the unetched surface where the focus is located. In the exemplary embodiment shown in FIG. 1K, the etch depth is 5 μm. The calibration target can be mounted below a histological sample carrier fixed to a mechanical stage. By imaging a calibration target mounted on the same mechanical stage stack as the sample itself, instabilities from both the MEMS mirror and the mechanical translation stage can be corrected. This method results in a design that is enabled to assist image reconstruction without the need for other sensor-based modalities (e.g., position-sensitive detectors (PSDs)). In a typical design, the detection beam can be picked up between the last element of the telescope and before the primary objective. This picked up beam can be sent to its own narrow bandwidth non-glass corrected objective lens and focused onto a calibration target mounted on the underside of the specimen carrier plate. In this manner, both mirror and stage instabilities can be calibrated out.

[0119] Additionally, resolution targets in silicon have been enabled to act as calibration targets for normalizing power levels and as health checks for PARS systems. A calibration target, for example, etched silicon, is ideal because it can provide PARS contrast that is repeatable and uniform across the surface of the reference target. This is because the sample is composed of pure crystalline silicon.

[0120] <5. Image reconstruction assistance method 5> In one-dimensional hybrid scanning, emissive and non-emissive signals are acquired at a fast pixel pitch at a selected excitation laser repetition rate. The detection wavelength is continuous wave (CW) and can be collected continuously between emissive and non-emissive excitation events at a much higher sampling rate. This higher resolution data from the detection channels aids in image reconstruction and can be used to complement the reconstruction aids described above.

[0121] A detailed discussion of multi-path detection architectures is provided here. In PARS, non-radiative absorption-induced perturbations in optical properties are visualized using a secondary confocal detection laser. The detection laser is confocal with the excitation spot, whereby absorption-induced modulations may be captured as changes in the backscattered intensity of the detection laser. For a given detection intensity I det Before the excitation pulse interacts with the sample, the signal is given by the following relation, PARS pre-ext ∝I det (R), where R is the unperturbed reflectance of the sample.

[0122] When the excitation pulse interacts with the sample, the signal is post-ext ∝I det The total PARS absorption contrast can then be approximated as (R + ΔR), where the pressure and temperature induced changes in reflectance are denoted by ΔR. sig ∝PARS post-ext -PARS pre-ext The above relationship can be approximated as pre-ext and PARS post-ext Regarding substitution, PARS sig ∝I det (R+ΔR)-I det (R) results.

[0123] Before the excitation pulse, the backscatter of multipath PARS (i.e., MPPARS) is given by the following relationship: pre-ext ∝(I det (R)) nwhere R is the unperturbed reflectivity of the sample and n is the number of times the excitation interacts with the sample. When the excitation pulse interacts with the sample, the signal is expressed as MPPARS post-ext ∝(I det (R+ΔR) n where the pressure and temperature induced changes in reflectivity are denoted by ΔR.

[0124] The total multipath PARS absorption contrast is then calculated as MPPARS sig ∝MPPARS post-ext -MPPARS pre-ext The above relationship is approximated as MPPARS pre-ext and MPPARS post-ext The following substitutions are made for MPPARS sig ∝(I det (R+ΔR) n -(I det (R)) n where n is the number of times the detector interacts with the sample. The PARS signal may be nonlinearly broadened by these repeated interactions of the backscatter detector with the sample. The detector may then be redirected to interact with the sample any number of times, resulting in a corresponding degree of nonlinear broadening in the nonradiative absorption contrast.

[0125] As noted above, a multi-path detection architecture may be oriented such that a path consists of a reflection or transmission event, which may occur at normal incidence to the sample, or at some related transmission or reflection angle. For example, if a target is characterized by a particularly strong Mie scattering angle, it may be advantageous for multiple paths to be oriented along this direction. The multiple paths may occur along a single (only one or exactly one) path (e.g., normal incidence reflection), or along multiple paths, e.g., a normal incidence transmission architecture, or even an architecture having additional (three or more) paths to exploit additional spatial nonlinearities.

[0126] 1L, in some cases, a multi-path detection architecture 100 may include an excitation source 102 (e.g., a 266 nm excitation source or laser), one or more detection sources 104 (e.g., a 405 nm detection source or laser), one or more photodiodes or photodetectors 106, a circulator 108, a collimator 110, one or more mirrors 112 for guiding the excitation and / or detection light, a prism 116, and a variable beam expander 118. Additionally, the multi-path detection architecture 100 may include a pair of alignment mirrors 120 for aligning the excitation and / or detection light, and one or more scanners or scan heads (122, 124) positioned on different sides of the sample. The one or more scanners may include a first scanner 122 for transmitting the excitation and detection light through the sample and a second scanner 124 positioned with a mirror 126 to enable multiple passes. The computer 128 analyzes the received signals and / or may be used to control the excitation source 102 and the detection source 104 .

[0127] Referring to FIG. 2A , in some cases, the multi-path detection architecture 200 may feature multi-path non-radiative detection, which may simultaneously capture transmitted and reflected signals (e.g., a transmission multi-path detection architecture). The multi-path detection architecture 200 may also feature a condenser lens used to focus divergent light. At the same time, the multi-path detection architecture 200 features two independent radiation collection paths, one on each side of the sample, as shown in FIG. 2A . That is, a first beam is focused onto the sample from the first side of the sample, and a second beam is focused onto the sample from the second side of the sample, the first side facing away from the second side. Due to this design, it will be appreciated that the multi-path detection architecture may facilitate imaging of certain samples, such as thin samples required for histological imaging, because the beam is allowed to pass through the specimen to enable imaging. For example, the sample thickness may range from less than 1 μm to more than 5 mm. More generally, the specimen can be of any thickness that allows light to pass through.

[0128] In these examples, the multi-path detection architecture 200 may include two excitation sources 202, 204 having different excitation wavelengths (e.g., a 266 nm excitation source or laser and a 532 nm excitation source or laser), one or more detection sources 206 (e.g., a 405 nm detection source or laser), two objective lenses 208, 210, one or more photodiodes or photodetectors 212, 214, 216, 218, a collimator 220, one or more mirrors 222 for guiding the excitation and / or detection light, one or more dichroic mirrors 224 for guiding the excitation light, and one or more spectral filters 226.

[0129] In the transmissive section, the multi-path detection architecture 200 may use two objective lenses 208, 210, each mounted on opposite sides of the sample as shown. The excitation and detection beams from the excitation and detection sources (202-206) may be confocal on the specimen 228 using the lower objective lens (210). A portion of the modulated detection beam and / or a portion of the radiation absorption and inelastic scattering signals may be captured by the upper objective lens (208). In some cases, a portion of the modulated detection beam may be returned to the specimen 228, and the returned detection beam may be captured by the lower objective lens (210). The returned portion of the modulated detection beam may range from less than 1% to 100% of the light collected by the upper objective lens. In some cases, a portion of the radiation absorption signal may also be captured using the lower objective lens (210). The detection beam may be separated from the radiation relaxation signal via chromatic isolation. The isolated detection and emission signals on the top and bottom of the sample are then directed to separate photodiodes, for example, photodiodes (212, 214, 216, 218).

[0130] As discussed above, with this architecture, both emissive and non-emissive signals may be collected on both sides of the specimen. By collecting both linear and non-linear representations of the non-emissive signal, there is an enhanced opportunity to capture additional intrinsic contrast within the material. As discussed below, this may facilitate an enhanced signal-to-noise ratio or may be used for techniques such as super-resolution or ultra-localization.

[0131] Although a multi-pass detection architecture may in some cases be described as a transmission-mode device (e.g., when the beam must pass through the specimen to generate and capture the signal), it is understood that the multi-pass detection architecture may also feature a reflection-mode design. In such a design, all signals may be collected from one side of the specimen (e.g., a single-pass, multi-pass detection architecture). This design may facilitate imaging of certain specimens, such as thick specimens, when the beam is not required to pass through the specimen to enable imaging.

[0132] It is also understood that a multi-pass detection architecture, as described above, is not a requirement for PARS imaging. In some cases, a multi-pass architecture may not be required or may not be used at all; in these cases, PARS may instead be described as a transmission-mode only or reflection-mode device. In such designs, the detection beam is required to interact with the sample only once.

[0133] Figure 2H may be substantially similar to the architecture of Figure 2A, but with the elimination of certain components. In particular, Figure 2H may eliminate the 50:50 splitter located between the spectral filter 226 and the objective lens 210. This elimination removes multipath effects.

[0134] Figure 2B may be substantially similar to the architecture of Figure 2A, but includes additional components. In particular, Figure 2B may include a spectral filter 226 (with a condenser lens) and a photodetector 242 between the excitation source 204 and one or more dichroic mirrors 224, and a spectral filter 226 (with a condenser lens) and a photodetector 244 between the excitation source 202 and one or more dichroic mirrors 224. These additions to Figure 2B may enable excitation scattering. This is achieved due to a beam sampler, which redirects a portion of the light returning from the sample to the spectral filter 226, the condenser lens, and the photodetector. The spectral filter isolates the excitation wavelength, and the photodetector measures the intensity, thereby measuring the scattered light intensity.

[0135] Figure 2I may be substantially similar to the architecture of Figure 2B, but with the elimination of certain components. In particular, Figure 2I may eliminate the 50:50 splitter located between the spectral filter 226 and the objective lens 210. This elimination removes multipath effects.

[0136] Figure 2C may be substantially similar to the architecture of Figure 2B, but includes additional components. In particular, Figure 2C may include a beam sampler and photodetector 248 with a spectral filter 226 (with a condenser lens), and a beam sampler and photodetector 246 with a spectral filter 226 (with a condenser lens), with the beam sampler and photodetector 248 and the beam sampler and photodetector 246 positioned between mirror 222 and photodetector 212. These additions shown in Figure 2C may enable transmission measurements.

[0137] Figure 2J may be substantially similar to the architecture of Figure 2C, but with the elimination of certain components. In particular, Figure 2J may eliminate the 50:50 splitter located between the spectral filter 226 and the objective lens 210. This elimination removes multipath effects.

[0138] Figure 2D may be substantially similar to the architecture of Figure 2B, but with additional components included and some components removed. In particular, Figure 2D may replace photodetectors 216, 218, 242, and 244 with cameras 250, 256, 258, and 260, respectively. Thus, with these changes, Figure 2D may be a camera-based system.

[0139] Figure 2K may be substantially similar to the architecture of Figure 2D, but with the elimination of certain components. In particular, Figure 2K may eliminate the 50:50 splitter located between the spectral filter 226 and the objective lens 210. This elimination removes multipath effects.

[0140] Figure 2L may be substantially similar to the architecture of Figure 2C, but with additional components included and some components removed. In particular, Figure 2L may replace photodetectors 212, 216, 218, 246, and 248 with cameras 254, 250, 256, 258, and 260, respectively. Thus, with these changes, Figure 2L may be a camera-based system.

[0141] Figure 2M may be substantially similar to the architecture of Figure 2L, but with the elimination of certain components. In particular, Figure 2M may eliminate the 50:50 splitter located between the spectral filter 226 and the objective lens 210. This elimination removes multipath functionality.

[0142] FIG. 2E may be substantially similar to the architecture of FIG. 2B, but with some components removed. In particular, all of the transmission components of FIG. 2B are removed in FIG. 2E. Thus, FIG. 2E may be FIG. 2B when in reflection mode. Reflection mode may be considered a mode for an architecture when all light is input and collected from the same side of the sample. Thus, light is not required to transmit through the specimen for measurement or imaging purposes.

[0143] FIG. 2F may be substantially the same as the architecture of FIG. 2C, but some components are removed. In particular, all of the transmissive components of FIG. 2C are removed in FIG. 2F. Thus, FIG. 2F may be FIG. 2C when in the reflection mode.

[0144] FIG. 2G may be substantially the same as the architecture of FIG. 2D, but some components are removed. In particular, all of the transmissive components of FIG. 2D are removed in FIG. 2G. Thus, FIG. 2G may be FIG. 2D when in the reflection mode.

[0145] FIG. 3 shows an excitation source architecture 300, which may be used to generate the excitation sources 202, 204 in any of the excitation sources in FIGS. 2A - 2M. The excitation source architecture 300 may include a pump laser 302 that provides light at a specific wavelength (e.g., 1030 nm), one or more half-wave plates 304, 306, one or more optical crystals 308, 310 (e.g., lithium triborate (LBO) or barium beta borate (BBO)), one or more mirrors 312 for guiding the light from the pump laser 302, one or more dichroic mirrors 314 for guiding the light from the pump laser 302, a prism 316, and two variable beam expanders 318, 320.

[0146] <PARS extraction method> Upon an excitation event at an excitation location (e.g., a location where the excitation beam is focused), the PARS system of the present disclosure may collect all, substantially all, or a portion of the generated signal to extract information about the sample at the excitation location. For example, using one or more PARS signal extraction methods, the PARS system may extract information, e.g., one or more unique feature vector characteristics, at the excitation location of the sample. The extracted information (e.g., feature vector characteristics) may contain absorption and / or scattering characteristics at the excitation location, and more specifically, at one or more pixel locations within the excitation location. As a result, the extracted information (e.g., feature vector characteristic(s)) may contain details about the sample composition, constituent biomolecules, etc. It is understood that the PARS system of the present disclosure may collect and / or extract multiple feature vector characteristics at multiple excitation locations throughout a tissue sample / specimen.

[0147] Each PARS data vector (data vector) contains "n" PARS features, as determined by the user or a control algorithm. The feature vector can contain any number of extracted signals, such as signal energy, total non-radiative or radiative energy, total absorption, quantum efficiency ratio, absorption derivative, relative relaxation fraction, time-domain signal features, blind clustering / dimensionality reduction, isolated non-radiative initial temperature and pressure signals, and filter bank / frequency-based extraction. These measurements may be extracted at any wavelength or combination of wavelengths. The feature vector can also contain any number of secondary measurements extracted as different combinations, calculations, or ratios of primary features. An example of a secondary measurement could be the difference between quantum efficiency ratios (QER) at two different excitation wavelengths. Overall, the PARS feature vector can contain any information collected and extracted from each PARS event.

[0148] The feature vectors may then be used at the pixel level or in aggregate to analyze aspects of the sample. In one example, PARS feature vectors can be correlated at the pixel level to ground truth (e.g., histochemical or immunohistochemical stains). This can provide a one-to-one mapping between PARS data vectors and different histochemical stains or their underlying biomolecular targets. This process allows a PARS "signature / fingerprint" or ground truth PARS data vector to be calculated for a given biomolecule or mixture of biomolecules. Alternatively, PARS vectors may be analyzed in aggregate, and the distribution of vectors within a sample can indicate the specimen's underlying diagnostic characteristics, such as malignancy, tumor grade, molecular expression, etc.

[0149] The PARS system of the present disclosure may further process the extracted information to form a visualization, e.g., an image. In examples where the visualization is an image, the image may be a digital image, e.g., a raster image (e.g., JPEG, PNG, BMP, GIF, etc.), which comprises a plurality of pixels or one or more sets of pixels. As used herein, the term "pixel" refers to the smallest addressable element in an image. In these cases, the PARS system may assign to the pixel a portion of the information, e.g., a feature vector characteristic, extracted from the signal generated at a single excitation position. In some cases, the location of the pixel (i.e., pixel location) corresponds to a unique portion of the excitation position. In cases where the visualization comprises one or more sets of pixels, the single set of pixels may correspond to all or substantially all of the information extracted from the signal generated at a single excitation position of the sample. That is, the visualization may contain all or substantially all of the information extracted from the signal generated at a single excitation position or multiple excitation positions. In some cases, the visualization may include all or substantially all of the information extracted from signals generated at multiple excitation positions, which may range across all or substantially all of the sample.

[0150] In some examples, the visualization may then be used directly for histological diagnosis or in conjunction with other tools, e.g., AI, to perform a diagnosis or generate more advanced colorization. For example, in cases where the extracted information includes absorption and / or scattering properties, the absorption and / or scattering properties at pixel locations may be further assigned one or more values, e.g., color and / or intensity values, corresponding to a color space. The color space may be RGB, YCbCr, CIELAB, sRGB, YPbPr, scRGB, HSV, CMYK, or any other known color space. These color and / or intensity values ​​may be calculated using one or more PARS signal extraction methods. In some examples, the PARS system may include one or more processors, which are configured to extract information from the generated signals, process the extracted information, and / or generate the visualization, as discussed above.

[0151] The PARS signal extraction method of the present disclosure may be any of the signal extraction methods, signal processing methods, and / or signal acquisition methods described in U.S. Patent Application No. 17 / 010,500, filed August 5, 2021 (entitled Pars_imaging_methods), and Patent Cooperation Treaty Application No. PCT / IB2021 / 055380, filed May 12, 2022 (entitled Photoabsorption_remote_sensing(PARS)imaging_methods), each of which is incorporated herein by reference. PARS signal extraction methods may be used for both radiative and non-radiative channels. It is understood that each PARS signal may be extracted in several different ways to capture different salient details of the signal, or the time-evolving signal may be used in its entirety. The following PARS signal extraction methods may be used in a PARS system.

[0152] <1. Signal Energy> The PARS signal extraction method can be a signal energy measurement process. In this process, to determine the total level of absorbed energy at a given pixel location, the process determines the integral of the modulation in the time-domain signal (FIG. 4). The signal's direct current DC value (intensity before excitation) is calculated and subtracted from the signal. The remaining modulation is then integrated to calculate the highlighted area shown in FIG. 4. That is, the integral of the modulation in the signal is extracted independently of the signal's direct current DC offset. Higher absorption results in larger and longer modulations corresponding to larger integrals. This is a simple, fast method that is robust to additive noise. This method may be applied directly to either the generated emissive or non-emissive signal to obtain an estimate of the total absorption level.

[0153] This method may be further enhanced by applying noise reduction / filtering before extracting the integral. For example, matched filtering (e.g., based on k-means extraction) may be used to extract the total signal energy. This technique is designed to optimally filter the signal based on its expected time-domain shape. This provides a robust, noise-resistant method for determining absorption amplitude or pixel "brightness."

[0154] 2. Total non-radiative or radiative energy In PARS architectures and systems that use multiple excitation sources at different wavelengths, there are radiative and non-radiative absorption measurements (e.g., extractable information) for each wavelength. That is, by isolating the generated signal by wavelength, the PARS system can extract and / or collect radiative and non-radiative absorption measurements for each wavelength at the excitation location. Therefore, it can be advantageous to display the combined radiative or non-radiative relaxation energy across all wavelengths. For example, exemplary images showing the total radiative and non-radiative energy are presented in FIGS. 5 and 6, respectively.

[0155] <3. Total absorption> By capturing both radiative and non-radiative absorption fractions (e.g., extractable information) at the excitation location, PARS architectures and systems can facilitate enhanced visualization. Unlike conventional modalities that capture some of the radiative or non-radiative absorption independently, in PARS, contrast can be unconstrained by efficiency factors, such as photothermal conversion efficiency or fluorescence quantum yield. Thus, PARS can provide enhanced sensitivity to any range of chromophores. The total absorption TA(λ) for any excitation wavelength is calculated as the sum of all radiative absorption signals (P r (λ)) and non-radiative absorption signal (P nr (λ) can be calculated as the sum of the absorption magnitudes of

[0156] TA(λ)=P r (λ)+P nr (λ). An exemplary total absorption image captured using a 266 nm excitation source is shown in FIG. 7, and another exemplary total absorption image captured using a 532 nm excitation source is shown in FIG.

[0157] Total absorption may also be calculated across several wavelengths. For example, a similar total absorption may be calculated for both 266 nm and 532 nm excitation. This is equivalent to the sum of the total absorption features for each wavelength independently. An example image including both radiative and non-radiative relaxation effects at 266 nm and 532 nm is shown in Figure 9. As shown, Figures 7-9 capture the full interaction. In other words, this modality is enabled to capture the effects simultaneously.

[0158] 4. Quantum efficiency ratio Once information has been extracted from the signal generated at the excitation location, the PARS system may extract / gather additional biomolecule-specific details from the extracted information. For example, additional biomolecule-specific details may be extracted based on proportional radiative and non-radiative relaxation characteristics. Different biomolecules may exhibit stronger radiative or non-radiative relaxation tendencies. This is specified by known intrinsic material properties, such as fluorescence quantum efficiency. Therefore, additional biomolecule-specific details may be further extracted from the relative proportions of radiative and non-radiative absorption fractions. This is presented as the quantum efficiency ratio, or QER, which is calculated as follows:

[0159]

number

[0160] In a PARS system, this may be done at any given excitation wavelength. An exemplary image is shown in Figure 10 for a PARS system using 266 nm excitation, and an exemplary image is shown in Figure 11 for a PARS system using 532 nm excitation.

[0161] The quantum efficiency ratio QER can also be calculated across several wavelengths. For example, a similar quantum efficiency ratio QER can be calculated for both 266 nm and 532 nm excitation. An example image produced by a PARS system using both 266 nm and 532 nm excitation is shown in FIG. 12.

[0162] The quantum efficiency ratio (QER) can then be used in combination with other aspects to provide useful visualization, such as colorization. For example, in a PARS system using both 266 nm and 532 nm excitation, the quantum efficiency ratio (QER) may be combined with total absorption, as shown in FIG. 13. In this example, the quantum efficiency ratio (QER) is used to define the color of the image, and the total absorption is used to define the intensity. In this case, the quantum efficiency ratio (QER) is scaled between [0, 1] and used to define the hue value in the HSV color space. At the same time, the total absorption is scaled between [0, 1] and used to define the saturation and value. Thus, in this example, the color of each pixel (from the quantum efficiency ratio (QER)) can provide details about the type of biomolecule at a given pixel, and the brightness (from the total absorption) can provide some information about the concentration of the biomolecule. A similar approach can be envisioned using any other color space, e.g., YCbCr, Lab, RGB, etc., where intensity is defined by total absorption and color is defined by quantum efficiency ratio (QER). The PARS system may be configured to generate a combined quantum efficiency ratio (QER)-total absorption image.

[0163] <5. Absorption Differential> Once information has been extracted from the signal generated at the excitation location, the PARS system may further process the extracted information, for example, by assigning a color value to the extracted information. For example, when excitation sources having different wavelengths are used in the PARS system, any two (or more) absorption features may be comparatively observed using absorption derivative visualization. This visualization provides a way to assess the relative difference in absorption intensity at each pixel location of the excitation location for a given wavelength. The absorption derivative for two wavelengths is calculated as follows:

[0164]

number

[0165] where S1 is the absorption signal (radiative or non-radiative) at a given wavelength and S2 is the corresponding radiative or non-radiative absorption signal at a second wavelength. Evaluating the difference in absorption between two wavelengths can provide a significant enhancement in the ability to separate different biomolecules when compared to observing their absorption independently or in combination.

[0166] This method may be applied to images as shown in Figures 14-15, where color maps are assigned according to absorption derivative values. Red indicates stronger 266 nm absorption, and blue indicates stronger 532 nm absorption. The non-radiative derivative contrast image is shown in Figure 14, and the radial derivative contrast image is shown in Figure 15.

[0167] <6. Relative Relaxation Fraction> In some examples, the extracted radiative or non-radiative absorption characteristics, or different combinations thereof, may be expressed as a fraction of the total absorption. This allows the proportional absorption at each wavelength to be evaluated independently of the concentration of biomolecules in each area. This may also allow for easier comparison between signals captured at different wavelengths. For example, the non-radiative signal at a single wavelength may be expressed as a fraction or percentage of the total non-radiative relaxation. In another example, the non-radiative relaxation at a single wavelength may be expressed as a fraction or percentage of the total absorption (including radiative and non-radiative relaxation).

[0168] The radiation relaxation signal is an emission spectrum. This emission spectrum can be uniquely associated with the excited biomolecule. In other words, for a given excitation wavelength, two different biomolecules can be expected to exhibit different spectral emissions. If the spectral difference is significant, this can facilitate better separation of the two biomolecules.

[0169] 7. Time domain signal characteristics Additional PARS contrast exists within the time evolution of the PARS signal. Important information about the biomolecular composition of a sample may be encoded in various time-domain signals. Unique features may be accessed in signal frequency, rise time, fall time, amplitude, etc. That is, by measuring these features, a PARS system can then gather and / or extract additional information (e.g., chromophore-specific information) from a single excitation event. For example, this may enable chromophore unmixing (e.g., detecting, separating, or otherwise discretizing constituent species and / or subspecies) from a limited number of excitation events.

[0170] In some instances, the shape of the non-radiative PARS signal depends on the evolution of pressure- and temperature-induced fluctuations within the sample. In these instances, the shape of the time-evolving modulation then captures details about the local material properties. On shorter time scales, the signal decay rate can be correlated to the sound speed of the material. On longer time scales, the signal decay rate can be correlated to the heat propagation speed. Additionally, the signal decay rate can be correlated to different velocities when shear waves are generated. The signal rise time can also provide additional information about mechanical properties (e.g., sound speed). This means that the PARS time-domain signal can capture properties such as thermal diffusivity, conductivity, sound speed, density, heat capacity, and acoustic impedance, which can be extracted and collected by the PARS system for further processing.

[0171] Similar material-specific features may be encoded in the time evolution of PARS emission modulation. As with non-radiative signals, the radiative signal time-domain shape is dictated by material properties. For example, if a fluorescent signal is isolated from radiative relaxation, time-resolved fluorescence lifetimes may be captured. Fluorescence lifetimes are biomolecular-specific properties that can be used to directly identify biomolecular constituents.

[0172] These time-evolving signals may be decomposed into measurable, intrinsic (e.g., intrinsic, measurable in amplitude or magnitude and / or evolution time) characteristic features. This wealth of information may then be used to improve the available contrast, provide additional multiplexing capabilities, and provide characteristic molecular signatures of the constituent chromophores. Several methods may be used to capture features indicative of material properties from the time-domain signals. Methods may include techniques such as principal component analysis, Fourier analysis, frequency decomposition, principal components of Fourier analysis, fitting methods, k-means methods, or wavelet extraction methods. These methods may be applicable to both radiative and non-radiative time-domain signals. Any information or features extracted from the time-domain signals may be included as additional information in the PARS data vector, or the additional information may be used to augment the data for further analysis.

[0173] 8. Blind Clustering / Dimensionality Reduction In some examples, to gather or extract additional information, the PARS system may use blind clustering and / or dimensionality reduction methods to compress the time-domain signal into fewer representative features, e.g., identify time-domain features related to the underlying sample features. Some examples of potential techniques include k-means clustering, principal component analysis, principal linear component decomposition, or other signal decomposition methods. Alternatively, intelligent AI-based clustering methods may be used. An advantage of these blind techniques is that they do not require prior information about the signal. This may facilitate processing when a sample, e.g., an analyte, is complex and may have numerous time-domain features to exploit.

[0174] If implemented correctly, clustering can identify signal features that capture material-specific information of the underlying analyte, such as thermal diffusivity, conductivity, speed of sound, density, heat capacity, and acoustic impedance. These feature intensities may then be extracted and used directly. Alternatively, the extracted feature intensities may be used with further processing to produce unique visualizations.

[0175] For example, a specialized k-means clustering method may be used to extract / collect signal features from the generated time-domain signal at the excitation location, as shown in FIGS. 16A-16D. FIGS. 16A-16C show k-means feature extraction applied to a thin section of preserved human breast tissue. In this example, UV excitation (e.g., 266 nm) may target several biomolecules, such as collagen, elastin, myelin, DNA, and RNA, and FIG. 16D is a graph illustrating the feature extraction of FIGS. 16A-16C. As shown, the feature intensity present in each time domain is shown along with the corresponding feature in the time-domain signal. Clustering may then be used to identify three unique time-domain features. The intensity or prevalence of each feature may be extracted from the time domain and presented in a corresponding feature representation, e.g., Feature 1 (FIG. 16A), Feature 2 (FIG. 16B), and Feature 3 (FIG. 16C).

[0176] The extracted feature-specific images may then be used for further processing or may be displayed directly. In one example, for example, when further processed, each pixel may be assigned a color value based on the extracted feature intensity, as shown in FIGS. 16A-16C. In another example, each of the feature images from FIGS. 16A-16C may be assigned to one of the red, green, or blue (RGB) channels to generate a colorized image, as shown in FIG. 17. The color and intensity of each pixel may then be described by the proportional presence of the three features in the time-domain signal. In this example, the nuclei (which may appear as a first color, in this example, green) are unmixed with the surrounding connective tissue (which may appear as a second color, in this example, blue / purple).

[0177] In another example, the extracted time-domain features may be used together with other PARS features to form a visualization, e.g., a colorization, as shown in FIG. 18. In these examples, two features may be further extracted (e.g., further processed) by the PARS system using a K-means method. Any color space (e.g., RGB, YCbCr, CIELAB, sRGB, YPbPr, scRGB, HSV, CMYK, etc.) may be used to form a colorization from these features. For example, in the YCbCr color space, the color value (i.e., chrominance blue (Cb) and chrominance red (Cr) channels) of each pixel is defined by the presence of the extracted features using the K-means method. At the same time, the intensity value of the color channel (i.e., luminance (Y)) is assigned based on the energy of the PARS non-radiative signal. That is, the color value assigned to a pixel may be determined from the extracted features (e.g., k-means features), and the color intensity may be determined by the amount of absorbed energy. In this example, measurements are scaled in the [0,1] range to utilize the full range of the YCbCr color mapping. As shown in Figure 18, nuclei, for example, may appear in one color (white), while connective tissue appears in different colors (shades of blue and orange) depending on the signal composition. That is, white represents a color value of 235, 128, 128, representing an intensity range of 16 to 235 from black to white depending on the intensity at the specific pixel, while exemplary shades of blue and orange represent color values ​​of X, 16, 240 and X, 220, 36, respectively, representing an intensity range of 16 to 235. In some cases, it may be desirable to convert the color and / or intensity values ​​of each pixel in the YCbCr color space to another color space, such as the RGB color space. It is understood that blind clustering and / or dimensionality reduction methods can help highlight the presence of proportional time-domain features independently of biomolecule concentration.

[0178] 9. Isolation of Non-Radiative Initial Temperature and Pressure Signals As discussed above, in PARS, the initial non-radiative signal results from pressure (photoacoustic signal) and temperature (photothermal signal) induced modulation of the local material properties of the specimen. Heat always accumulates, but photoacoustic pressure can only be generated under specific conditions. In cases where pressure is generated, the initial pressure and temperature signals may be isolated and evaluated independently.

[0179] When non-radiative signals are captured as amplitude modulations of a confocal detection source, pressure-induced modulation, as specified by the speed of sound, is expected to be several orders of magnitude faster and higher than thermal modulation, as specified by thermal conductivity, for most samples / analytes. For example, a photothermal signal used to generate an image of the analyte may be measured within 500 ms of an excitation event, and a photoacoustic signal used to generate an image of the analyte may be measured within 500 ns of an excitation event. More generally, temperature (e.g., photothermal signal) can be measured as soon as pressure (e.g., photoacoustic signal) leaves the area. This can correlate to measuring temperature in the μs-ms range (e.g., 1 μs-500 ms) and pressure in the ps-ns or longer range (e.g., 1 ps-500 ms). While decay can occur over μs-ms, it is contemplated in the present disclosure that faster measurements are possible. Temperature effects can occur faster than the given range in certain cases (e.g., metallic samples). This is discussed further below.

[0180] For example, Figure 19 shows PARS non-radiative pressure (photoacoustic) and temperature (photothermal) induced modulations in the local optical properties of a sample as observed by the detection source of a PARS system. Figures 20A-20C show PARS non-radiative pressure (photoacoustic signal) and temperature (photothermal signal) induced modulations in the local optical properties of a sample as observed by the detection source of a PARS system. In Figures 19 and 20A-20C, the pressure signal is captured by the initial rapid signal, and the temperature is a slower transient signal. This means that the pressure-induced signal can be isolated from the thermal signal, allowing for specific assessment of independent features.

[0181] Specifically, Figure 20B shows an example time domain where the temperature dominates and the pressure signal is not discernible. Figure 20C shows an example time domain where the PARS detection spot position is shifted relative to the excitation spot. This means that the fast-propagating acoustic signal arrives at the detection spot before the slower-propagating thermal signal, resulting in a directly discernible signal region.

[0182] By isolating the pressure signal, pressure-specific properties, such as sound speed, acoustic impedance, and absorber size, may be extracted directly from the PARS pressure modulation. For example, this may allow the sound speed at the detection focal spot to be measured directly. Alternatively, this may facilitate super-resolution imaging when the pressure response is correlated to absorber size.

[0183] Conversely, if the thermal signal is isolated, properties such as heat propagation rate or specific heat capacity may be determined in isolation. Furthermore, by isolating the two signals, the relationship between the two may be evaluated. Properties such as photothermal conversion efficiency, isothermal compressibility, and elasto-optical properties may be measured (and thus extracted and / or collected by the PARS system) based on the relative presence and rate of pressure and temperature modulation.

[0184] In some samples, specific portions of the non-radiative signal may be targeted as a means of achieving an enhanced signal-to-noise ratio (SNR). Initial pressure modulation can provide modulation of intensity several orders of magnitude greater in some samples, allowing for significant enhancement of image fidelity.

[0185] In another example, pressure modulation may be targeted in isolation from thermal modulation to enable faster imaging. Pressure signals propagate through a sample proportional to the speed of sound in the medium, typically on the order of 1000 m / s in biological samples. Thermal signals propagate through a sample proportional to the thermal conductivity, typically on the order of 0.001 m / s. Thus, pressure dissipates several orders of magnitude faster than temperature. Targeting only pressure signals may enable imaging to be several orders of magnitude faster because pressure signals propagate much faster than their thermal counterparts.

[0186] 10. Filter Banks / Frequency-Based Extraction Another method for capturing signal features may rely on the presence of frequency information encoded in the time-domain signal. A series of analog or digital filters may be used to isolate specific frequency bands in the time-domain signal to isolate signal intensities associated with specific frequencies or frequency bands. This may be performed by splitting the original analog signal from the photodetector and recording it in two separate channels, or by digital means. The specific frequency bands may then be processed according to any of the described methods, such as signal integration / energy extraction, blind clustering, or any other processing method.

[0187] 11. High-speed signal acquisition In some PARS architectures, non-radiative absorption signals are detected as modulations in the backscattered detection intensity. Extracting the modulations and their energy from the scattered signal is an essential step in the formation of the image. One such method for extracting the PARS signal is to use optimized analog or digital filtering techniques that specifically isolate non-radiative induced modulations using targeted high- and low-pass filters. This method is clearly beneficial because it ensures that non-radiative modulations occur at higher frequencies than the local scattering contrast.

[0188] Thus, in PARS, the high-pass filter can be selected to completely remove the scattered signal while retaining the PARS non-radiative modulation. To remove the scattered signal, the maximum spatial frequency of the scattering can be calculated as follows.

[0189]

Equation

[0190] where F s_max is the maximum frequency of the scattered signal. In the case of a 1 MHz excitation source hybrid scanning embodiment, this level is approximately 3.5 MHz. Together, the low-pass filter can be used to remove excessive high-frequency noise from the signal while aiming to retain as much information as possible in the initial PARS signal. For example, in this system, since the photodiode bandwidth is 50 MHz, a 50 MHz low-pass is applied to remove the irrelevant electrical noise in the signal. By using a 3.5 MHz high-pass filter and a 50 MHz low-pass filter, the non-radiative signal can be directly isolated from the scattered signal. Then, the modulation energy can be calculated from the filtered signal using any number of techniques, including maximum amplitude projection, matched filtering, and the like.

[0191] <PARS Optimization> PARS (including related architectures, systems, and methods) provides a unique set of contrasts, so it may be beneficial to further optimize PARS. Some of these optimization techniques or methods for PARS architectures and systems (e.g., system operation) and / or PARS extraction methods (e.g., image processing methods) may include, but are not limited to, the following.

[0192] <PARS Architecture and System Optimization> <1. System Alignment> The PARS architecture and system features several excitation and detection spots that are aligned in an appropriate configuration, e.g., in some embodiments, a confocal configuration, to derive the intended pressure (photoacoustic) and temperature (photothermal) signals, similar to those discussed with respect to Figures 20B-20C.

[0193] <2. Automatic focusing> In some cases, it may be important to acquire sample images and data from an optimal focal plane. As used herein, the term "optimal focal plane" refers to a specific position of any detection or excitation source in a PARS system where the system can acquire the sharpest or clearest images and / or most precise data from the sample. An autofocus algorithm may be used to determine the optimal focal plane for the scatter, non-emission, and / or emission channels for any detection or excitation beam. For a given region of interest (ROI), multiple acquisitions are performed at a specific spacing over a given depth range (e.g., the Rayleigh range of the beam). An exemplary illustration of this is shown in FIG. 21. FIG. 21 shows an example of an axial depth scan over a range of + / - 2 μm with a step size of 500 nm. The focus is determined using a depth scan area that is much smaller than the entire region of interest ROI scan area.

[0194] At each axial position, a scalar focus metric is then computed for the relative sharpness of the given layer. A suitable function (e.g., parabolic or Gaussian) is then fit to the sharpness versus axial position curve. The peak of this fitted function corresponds to the optimal focal plane. An example of this (based on FIG. 21) is shown in FIG. 22.

[0195] A subset of the area from the entire scanned region of interest ROI may be used as the depth scan region. In these cases, the optimal focal plane for the entire region of interest ROI is determined from this representative subset. This is shown in FIG. 21.

[0196] In some PARS systems, all excitation and detection beam spots may be axially co-aligned and therefore share the same optimal focal plane. Therefore, the autofocus algorithm may be implemented for a single wavelength or data channel. For example, only the detection scatter channel may be used to determine the focal plane for all data channels collected using the PARS system.

[0197] The autofocus algorithm may also be used as a tool to guide the axial co-alignment of all detection and excitation spots present in the PARS system. The autofocus algorithm may determine the optimal focal plane for each excitation and detection beam. Using this information, each spot position may be adjusted until their optimal focal planes match / align, thus achieving optimal axial beam overlap.

[0198] 3. Automated Whole Slide Imaging In some instances, PARS is used to measure the entire slide sample (e.g., 1 cm 2 ) at high resolution (e.g., 250 nm per pixel) to capture individual subregions (e.g., 0.5 mm 2 ) can be imaged by scanning them separately, which may later be recombined and stitched together using automated whole-slide imaging. In these cases, these sub-regions, or tiles, are arranged in a grid-like pattern to optimally cover the entire sample area. An example of these sub-regions for a tissue slide is shown in FIG. 23.

[0199] In automated whole-slide imaging, a camera image or slide preview determines tissue(s) boundaries and is then used to divide the sample area into subregions. Tissue boundaries or regions of interest may be manually tracked or automatically determined with a boundary detection algorithm. As an alternative or in addition to a camera, a slide image or preview may also be generated from low- and / or high-resolution scatter from a detection laser. In other architectures, any camera-based imaging (e.g., attenuation, autofluorescence, or brightfield) may be used to guide PARS acquisition while also providing additional details about the specimen.

[0200] In some cases, there may be a small amount of intentional overlap between adjacent tiles, which provides some image redundancy between tiles to aid in stitching and contrast / brightness leveling across all slide images.

[0201] Tiles may be imaged at their optimal focal planes, which may be determined using the autofocus algorithm described above. Referring to FIG. 23, tiles may be flagged as edge tiles (blue) or inner tiles (red). As used herein, the term "edge tiles" refers to tiles that do not cover enough tissue area for the autofocus algorithm to accurately assess focus. For example, if a tile contains mostly background glass, the algorithm may focus on the slide instead of the tissue. Edge tiles may be scanned at their best approximate optimal focus from the same focus as the nearest red tile. Alternatively, the focus of the edge tiles may be found by running the focusing algorithm after masking out pixels corresponding to the glass layer.

[0202] <4. Full slide stitching and contrast leveling> In a full slide image, subtle contrast and brightness variations can exist between adjacent tiles. In these images, these variations exist in each image channel. An example of these contrast variations for the scatter channel is shown in FIG. 24.

[0203] A contrast leveling algorithm may be used to address contrast and brightness variations. The contrast leveling algorithm may be divided into two separate algorithms that are executed sequentially. The first algorithm is the bulk leveling algorithm, which shifts and scales the histogram of each tile by the difference in the mean and standard deviation between the overlapping pixels of the surrounding tiles and the inner tile. The second algorithm corrects the two-dimensional contrast gradient shift between tiles. These shift and scale the individual pixels of each tile based on local intensity statistics interpolated from the difference between the overlapping pixels of the surrounding tiles and the inner tile. FIG. 25 shows the result after executing the algorithm on the image in FIG. 24.

[0204] It is understood that the aspects disclosed in this section may be used with various types of systems and / or architectures, such as hybrid scanning, 2D optical scanning, cameras, line scanning.

[0205] <PARS Extraction Method Optimization> In addition to PARS architecture and system optimization, there may be PARS extraction method optimization methods, for example, image processing methods that are clearly beneficial for PARS compared to images collected from conventional modalities. These methods are developed to specifically function with the intrinsic time-evolving PARS radiative and non-radiative (e.g., photoacoustic and photothermal) data. The methods described herein specifically utilize the intrinsic characteristics of the PARS data channels to filter, enhance, or modify the data in a desired manner. Some of these methods may be, but are not limited to, the following paragraphs.

[0206] <1. Local spatiotemporal averaging> In some cases, the signal-to-noise ratio can be affected by the presence of additive or measurement noise in the system. It may be desirable to mitigate these effects by performing differential filtering or averaging. In the case of PARS time-domain signals, there is high spatial and temporal correlation in the sample. This high degree of correlation can be exploited to denoise the time-domain signals before intelligent clustering or signal extraction operations.

[0207] For example, the data volume is reconstructed according to two spatial axes, with the third axis containing the time-domain signal. This can facilitate spatial-domain processing operations before time-domain signal extraction. The signal is locally averaged in the spatial axes to provide smoothing while preserving information in the time axis. Figure 26 is an example of a local spatiotemporal filtering implementation. The signal is spatially reshaped in the horizontal and vertical dimensions, with time comprising the third dimension. The signal can then be spatially filtered while preserving the temporal signal quality.

[0208] Similar non-intelligent techniques may be implemented in any or all of the PARS radiative, non-radiative, and scattering channels. This method may be applied to a task, such as noise removal, followed by k-means clustering to explore the signal shape as previously explored. The same technique may be applied before extracting the absorbed energy of the signal, as described above.

[0209] <2. Local Statistics Image Smoothing> In some PARS systems, the imaging noise of the non-radiative channel can be closely related to the measurement noise of the detection source. Additive noise becomes larger for PARS non-radiative amplitude modulation as modulation intensity decreases. That is, the signal-to-noise ratio is expected to decrease as the PARS signal level decreases. Given this relationship, PARS images may be filtered based on the assumption that lower signal levels indicate lower SNR. The filtering acts as an adaptive outlier removal method, which aims to correct for local variance in the PARS data based on expected variance and intensity. To perform this filtering, the mean and standard deviation of a local neighborhood are calculated. The center pixel of the neighborhood is corrected to lie within a given variance of the local region. The allowed variance is scaled based on the local intensity, with lower intensities allowing less variance because noise is expected to be higher. It will be appreciated that this filtering method can remove bright or dark outliers from an image without affecting image sharpness or introducing any blurring. An example of this PARS local statistics image filtering method applied to the non-radiative contrast channel of an image captured in a thin section of human skin tissue is shown in FIG.

[0210] <3. Total absorption dispersion correction> In some PARS systems, there can be a high degree of correlation between emissive and nonemissive collections at each pixel location. In these cases, because both contrasts are generated from the same excitation event, common-mode noise associated with the excitation pulse can be present in both the emissive and nonemissive data sets. Also, structures can have significant spatial similarities in both visualizations. However, because each contrast uses a slightly different collection path and mechanism, this is equivalent to making two independent measurements of contrast and excitation noise at each pixel. This high degree of correlation, along with measurement independence, can be exploited to denoise the resulting emissive and nonemissive images.

[0211] The total absorption dispersion correction method may be used to isolate common-mode excitation pulse energy noise from the image. The excitation noise may occur at spatial frequencies above the system resolution in both data sets. To extract sub-resolution excitation-induced fluctuations, the image may be high-pass filtered. This provides two independent measurements of local excitation noise (radiative and non-radiative based). The extracted measurements from each data set (radiative and non-radiative) may then be used to correct for excitation-induced dispersion in the opposite data set. This is equivalent to performing a reference correction of the pulse energy. It will be appreciated that this filtering and correction method does not result in any image blurring, as the corrections are derived from independent sources. An example of total absorption dispersion correction applied to a thin section of human skin tissue is shown in FIG. 28.

[0212] <Example> The following examples describe mechanisms, architectures, systems, and methods according to the present disclosure. However, it should be understood that these examples are provided by way of example, and that nothing in the examples should be construed as a limitation on the overall scope of the present disclosure.

[0213] Example 1 The proposed embodiment uses two excitation sources with a shared detection source. Each excitation wavelength is selected to target the unique radiative and non-radiative absorption properties of local biomolecules. In this example, the first wavelength is 266 nm, which is highly absorbed by DNA. This is shown in Figure 29, which primarily reveals nuclear structure and connective tissue. As shown, this promotes strong non-radiative contrast within the nucleus. The second wavelength is 532 nm, which induces strong non-radiative contrast from heme proteins, which reveals red blood cell structure and connective tissue, as shown in Figure 30. In this example, the detection source is a 405 nm continuous wave source. This was selected to provide high resolution and high sensitivity.

[0214] Radiation relaxation and inelastic scattering at the two wavelengths capture a wide range of the most common biomolecules, such as collagen, elastin, and myelin. There is slight variation in the response of each tissue to the 266 nm (Figure 31) and 532 nm (Figure 32) excitations. In particular, Figures 31 and 32 primarily reveal connective tissue structure and dust contamination artifacts.

[0215] Finally, PARS can also provide light scattering from each of the beams interacting with the sample. This includes both the excitation and detection sources. Scattering images primarily reveal the structural morphology of the sample. An example of the light scattering contrast resulting from a 405 nm detection source is shown in FIG. 33, which mostly captures the structural morphology of the sample. An example of the excitation scattering contrast resulting from a 266 nm excitation source is shown in FIG. 35.

[0216] In addition to morphological information, scattering can also carry details of indirect absorption. In some cases, the scattering intensity of the excitation and detection beams has a small dependence on the local absorption properties of the sample. As shown in Figure 34, red blood cells (encircled in dashed outline) exhibit very high absorption at the detection wavelength of 405 nm. The red blood cells then appear in the scattering image as slightly darker spots. Biomolecules or targets of interest may appear in the light scattering contrast image as spots that are relatively darker than the surrounding non-absorbing medium.

Claims

1. 1. A device as an imaging device for histological and / or molecular imaging of tissue samples, said device comprising: one or more light sources, wherein the one or more light sources (i) one or more excitation beams configured to be directed to an excitation location focused on the tissue sample to generate a signal at the tissue sample; and (ii) one or more interrogation beams configured to be directed at a detection location, wherein a portion of the one or more interrogation beams returning from the tissue sample is indicative of at least some of the generated signals; one or more of said light sources configured to generate a photodetector configured to detect an emission signal from the tissue sample; one or more processors; It is equipped with One or more of the processors generating an image of the tissue sample using only pressure (photoacoustic) signals from the generated signals; generating an image of the tissue sample using only the temperature (photothermal) signal from the generated signals; generating an image of the tissue sample using both photoacoustic and photothermal signals from the generated signals; configured to run Device.

2. the photoacoustic signal used to generate an image of the tissue sample is measured in the range of 1 picosecond to 500 milliseconds of an excitation event caused by one or more of the excitation beams; 10. The apparatus of claim 1.

3. the photothermal signal used to generate an image of the tissue sample is measured in the range of 1 microsecond to 500 milliseconds of the excitation event caused by one or more of the excitation beams; 3. The apparatus of claim 2.

4. The one or more light sources are: a first excitation light source configured to emit light at a first wavelength; and a second excitation light source configured to emit light at a second wavelength different from the first wavelength; Equipped with 10. The apparatus of claim 1.

5. the first and second wavelengths of light are configured to target unique radiative and non-radiative absorption properties of localized biomolecules in the tissue sample.

5. The apparatus of claim 4.

6. One or more of the processors excitation using only the first wavelength; and a photoacoustic signal and / or a photothermal signal from excitation using only the second wavelength; and generating an image based on the 5. The apparatus of claim 4.

7. One or more of the processors (1) the photoacoustic and photothermal signals from excitation using only the first wavelength; and (2) the relative derivative of the photoacoustic and photothermal signals from excitation using only the second wavelength; and and generating an absorption differential image based on the 5. The apparatus of claim 4.

8. the one or more processors are configured to generate transmission and reflection attenuation maps via light scattering contrast images of the one or more interrogation beams or excitation beams.

10. The apparatus of claim 1.

9. Biomolecules or targets of interest appear in the light scattering contrast image as spots that are relatively darker than the surrounding non-absorbing medium.

9. The apparatus of claim 8.

10. The tissue sample comprises one or more of a freshly excised tissue specimen, a preserved tissue specimen, a prepared tissue specimen, an extracted tissue specimen, or an in vivo tissue; 10. The apparatus of claim 1.

11. The apparatus further comprises a temperature control device configured to regulate the temperature of the tissue sample.

10. The apparatus of claim 1.

12. the apparatus further comprises a slide for receiving the tissue sample; the slide comprises a UV transparent material configured to allow imaging through the slide; 10. The apparatus of claim 1.

13. One or more of said processors further comprise: calculating the intensity of the generated signal before excitation; determining a residual modulation by subtracting the calculated intensity before excitation from the generated intensity of the signal after excitation; integrating the residual modulation so as to be integrated; using said integral to estimate the total absorption level of a radiative or non-radiative signal; configured to run 10. The apparatus of claim 1.

14. the one or more processors are configured to apply noise removal or filtering before extracting the integral.

14. The apparatus of claim 13.

15. the one or more processors are configured to generate an image using all of the photoacoustic signal, the photothermal signal, and the radiation signal.

10. The apparatus of claim 1.

16. The one or more processors are configured to generate an image using a quantum efficiency ratio (QER) ratio of (1) the photoacoustic signal and the photothermal signal to (2) the radiative signal.

10. The apparatus of claim 1.

17. One or more of the processors (i) using the Quantum Efficiency Ratio (QER) ratio to define the color of the combined QER-total absorption image; and (ii) using all of the photoacoustic signal, the photothermal signal, and the radiative signal to define the intensity of the combined quantum efficiency ratio (QER)-total absorption image; configured to generate the combined quantum efficiency ratio (QER)-total absorption image; 17. The apparatus of claim 16.

18. the color provides information about the type of biomolecule in the combined quantum efficiency ratio (QER)-total absorption image; the intensity of the combined quantum efficiency ratio (QER)-total absorption image provides information about the concentration of the biomolecule.

18. The apparatus of claim 17.

19. the one or more processors are further configured to generate a visualization of the extracted time-domain features, distinguishing different biomolecules with different colors.

10. The apparatus of claim 1.

20. the apparatus further comprises a secondary imaging head; the secondary imaging head is a camera-based detector configured to perform wide-area, high-resolution imaging at high rates; 10. The apparatus of claim 1.

21. The one or more light sources are: (i) a white light source; and (ii) isolated wavelengths; and and one or more of: The one or more light sources are: (i) a bright field image; (ii) measuring light attenuation in the specimen; and (iii) measuring autofluorescence in the specimen; configured to provide one or more of:

10. The apparatus of claim 1.

22. one or more of the excitation beams and / or one or more of the interrogation beams underfill an objective lens used for histological and / or molecular imaging of the tissue sample; 10. The apparatus of claim 1.

23. wherein said excitation beam(s) and / or said interrogation beam(s) accurately fill or overfill an objective lens used for histological and / or molecular imaging of said tissue sample; 10. The apparatus of claim 1.

24. the one or more processors are further configured to generate an image using the radiation signals detected by the photodetector; the emitted signal is autofluorescence; 10. The apparatus of claim 1.

25. the photodetector is configured to detect a non-radiative signal dominated by the temperature (photothermal) signal; 10. The apparatus of claim 1.

26. the photodetector is configured to detect a non-radiative signal dominated by the pressure (photoacoustic) signal; 10. The apparatus of claim 1.

27. 1. A method for configuring a scanning system to scan a sample at different pixels spaced a determined distance from each other, the different pixels corresponding to different locations of an excitation event, the method comprising: performing a first scan at two or more pixels in an area of ​​the sample; determining the determined distance between pixels from the first scan, the determined distance corresponding to a minimum distance that allows for extracting a signal from a particular pixel such that the particular pixel is completely isolated and the sample is allowed to return to thermal equilibrium before being excited again; Synchronizing the oscillation frequency of the MEMS mirror with the pulse repetition frequency PRF of the laser source; optically scanning the beam generated by the laser source at each pixel in the subgroup via the MEMS mirror across the sample, the oscillation frequency synchronized with the pulse repetition frequency PRF allowing the beam to be pulsed at the determined distance onto the sample; The method comprises:

28. The method further comprises: generating one or more additional beams generated by the laser source and / or one or more additional laser sources, each beam of the one or more additional beams corresponding to a distinct sub-group of pixels; optically scanning each of the one or more additional beams across the sample via the MEMS mirror at each pixel in the corresponding subgroup, wherein a synchronized scanning frequency and the pulse repetition frequency PRF enable the one or more additional beams to be pulsed onto the sample at the determined distance; Equipped with 28. The method of claim 27.

29. the beam and each of the one or more additional beams are spaced apart from one another by the determined distance; 29. The method of claim 28.

30. 1. A method for configuring a scanning system to scan a sample at different pixels spaced a determined distance from each other, the different pixels corresponding to different locations of an excitation event, the method comprising: performing a first scan at two or more pixels in an area of ​​the sample; determining the determined distance between pixels from the first scan, the determined distance corresponding to a minimum distance that allows for extracting a signal from a particular pixel such that the particular pixel is completely isolated and the sample is allowed to return to thermal equilibrium before being excited again; pulsing a plurality of beams generated by one or more laser sources at a pulse repetition frequency PRF, each beam of the plurality of beams corresponding to a distinct subgroup of pixels; optically scanning the plurality of beams across the sample via a MEMS mirror at each pixel in the corresponding distinct subgroup, whereby the plurality of beams are pulsed onto the sample at the determined distance; The method comprises:

31. the pulse repetition frequency PRF is greater than the vibration frequency of the MEMS mirror; 31. The method of claim 30.

Images