3D pathology-guided microdissection system

Abstract

A 3D pathology-guided microdissection (3DPM) system includes a 3D imaging system, a tissue extraction system, and a computing system. The computing system operates the 3D imaging system to collect a 3D image of a sample. One or more volumes of interest (VOIs) are identified in the 3D image. The computing system generates cutting instructions for the VOIs. The computing system operates the tissue extraction system to remove the VOIs from the sample. Further analysis may then be performed on the removed volumes. Information gained from the analysis may be related back to the 3D relationship between the VOIs in the sample from the 3D image.

Description

3D PATHOLOGY-GUIDED MICRODISSECTION SYSTEMCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the filing benefit of U.S. Provisional Application No.63 / 790,979, filed April 18, 2025. This application is incorporated by reference herein in its entirety and for all purposes.FEDERALLY SPONSORED RESEARCH AND DEVELOPMENT

[0002] This invention was made with government support under Grant No. R01EB031002, awarded by the National Institute of Biomedical Imaging and Bioengineering. The government has certain rights in the invention.BACKGROUND OF THE INVENTION

[0003] Image-guided tissue microdissection enables the extraction of cells or tissue regions of interest (ROIs) identified in imaging data. The extracted tissue can be further used in downstream molecular assays to integrate image with molecular profiles. For example, pathologists dissect morphologically distinct tumor regions from 2D H&E slides to study tumor phylogeny using genomics tools, while neuroscientists cut specific brain regions and apply mass spectrometry to analyze their protein composition.

[0004] However, current methods — such as manual macrodissection and laser-capture microdissection (LCM) — are limited to processing thin slices guided by 2D images (<15 pm thick). As biological samples are intrinsically 3D structures, traditional 2D imaging is inadequate to capture the spatial context useful for advanced biological research. In contrast, 3D imaging has emerged as a key tool across various fields, enabling comprehensive spatial characterization.

[0005] Despite these advancements in 3D imaging, no existing methods allow for 3D microdissection. There may be a need for 3D guided microdissection, for example to facilitate the integration of 3D imaging and molecular profiles to advance research in cancer evolution and other biomedical fields.BRIEF SUMMARY OF THE INVENTION

[0006] In at least one aspect, the present disclosure relates to a system including a 3D imaging system that generates a 3D image of at least a portion of a sample, a computing 1 4898-7105-8594 1system, and a tissue extraction system. The computing system includes a processor and a memory encoded with non-transitory instructions which, when executed by the processor, cause the system to generate 3D coordinates of a boundary of a volume of interest (VOI) identified in the 3D image of the at least a portion of the sample, translate the 3D coordinates of the VOI from an imaging coordinate system of the 3D image to an extraction coordinate system, and generate cutting path instructions in the extraction coordinate system for the VOI based on the registered 3D coordinates of the VOI. The tissue extraction system extracts at least a portion of the VOI from the sample using the cutting path instructions.

[0007] The system may also include a sample holder that includes at least one registration mark. The sample holder holds the sample. Coordinates of the at least one registration mark may be co-registered to both the imaging coordinate system and the cutting coordinate system. The computing system may perform the translation of the 3D coordinates of the VOI based, at least in part, on the co-registered coordinates of the at least one registration mark. The 3D imaging system may include an imaging stage with a recess that holds at least a portion of the sample holder and the tissue extraction system may include an extraction stage with a recess that holds at least a portion of the sample holder. The sample holder may rest in the recess of the imaging stage in an inverted position and rest in the recess of the extraction stage in an upright position. The instructions, when executed by the processor, may cause the system to form the registration mark on the sample holder with the tissue extraction system with first coordinates in the extraction coordinate system, image the registration mark with the 3D imaging system to record second coordinates of the registration mark in the imaging coordinate system, and co-register the imaging coordinate system to the extraction coordinate system based on the first coordinates and the second coordinates of the registration mark.

[0008] The 3D imaging system may include an open-top light sheet microscope. The 3D imaging system may include a two-photon microscope, a confocal microscope, a light-field microscope, a wide field deconvolution system, an optical coherence tomography system, a reflectance OCT system, a reflectance confocal system, a non-linear microscopy system, a phase-contrast imaging system, a holotomography system, a microCT imaging system, other forms of non-contact 3D imaging, or combinations thereof. The tissue extraction system may include a 5-axis computer numerical control (CNC) mill, one or more robotic arms, or combinations thereof.2 4898-7105-8594 1

[0009] The instructions, when executed by the processor, may cause the computing system to collect a sequence of images from the 3D imaging system in a stack and generate the 3D image from the stack. The 3D image may include a fluorescent image, a reflectance optical coherence tomography image, a reflectance confocal image, a non-linear microscopy image, a phase-contrast image, holotomography, a microCT image, or combinations thereof.

[0010] The instructions, when executed by the processor, may cause the system to generate and optimize a set of cutting parameters for the VOI based on the 3D coordinates of the boundary of the VOI, and generate the cutting path instructions based on the optimized set of cutting parameters. The instructions, when executed by the processor, may cause the system to generate a plurality of sets of cutting parameters, select a best one of the plurality of sets of cutting parameters based on the simulated tissue extraction, and determine if the best one of the plurality of sets of cutting parameters meets one or more goals and (1) if the best one of the plurality of sets of cutting parameters does not meet the one or more goals generate a new plurality of sets of cutting parameters, based at least in part on the best one of the plurality of sets of cutting parameters and iteratively repeat the simulated tissue extraction with each of the new plurality of sets of cutting parameters, or (2) if the best one of the plurality of sets of cutting parameters does meet the one or more goals provide the best one as the optimized set of cutting parameters. The instructions, when executed by the processor, may cause the system to generate and optimize the set of cutting parameters with a genetic algorithm.

[0011] In at least one aspect, the present disclosure relates to a method that includes imaging a sample with a 3D imaging system to generate a 3D image of the sample, identifying 3D coordinates of a volume of interest (VOI) based on the 3D image with a computing system, translating the 3D coordinates of the VOI from a coordinate system of the 3D imaging system to a coordinate system of a tissue extraction system with the computing system, generating cutting path instructions for the VOI based on the translated 3D coordinates of the VOI with the computing system, and cutting at least a portion of the VOI from the sample with a tissue extraction system using the cutting path instructions.

[0012] The method may also include mounting the sample to a sample holder, positioning the sample holder in a stage of the 3D imaging system for the imaging, and positioning the sample holder in a stage of the tissue extraction system for the cutting. The method may also include positioning the sample holder in the stage of the 3D imaging system in an inverted position, and positioning the sample holder in the stage of the tissue extraction3 4898-7105-8594 1system in an upright position. The method may also include mounting the sample to the sample holder by gluing the sample to the sample holder.

[0013] The method may also include making a registration mark on the sample holder with the tissue extraction system, imaging the registration mark on the sample holder with the 3D imaging system, and registering the coordinate system of the 3D imaging system to the coordinate system of a tissue extraction system based on a registration mark. The method may also include operating the tissue extraction system with the computing system based on the cutting path instructions.

[0014] The method may also include generating the cutting path instructions by optimizing one or more cutting parameters based on one or more metrics using an iterative optimization method. The one or more parameters may include a depth of a plane, angles of one or more cuts that pass through an edge inscribed in the plane, or combinations thereof. The one or more metrics may include recall, precision, amount of damage to surrounding tissue, or combinations thereof. The iterative optimization method may be a genetic algorithm, swarm-based algorithms, reinforcement learning, or combinations thereof.

[0015] The method may also include identifying respective 3D coordinates of a plurality of VOIs in the sample based on the 3D image, generating cutting path instructions with the computing system for the plurality of VOIs by sequentially optimizing a set of cutting parameters for each of the plurality of VOIs based on the respective 3D coordinates, and cutting a plurality of volumes from the sample based on the cutting instructions with the tissue extraction system, wherein each of the plurality of volumes is associated with one of the plurality of VOIs. The method may also include performing an omics analysis on each of the plurality of volumes, and relating omics information associated with each of the plurality of volumes to a 3D relationship between the plurality of VOIs in the sample.BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a block diagram of an example 3D pathology-guided microdissection (3DPM) system according to some embodiments of the present disclosure.

[0017] Figure 2 is a flow chart of a method of 3DPM according to some embodiments of the present disclosure.

[0018] Figure 3A is an isometric schematic drawing of a 3D imaging system stage according to some embodiments of the present disclosure.

[0019] Figure 3B is an isometric schematic drawing of a tissue extraction stage according to some embodiments of the present disclosure.4 4898-7105-8594 1

[0020] Figure 4 is an image of an example registration mark on a sample holder according to some embodiments of the present disclosure.

[0021] Figure 5 is a flow chart of a visualization of cutting instruction generation and optimization according to some embodiments of the present disclosure.

[0022] Figure 6 is a flow chart of an example implementation of a method of cutting parameter optimization according to some embodiments of the present disclosure.

[0023] Figure 7 is a flow chart of a method of generating cutting path instructions based, in part on optimized cutting parameters according to some embodiments of the present disclosure.

[0024] Figure 8 is flow chart of a method of using a 3DPM system according to some embodiments of the present disclosure.

[0025] Figure 9 is a figure showing an image of phylogenetic information determined from using a 3DPM on a tumor biopsy according to some embodiments of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0026] The following description of certain embodiments is merely exemplary in nature and is in no way intended to limit the scope of the disclosure or its applications or uses. In the following detailed description of embodiments of the present systems and methods, reference is made to the accompanying drawings which form a part hereof, and which are shown by way of illustration specific embodiments in which the described systems and methods may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice presently disclosed systems and methods, and it is to be understood that other embodiments may be utilized and that structural and logical changes may be made without departing from the spirit and scope of the disclosure. Moreover, for the purpose of clarity, detailed descriptions of certain features will not be discussed when they would be apparent to those with skill in the art so as not to obscure the description of embodiments of the disclosure. The following detailed description is therefore not to be taken in a limiting sense, and the scope of the disclosure is defined only by the appended claims.

[0027] There are a number of applications where it is useful to identify one or more structures of interest in a tissue sample and then extract those structures for further processing. 3D imaging techniques, such as 3D microscopy, may be especially useful for identifying structures of interest. For example, the 3D morphology of the structures may be visualized, biomarker specific contrast agents (e.g., fluorescent dyes) may be used to identify specific tissues of interest, or combinations thereof. It may also be useful to determine5 4898-7105-8594 1information about region-specific properties within a tissue of interest. For example, by extracting the structure of interest to study the molecular biology of the extracted tissue, such as to perform genome sequencing on the extracted tissue to track gene expression. There is a growing recognition that it is useful to link the information gained from the extracted tissue to the spatial context that the tissue occupied in situ. However, common techniques for both imaging and for extraction may generally be limited to relatively thin sections that lose the 3D spatial context of the original sample. There remains an unmet need for technologies to enable integrated spatial-molecular analyses with comprehensive high-depth omics (e.g., genomics, proteomics, metabolomics, or combinations thereof) at the cell-population level in conjunction with high-resolution 3D morphological information across relatively large (e.g., mm to cm) length scales.

[0028] The present disclosure is drawn to apparatuses, systems and methods for 3D pathology-guided microdissection. A 3D pathology-guided microdissection (3DPM) system of the present disclosure includes a 3D imaging system and a tissue extraction system. A sample (e.g., biopsied tissued) is attached to a sample holder. The 3D imaging system generates 3D images, or stacked 2D images, of at least a portion of the sample. A volume of interest (VOI) is identified in the 3D images. The VOI has a boundary with 3D coordinates that are registered to both an imaging coordinate system of the 3D imaging system and an extraction coordinate system of the tissue extraction system. For example, the sample holder may include a registration mark that has coordinates co-registered to both the coordinate system of the 3D imaging system and the coordinate system of the tissue extraction system. Cutting path instructions are generated for at least a portion of the VOI based on the co-registered coordinates of the VOI. The sample holder and sample are transferred to the tissue extraction system. The tissue extraction system uses the cutting path instructions to extract at least a portion of the VOI from the sample.

[0029] In an example implementation, a 3D microscope is used as the imaging system and a five-axis computer numerical control (CNC) milling machine is used as the tissue extraction system. The sample is mounted to a sample holder to prevent sample movement during microdissection, imaging, and movement between the 3D imaging system and the tissue extraction system. For example, the bottom of the tissue is glued to the tissue holder. This may be particularly useful in embodiments where the 3D microscope is an inverted microscope, such as an open top light sheet (OTLS) microscope. When placed on the inverted microscope, the sample holder may be in an inverted position (e.g., with the sample hanging down below the sample holder) and fit onto the sample holder on the6 4898-7105-8594 1microscope. Before imaging, co-registration of the 3D imaging system and the five-axis CNC milling machine is performed based on the registration mark. This allows the coordinates of the 3D images to match the coordinates of the 5-axis CNC milling machine. The coregistered 3D imaging system is then used to capture 3D images, such as 3D pathology images. A manual process, an automatic process, or combinations thereof is used to identify one or more VOIs in the sample based on the 3D image. These annotations are then used to compute the optimized 3D cutting instructions for the selected VOIs. The sample and sample holder is transferred to the five-axis CNC milling machine for 3D microdissection. The CNC milling machine extracts at least a portion of each of the selected VOIs based on the optimized 3D cutting instructions. Subsequent analysis, such as an omics analysis (e.g., genomics, proteomics, metabolomics, etc.) may then be performed on the extracted VOIs and the information from the analysis related to the original 3D relationship between the VOIs in the tissue.

[0030] Figure 1 is a block diagram of an example 3D pathology-guided microdissection (3DPM) system according to some embodiments of the present disclosure. The 3DPM system 100 includes a 3D imaging system 110, a tissue extraction system 120 and a sample holder 102 that is configured to be placed in either the 3D imaging system 110 or the tissue extraction system 120. Also shown as part of the 3DPM system 100 is a computing system 130 that is used to operate the 3D imaging system 110 and the tissue extraction system 120. In the example illustration of Figure 1, a single computing system 130 is shown, but in other examples multiple computing systems may be used, such as a first computing system that operates the 3D imaging system 110 and a second computing system that controls the tissue extraction system 120. Similarly, Figure 1 shows the computing system 130 as being separate from the 3D imaging system 110 and tissue extraction system 120. However, in other embodiments, all or part of the computing system 130 may be integrated into one or both of the 3D imaging system 110, tissue extraction system 120, or combinations thereof.

[0031] The 3DPM system 100 is used to generate a 3D image of a sample 104, such as a tissue of interest, identify one or more VOIs within that sample 104 based on the 3D image, and then extract at least a portion of at least some of the identified VOIs. The sample 104 is mounted on a sample holder 102. The sample holder 102 is positioned on the 3D imaging system 110 and the computing system 130 operates the 3D imaging system 110 to generate a 3D image of all or a portion of the sample 104. The computing system 130 is then used, either through a manual process, an automatic process, or combinations thereof, to identify VOIs in the 3D pathology image. The computing system 130 generates cutting path7 4898-7105-8594 1instructions based on the VOIs and co-registered coordinate systems of the 3D image and the tissue extraction system 120. The sample holder 102 and sample 104 are moved to the tissue extraction system 120. The computing system 130 operates the tissue extraction system 120 using the cutting path instructions to extract the VOIs.

[0032] In an example use case, the extracted VOIs may then be used for subsequent processing, for example to analyze omics information in the extracted material. Since the VOI was identified from the 3D image, the broader 3D context of the VOI is maintained and the omics information may be related back to the original 3D spatial relationship between the VOI(s) and the sample 104. For example, if multiple VOIs are identified and extracted, then the omics information for each of the extracted VOIs may be related to the 3D spatial relationship that was originally present between those VOIs.

[0033] The 3D imaging system 110 generates 3D images of the sample 104. In some embodiments, the 3D imaging system may generate 'true' 3D images. In some embodiments, the 3D imaging system may generate multiple 2D (or thin 3D) images in an image stack. The images of the image stack may be combined by the computing system 130 to form a 3D view of all or a portion of the sample 104. Different imaging modalities, such as optical imaging (e.g., microscopy), tomographic imaging (e.g., computed tomography), magnetic resonance imaging (MRI), or other imaging modalities may be used by the imaging system 110 in different example embodiments. In the example of Figure 1, a 3D microscope is shown as the 3D imaging system 110, in particular an open-top light sheet (OTLS) microscope. For the sake of consistency, the present disclosure is generally described with respect to an OTLS microscope as the 3D imaging system. However, other microscopes, and other imaging modalities may be used in other example implementations.

[0034] The OTLS 3D imaging system 110 includes an illumination source 112, illumination optics 114, collection optics 116, a detector 118, and a stage 117 coupled to an optional actuator 119. Tn illumination source 112 generates light that is directed to the illumination optics 114. The illumination optics 114 generate illumination light in the form of a generally planar light sheet. The light sheet illuminates a region of the sample 104 when the sample 104 and sample holder 102 are placed on the stage 117. Light from the illuminated region of the sample 104 is collected by collection optics 116 and directed to a detector 118 that generates images of the collected light from the illuminated region of the sample 104.

[0035] The illumination source 112 generates light used to illuminate a region of the sample 104. The illumination source 112 may be a broad spectrum source or may generate light over a relatively narrow frequency rage. In some embodiments, the light source 112 may8 4898-7105-8594 1be an incandescent source, an LED source, a laser source, or any other light source. The illumination optics 114 include one or more optical elements that shape, focus, filter, redirect, otherwise condition, or combinations thereof the light from the illumination source 112 to the sample 104. The illumination optics 114 may include one or more mirrors, filters, lenses, apertures, or combinations thereof. In some embodiments, the illumination optics 114 may include scanning optics. For example, if the illumination source 112 is a laser, then the scanning optics may scan the light across a target region within the sample 104.

[0036] The collection optics 116 receive light from a region of the sample 104 illuminated by the illumination light from the illumination optics 114. The collection optics 116 include one or more mirrors, filters, lenses, apertures, or combinations thereof. The collection optics 116 shape, focus, filter, redirect, otherwise condition, or combinations thereof light from the illuminated region of the sample 104 to the detector 118. The detector 118 generates an electronic signal based on the received light. For example, the detector may be a CCD device. The detector generates image information (e.g., one or more image files) based on the received light and sends it to the computing system 130.

[0037] In the example OTLS system shown as the 3D imaging system 110 in Figure 1, the illumination optics 114 include an illumination objective lens, and the collection optics 116 include a collection objective lens different than the illumination objective lens. The illumination objective directs a light sheet into the sample 104. The light sheet is generally elongated in one direction and generally flat in a second direction orthogonal to the first direction, which may be visualized as a relatively flat sheet of light extending along a principle optical axis of the illumination objective. The light sheet allows the light to penetrate relatively far into the sample 104, illuminating a 'slice' of the sample. In some embodiments, the light sheet directed from the illumination objective 114 and the principle optical axis of the collection objective may be angled relative to each other, for example at 90°. In some embodiments, both the light sheet and the principle optical axis of the collection objective may also be angled with respect to the sample holder 102 when the sample holder is properly positioned in the stage 117. For example, the light sheet and the principle optical axis of the collection objective may both be angled at about 45° with respect to a normal of the sample holder 102. Other angles may be used in other example embodiments.

[0038] The 3D imaging system 110 includes a stage 117 that supports the sample holder 102. In the example geometry of Figure 1, the OTLS is an inverted microscope, where the illumination optics 114 and collection optics 116 are below the stage 117. The sample9 4898-7105-8594 1holder 102 is positioned on the stage 117 inverted with the sample 104 on an 'underside' of the sample holder 102 so that it faces the illumination optics 114 and collection optics 116. The sample 104 may be mounted to the sample holder 102 to allow it to 'hang' down from the sample holder 102 in the inverted position. Other geometries are possible in other example embodiments. In some embodiments, the stage 117 may be at least partially made from a transparent material to allow the illumination light to reach the sample 104 and light from the sample to reach the collection optics 116. In some embodiments, the stage 117 may include one or more lenses and / or windows to allow light to pass.

[0039] In some embodiments, index matching fluids may be used in the OTLS system. For example, one or both of the illumination objective and collection objective may be immersion objectives, and immersion oil may be used. In some embodiments, the sample 104 may be suspended in an immersion fluid. For example, in the geometry shown in Figure 1, the stage 117 includes an immersion bath that can be at least partially filled with an immersion fluid, and the sample is 104 is suspended in the immersion fluid when the sample holder 102 is placed on the stage 117.

[0040] In some embodiments, the 3D imaging system 110 may adjust where the image is being collected in the sample 104. For example, the illumination light, the region light is collected from, or both may be moved relative to the sample, the sample may be moved relative to the illumination light, the region light is collected from, or both, or combinations thereof. In the example of Figure 1, the stage 117 is coupled to an optional actuator 119. The actuator 119 may be operated, for example manually (e.g., via knobs by a user), automatically (e.g., by motors controlled by the computing system 130) or combinations thereof to move the stage 117, and thus the sample 104 relative to the imaging plane of the 3D imaging system 110. The imaging system 110 may track coordinates of where each image is captured within a coordinate system of the imaging system 110. While Figure 1 shows an embodiment where the stage is moved relative to the optics, other embodiments may adjust the imaging location using other methods.

[0041] The computing system 130 generates a 3D image of all or a portion of the sample 104 based on the image(s) collected from the 3D imaging system. In some embodiments, the 3D image may extend several mm into the tissue. For example of the 3D image may have a thickness (extending from the surface of the sample 104) of up to about 5 mm in some embodiments. The 3D image may take the form of a 3D pathology dataset in some example embodiments.10 4898-7105-8594 1

[0042] While an OTLS is shown as the example 3D imaging system in Figure 1, other imaging systems may be used in other example implementations. For example, the 3D imaging system may by a different type of open-top 3D microscopy system, such as a two-photon microscope, a confocal microscope, a light-field microscope, a wide field deconvolution system, an optical coherence tomography (OCT) system, or another type of system. In some embodiments, the 3D image may be based on reflectance OCT, reflectance confocal, non-linear microscopy (e.g., SHG, SRS, CARS, or others), phase-contrast imaging, holotomography, microCT imaging, other forms of non-contact 3D imaging, or combinations thereof. In some embodiments, the 3D images may include multiple imaging modalities. For example, fluorescence images may be overlaid on reflectance confocal images to generate an overlay of fluorescently labeled tissues on the reflectance image. In some embodiments, the same 3D imaging system 110 may perform multiple imaging modalities, such as performing both the fluorescent microscopy and reflectance micrsocopy.

[0043] The tissue extraction system 120 may be operated to remove material from the sample 104. For example, the tissue extraction system 120 may remove material around a VOI in order to allow the VOI to be removed from the sample 104. Various methods of removing material may be used. For example, the tissue extraction system 120 may include an optical tissue removal tool (e.g., a laser), a physical tissue removal tool (e.g., a cutting element) or any other method of removing material.

[0044] The example tissue extraction system 120 shown in Figure 1 includes a cutting tool 122 coupled to an actuator 124. The tissue extraction system includes a stage 126 configured to hold the sample holder 102. In some embodiments, the sample holder 102 may fit in the stage 126 of the tissue extraction system 120 in an upright position (e.g., with the sample 104 facing up). The actuator 124 of the tissue extraction system 120 may move the cutting tool 122 relative to the sample holder 102, the sample holder 102 relative to the cutting tool 122, or combinations thereof. In some embodiments, the tissue extraction system 120 may be a computer controlled cutting system, such as a CNC mill. In some embodiments, the actuator 124 may operate across multiple axes, such as a 5-axis CNC mill. In some embodiments, the tissue extraction system 120 may include one or more robotic arms.

[0045] The computing system 130 includes one or more processors 132, an input / output system 134 such as a mouse, keyboard, touchscreen, or combinations thereof, a display 138, and a communications module 136 that couples it via wired communications, wireless communications, or combinations thereof, to the imaging system 110 and the extraction11 4898-7105-8594 1system 120. The computing system also includes a memory 140 which stores various information and encodes instructions 150 that cause the processor(s) to operate the 3D imaging system 110 and the tissue extraction system 120. In some embodiments, the computing system 130 may represent one or more general purpose computing devices such as desktops, laptops, tablets, smart phones, other devices, or combinations thereof. In some embodiments, the computing system 130 may represent specialized computing devices such as integrated hardware of the 3D imaging system, the tissue extraction system 120, or combinations thereof. In some embodiments, the computing system 130 may represent a combination of general purpose and specialized computing devices. In some embodiments, the computing system 130 may be distributed across multiple devices. For example, a first computer may operate the 3D imaging system while a second computer operates the tissue extraction system. In some embodiments, the computing system 130 may represent one or more networked components, cloud computing elements, or other distributed computer resources. For example, the portions of the computing system 130 that a user interacts with may be remote from the 3D imaging system 110 and / or tissue extraction system 120.

[0046] The computing system 130 includes a memory 140 that includes information and instructions 150 to operate the computing system 130. For example, the memory 140 may represent a non-transitory memory loaded with digital information. The instructions 150 may represent programs loaded on the memory 140.

[0047] The memory 140 stores imaging coordinate information 142 and extraction coordinate information 144. The imaging coordinate information 142 represents a coordinate system used by the 3D imaging system 110. The extraction coordinate system 144 represents a coordinate system used by the tissue extraction system 120. For example, the imaging coordinate information 142 may indicate how the coordinate system used by the 3D imaging system 110 (e.g., used by the actuator 119) relates to physical locations relative to the sample holder 102. Similarly, the extraction coordinate information 144 may indicate how the coordinate system used by the tissue extraction system 120 (e.g., used by the actuator 124) relates to physical locations relative to the sample holder 102.

[0048] The instructions 150 include a first set of instructions 152 for registering the imaging coordinate system 142 and extraction coordinate system 144 to each other. The computing system 130 may co-register these coordinate systems so that coordinates of images obtained by the imaging system 110 may be translated to coordinates that can be used by the tissue extraction system 120 to locate selected VOIs in the sample 104 when the sample12 4898-7105-8594 1holder 102 is positioned on the tissue extraction system 120. In an example registration procedure, the sample holder 102 is placed on the stage 126 of the tissue extraction system 120. The instructions 152 are used to operate the tissue extraction system 120 to inscribe one or more registration marks on the sample holder 102. The sample holder 102 is then moved to the stage of the 3D imaging system 110 and the computing system 130 directs the 3D imaging system 110 to capture an image of the registration mark(s). The registration mark(s) has a fixed location relative to the sample holder 102, and also has coordinates in both the imaging coordinate system from the image and the extraction coordinate system from when the mark was made. This allows the computing system 130 to register the location of the registration mark(s) relative to both coordinate systems, allowing the coordinate systems to be co-registered to each other. For example, the computing system 130 may develop a translation to convert between the two coordinate systems 142 and 144 in order to co-register the coordinates.

[0049] In some embodiments, a contrast agent such as a fluorescent dye may be used to aid imaging of the registration mark. For example, the contrast agent may be applied to the sample holder 102 after the tissue extraction system 120 has made the registration mark(s).

[0050] In another example implementation, the co-registration of the coordinate systems may be performed by a separate system (e.g., a camera, a stereoscopic imaging system, a profilometer, or other) which captures an image of the sample holder 102, and which has an imaging coordinate system registered to both the 3D imaging system's coordinate system 142 and the tissue extraction system's coordinate system 144. In another example implementation, the 3D imaging system 110 and the tissue extraction system 120 may be integrated with a common translation stage which allows the imaging coordinate system 142 and the tissue extraction coordinate system 144 to be unified into a single coordinate system. For example, the stages 117 and 126 may be the same stage in some embodiments, and the actuators 119 and 124 may be the same actuator in some embodiments.

[0051] The instructions 150 include a second set of instructions 154 that cause the computing system 130 to operate the 3D imaging system 110 to generate a 3D image of the sample 104. For example, the instructions 154 may capture a first image from the detector 118, then cause the actuator 119 to shift the sample 104 relative to the imaging field of view so a second image can be captured. By building up a stack of such images, a 3D image of all or a portion of the sample may be generated by the computing system 130. The images may have known coordinates in the coordinate system of the imaging system 142, for13 4898-7105-8594 1example based on the known movements of the actuator 119 to obtain those images. In some embodiments, the computing system 130 may direct the 3D imaging system 110 to capture multiple images or multiple sets of images (e.g., multiple depth stacks) that are mosaiced together to form an overall 3D image.

[0052] The instructions 150 may allow for generation of 3D coordinates of one or more VOIs in the 3D image. In some embodiments, an automatic process may be used to locate the borders of a VOI of interest. In some embodiments, a manual process may be used. In some embodiments, a mixture of automatic and manual processes may be used. In an example implementation, a user may review the images on the display 138 and mark the borders of the VOI using the I / O system(s) 134. In an example implementation, a user may review each slice in the stack of images and draw a 2D border around a region of interest, and those 2D borders are combined to form the 3D coordinates of the VOI boundary. The 3D coordinates of the VOI may be in the coordinate system 142 of the 3D imaging system. The computing system 130 may use the translation set up by the co-registration to translate the 3D coordinates of the VOI into the coordinate system of the tissue extraction system.

[0053] The instructions 150 include instruction subset 156 for generating cutting instructions for the tissue extraction system 120 based on the registered 3D coordinates of the VOI. The cutting path instructions may be generated based on the 3D coordinates of the VOI translated into the coordinate system 144 of the tissue extraction system based on the co-registration of the two coordinate systems. The cutting path instructions may represent a sequence of movements of the actuator 124 of the tissue extraction system 120 to remove the VOI from the sample 104. In some embodiments, an automated process or semi-automated process may be used to generate the cutting path instructions. For example, an iterative process such as a machine learning model, genetic algorithm, swarmbased algorithm, reinforcement learning, or combinations thereof may be used to optimize capturing the maximum volume of the VOI without taking too much extra material. Figure 1 shows an example implementation where a cutting path genetic algorithm 146 is implemented by the computing system 130.

[0054] In some embodiments, the cutting path instructions may be generated based on an iterative process that optimizes one or more parameters in light of one or more metrics. For example, the parameters may include a depth of an initial milling step, the angle of one or more detachment cuts (e.g., four quadrant detachment cuts) , or both and the metrics may include recall, precision, amount of damage to surrounding tissue (e.g., surrounding VOIs), or combinations thereof. In some embodiments, the amount of the sample removed14 4898-7105-8594 1may be on a mm scale. For example, the removed area may be as small as 1mm x 1mm on a side and about 0.5mm thick in some embodiments.

[0055] Figure 2 is a flow chart of a method of 3DPM according to some embodiments of the present disclosure. The method 200 represents an example workflow for 3DPM. In some embodiments, the method 200 may be performed by the 3 DPM system 100 of Figure 1. The method 200 is represented by a sequence of images that show example components (such as the components of the 3DPM system 100 of Figure 1) that may be used to implement the method 200. The method 200 will generally be described with respect to the components shown in Figure 2, however other components may be used in other example embodiments.

[0056] The method 200 includes operation 210 that includes mounting a tissue sample 202 to a sample holder 204 (e.g., sample holder 102 of Figure 1). Operation 210 is generally followed by operation 220 that includes performing 3D imaging of the sample 202 using a 3D imaging system 222 (e.g., 110 of Figure 1). Operation 220 is generally followed by operation 230 that includes generating a 3D image, in this case a 3D pathology dataset 232 based on the 3D imaging. Operation 230 is generally followed by operation 240 that includes annotating one or more VOIs 242 in the 3D pathology dataset 232. Operation 240 is generally followed by operation 250 that includes generating cutting path instructions for the one or more VOIs. Operation 250 is generally followed by operation 260 that includes performing microdissection with a tissue extraction system 262 (e.g., 120 of Figure 1) to remove the one or more VOIs from the sample 202. Operation 270 includes an example workflow for removing a given VOI from the sample 202.

[0057] Operation 210 includes mounting the tissue sample 202 to the sample holder 204.For example, the sample 202 may be adhered to the sample holder 204. In an example mounting operation, a glue, such as ethyl cyanoacrylate, may be used to adhere the sample 202 to the sample holder 204. The sample 202 may be a biological sample. The sample 202 may be prepared in vitro, grown in vivo, or a mixture thereof (e.g., biopsied cells subsequently cultured). In some embodiments, the sample 202 may be a cultured tissue. In some embodiments, the sample 202 may be extracted from a subject, such as a human or veterinary subject. For example, the sample 202 may represent a biopsy sample from the subject.

[0058] In some embodiments, the method 200 may include performing one or more sample preparation steps on the sample 202 before mounting the sample 202 on the sample holder 204, after mounting the sample 202 on the sample holder 204, or both. For example, optical15 4898-7105-8594 1clearing may be performed, a contrast agent may be applied, or combinations thereof. In some embodiments, the imaging system may be a fluorescent microscope and one or more fluorescent dyes may be applied to the sample. In some embodiments, the fluorescent dyes may be targeted to stain biomarkers associated with particular tissues of interest.

[0059] Operation 220 includes performing 3D imaging of the sample 202 using a 3D imaging system 222. Similar to Figure 1, Figure 2 shows the 3D imaging system as an OTLS microscope. The imaging may include placing the sample holder 204 on a stage 224 of the OTLS microscope 222. In the example geometry of Figure 2, placing the sample holder 204 on the stage 224 includes placing the sample holder 204 in an inverted position with an 'upper' side of the sample holder 204 (the side that the sample 202 is mounted to) facing down and a lower side of the sample holder 204 opposite the upper side facing up. The operation 220 may include immersing the sample 202 in an immersion fluid 226. For example, the stage 224 may have an immersion bath filled with the immersion fluid 226 either before or after the sample holder 204 is positioned on the stage 224.

[0060] The imaging of operation 220 may include collecting a 'stack' of images. For example, a first generally 2D image may be collected at a first position, then a second 2D image collected at a second position, and so forth. In some embodiments, each image is offset from the others along a direction that is generally normal to the plane of the image. Other geometries for the image stack may be used in other example embodiments. In some embodiments, multiple images may be collected for each portion of the stack. For example, the method 200 may include capturing an image using brightfield illumination, then capturing a fluorescent image using filters for a first fluorescent dye, then capturing a fluorescent image for a second fluorescent dye, and so forth. In some embodiments, multiple stacks may be captured and mosaiced together to form an overall 3D image of a region of the sample 202.

[0061] The method 200 may include capturing coordinate information for the image(s). For example, the 3D imaging system 222 may have a coordinate system that relates each image to a location of the stage. This location information may relate the coordinates of the 3D image to the sample holder.

[0062] Operation 230 includes generating a 3D image such as 3D pathology dataset 232.The 3D pathology dataset 232 may be generated by a computing system (e.g., 130 of Figure 1) based on the images captured in operation 220. The 3D pathology dataset 232 represents a 3D image of all or a portion of the sample 202. In some embodiments, operation 230 includes combining multiple images in a stack to form a simulated 3D image16 4898-7105-8594 1(e.g., through mosaicking). In some embodiments, operation 230 includes presenting the images of the stack to represent their relationship to each other in three-dimensional space. In some embodiments, the 3D pathology dataset 232 may mimic the appearance of a stained pathology sample. For example, operation 230 may include applying a false color scheme to the 3D images, such as to mimic common stains such as H&E. The method 200 may include presenting the 3D pathology dataset 232 to a user, for example using display 138 of Figure 1. In some embodiments, the 3D pathology dataset 232 may show a region of the sample 202 that extends up to a few mm into the sample 202. For example, the 3D pathology dataset 232 may be around 5mm deep in some embodiments.

[0063] Operation 240 describes annotating one or more VOIs 242 in the 3D pathology dataset 232. The annotation may be performed automatically, manually, or combinations thereof. In an example automatic annotation process, a segmentation algorithm may be used to draw a border around regions of interest. How those regions of interest and borders are defined may depend on the application the 3DPM method 200 is being used for. In an example manual annotation process, the 3D pathology dataset 232 is presented to a user (e.g., a clinician) on a display (e.g., 138 of Figure 1). The user operates an input system (e.g., 134 of Figure 1) such as a mouse, touchscreen, stylus, or other input, to draw borders around one or more regions of interest. In some embodiments, a mixture of manual and automatic processes may be used. For example, an automatic process may suggest borders and a user may then manually adjust those borders.

[0064] Operation 250 includes generating cutting path instructions for the one or more VOIs 242 identified in operation 240. The cutting path instructions may depend on the operation and geometry of the tissue extraction system 262. In some embodiments, generating the cutting path instructions may include an iterative process for each volume of interest. For example, the cutting path instructions may have an optimization step that iteratively optimizes the cutting instructions to maximize an amount of VOI extracted while minimizing damage to surrounding areas of the sample 202. An example of cutting path instruction generation is described in more detail in Figures 5-7.

[0065] Operation 260 includes performing microdissection with the tissue extraction system 262. The operation 260 includes placing the sample holder 204 and sample 202 on a stage 264 (e.g., 126 of Figure 1) of the tissue extraction system 262. In the example geometry of Figure 2, the sample holder 204 is placed in an upright position on the stage 264. The coordinate systems of the 3D imaging system and the tissue extraction system 262 are registered to each other. For example, the coordinate systems may be registered using17 4898-7105-8594 1the process described with respect to the instructions 152 of Figure 1. The co-registration of the coordinate systems allows the tissue extraction system to relate the cutting path instructions generated in operation 250 with respect to the coordinate system of the image to be translated into coordinates for the operation of the tissue extraction system.

[0066] Operation 260 may include operating the tissue extraction system 262 to remove the one or more VOIs from the sample 202. Operation 270 is an example work flow that may be repeated for each VOI that is going to be removed in order to perform operation 260. For example, operation 270 shows milling out an area above the VOI, such as around a contour of the VOI when viewed from the direction of the cutting tool. One or more angled cuts are then performed to separate the VOI from the surrounding material. Once separated from the material, the VOI may then be removed from the sample 202 and the operation 270 repeated for other identified VOIs. For example in some embodiments, the tissue extraction system 262 may have a minimum extracted volume of about 1mm x 1mm and about 0.5mm thickness.

[0067] After removing the one or more VOIs, the operation 270 may include performing one or more analysis operations on the extracted VOIs. For example, the method 200 may include performing an omics analysis on the extracted VOIs. In some embodiments, the method 200 may include generating a map relating information gained from the analysis of the VOIs to their original location within the 3D pathology dataset 232.

[0068] Figure 3A is an isometric schematic drawing of a 3D imaging system stage according to some embodiments of the present disclosure. Figure 3B is an isometric schematic drawing of a tissue extraction stage according to some embodiments of the present disclosure. Figure 3A shows a portion of a 3D imaging system 310, in particular the stage 312 of the 3D imaging system 310. The 3D imaging system 310 may implement a portion of the 3D imaging system 110 of Figure 1 and / or 222 of Figure 2. The stage 312 may implement the stage 117 of Figure 1 and / or 224 of Figure 2. Figure 3B shows a portion of a tissue extraction system 350, in particular the stage 352 of the tissue extraction system 350. The tissue extraction system may implement a portion of the tissue extraction system 120 of Figure 1 and / or 262 of Figure 2 and the stage 352 may implement the stage 126 of Figure 1 and / or 264 of Figure 2.

[0069] Figures 3A and 3B both show a sample holder 302 (e.g., 102 of Figure 1 and / or 204 of Figure 2) with a mounted sample 304 (e.g., 104 of Figure 1 and / or 202 of Figure 2). The sample holder 302 is configured to be positioned in either the stage 312 or the stage 352. Figure 3A shows the sample holder 302 being positioned on the 3D imaging system 310 (e.g.,18 4898-7105-8594 1as part of operation 220 of Figure 2). Figure 3B shows the same sample holder 302 being positioned on the tissue extraction system 350 (e.g., as part of operation 260 of Figure 2). Figures 3A and 3B shows a particular geometry of how the sample holder 302 is positioned on the stages 312 and 352, however other geometries may be used in other example embodiments.

[0070] In the example of Figures 3A and 3B, the tissue holder 302 is shaped like a capital letter 'I', with a long middle section 306 and two ends 308 that are wider than the middle section 306. In some embodiments, optional alignment holes are drilled through the ends 308.

[0071] The stage 312 of the 3D imaging system 310 supports an imaging chamber 314. The imaging chamber 314 includes an immersion fluid chamber 316 that can be at least partially filled with an immersion fluid. The imaging chamber 314 has partially recessed sections 318. The ends 308 of the sample holder 302 rest in the recessed sections 318 with the middle section 306 going across a width of the top of the immersion fluid chamber 316. When the sample holder 302 is positioned in the imaging chamber 314 the sample 304 may hang below the sample holder and into the immersion fluid in the immersion fluid chamber 316.

[0072] The stage 352 of the tissue extraction system 350 includes an adapter 354. The adapter 346 may be mounted on the stage 352, for example using screws or other suitable mounting methods. The adapter 354 has a recessed portion 356 that the sample holder 302 fits into. For example, the recessed portion 356 may generally have a shape that matches the perimeter of the sample holder 302. In some embodiments, the recessed portion 356 may have extra cutouts, for example to allow fingers to access the sides of the sample holder 302 in order to more easily place or remove the sample holder 302 in the adapter 354.

[0073] In some embodiments, additional alignment guides may be used. For example either the imaging chamber 314, the adapter 354 or both may have pins that mate with corresponding alignment holes in the ends 308 of the sample holder.

[0074] Figure 4 is an image of an example registration mark on a sample holder according to some embodiments of the present disclosure. The image 400 represents an image captured by a 3D imaging system (e.g., 110 of Figure 1, 222 of Figure 2, and / or 310 of Figure 3A) of a registration mark 402 inscribed on a sample holder (e.g., 102 of Figure 1, 204 of Figure 2, and / or 302 of Figure 3) by a tissue extraction system (e.g., 120 of Figure 1, 262 of Figure 2, and / or 350 of Figure 3). For example, the registration mark 402 may be inscribed19 4898-7105-8594 1by the tissue extraction system as part of a co-registration process such as described by instructions 152 of Figure 1.

[0075] Before the image 400 is captured, the registration mark 402 is inscribed on the sample holder. For example, the tissue extraction system may inscribe the mark 402 on the sample holder. The tissue extraction system records the coordinates of the registration mark in the coordinate system of the tissue extraction system. In the embodiment of Figure 4, the registration mark is a + sign, with a vertical arm aligned along the +y and -y directions, and a horizontal arm aligned along the +x and -x directions. Other shapes may be used for the registration mark in other embodiments.

[0076] The image 400 is captured by the 3D imaging system. The 3D imaging system records the coordinates of the registration mark 402 in the coordinate system of the 3D imaging system. Since the registration mark 402 is a fixed position on the sample holder, the two coordinate systems now can be related and co-registered.

[0077] Figures 5-7 show an example workflow for developing cutting instructions for one or more VOIs. Figure 5 shows an example workflow with diagrams demonstrating the geometry of optimization parameters for the cutting instructions. Figure 6 shows a flow chart of a method of optimizing the cutting parameters for each VOL Figure 7 shows a flow chart for how the optimized cutting parameters are combined into cutting instructions usable by the tissue extraction system.

[0078] Figures 5-7 are all explained with respect to a certain method of extracting the VOIs from the sample, that involves a particular set of cutting parameters to be optimized. However, other methods of extracting the VOIs and / or other cutting parameters may be used in other example embodiments. Similarly, Figures 5-7 are generally described with respect to certain goals for how the cutting parameters should be optimized. Other goals, and other values for those goals may be used in other example embodiments.

[0079] Figure 5 is a flow chart of a visualization of cutting instruction generation and optimization according to some embodiments of the present disclosure. Figure 5 shows a flow chart 500 with a set of diagrams that visually represent a work flow for generating cutting instructions for each identified VOI in a sample. The flow chart 500 may represent the operation of 3DPM system such as 100 of Figure 1. For example the method 500 may represent operation such as operation 250 of Figure 2.

[0080] The flow chart 500 begins with a cross sectional diagram 510 showing an imaged region in a sample. For example, the cross-sectional diagram 510 may represent a cross section through a 3D pathology dataset such as 232 of Figure 2. Several VOIs 504 have been20 4898-7105-8594 1identified in the sample. In this example for ease of visualization, the VOIs are all shown extending through the same XZ plane represented by the diagram 510. However, it is not necessary for the VOIs to all be present in the same plane.

[0081] The method includes ordering the VOIs based on their distance to the surface of the sample. In this case there are five VOIs 504, and they are numbered in the diagram 510 to represent their order. After ordering the VOIs 504, the method 500 proceeds to cutting instruction generation for each of the VOIs. The cutting instruction generation may be an iterative process that optimizes one or more parameters in order to achieve a goal. An example set of goal includes high recall (e.g., most amount of VOI extracted), a high precision (e.g., only the VOI included in the extracted material), and low damage to surrounding VOIs. For example, the optimization goals may be set to maximize recall, achieve a precision of >85% and cause <15% damage to other VOIs. Other optimization goals, or other target values of those goals, may be used in other example embodiments.

[0082] In the example process of the method 500, the parameters to be optimized include zc, the depth of a contour plane, ei to e4, the locations of the edges of an inscribed quadrilateral within that contour plane, and angles 0i to 04, that define cutting planes through the edges ei to 64 respectively. More or fewer than four edges and cutting planes may be used in other example implementations.

[0083] Diagrams 520 to 550 represent the geometry of the cutting parameters for a single VOI 504. The cutting parameters shown in diagrams 520 to 550 are iteratively optimized for each VOI in sequence (e.g., in the order shown in diagram 510). An example optimization is described in more detail in Figure 6.

[0084] The diagram 520 is an isometric diagram showing the VOI 504 in the sample. A contour plane 514 is set at a depth zc. Once the distance zcis set, the edges ei to e4are drawn within the contour plane. For example, the edges ei to e4may represent the largest quadrilateral by area that can be drawn in the area of the VOI where the contour plane intersects it. The cutting path instructions include instructions for milling out the sample material down to a top of the VOI. Additional milling than removes material as close as possible to the edge of the VOI 504 down to the contour plane 514.

[0085] Diagrams 540 and 550 show an xz cross section and an zy cross section after the material above the VOI 504 has been milled out. Angles 0i to 0 are defined relative to the edges ei to 64 respectively. Each angle 0i to 04 represents an angle of a cutting plane to vertical that intersects the respective edge ei to 64. For example, the angle 0i represents a vertical angle of a first cutting plane that intersects the first edge ei and so forth.21 4898-7105-8594 1

[0086] Figure 6 is a flow chart of an example implementation of a method of cutting parameter optimization according to some embodiments of the present disclosure. The method 600 may be implemented a 3DPM system such as 100 of Figure 1 in some embodiments. For example, the method 600 may be implemented as instructions (e.g., 150 of Figure 1) performed by the computing system 130 of Figure 1. The method 600 may represent the operation 250 of Figure 2 and / or the method 500 of Figure 5 in some embodiments.

[0087] The method 600 generally begins with operation 610. Before operation 610 is performed, an operation 602 is shown that includes annotation of VOIs in a 3D image (e.g., operation 240 of Figure 2). The output of the operation 602 is a set of 3D coordinates for the boundaries of the identified VOIs. These 3D coordinates are co-registered to the image(s) used for annotation and to the coordinate system of a tissue extraction system.

[0088] Operation 610 includes sorting the VOIs into an order. The order is determined by distance of the VOI to the surface of the sample (e.g., the side of the sample opposite the side mounted to the sample holder). In some embodiments, the order may be determined based on the measured distance from the surface to a highest point of each VOI border. Once an order is determined, cutting parameters are optimized for each of the VOIs in order. For example, the cutting parameters for a VOI closest to the surface of the tissue are optimized first, then the cutting parameters for the VOI that is next closest to the surface of the tissue, and so forth. Box 615 shows a flow chart of operations that represent cutting path optimization. The method within box 615 is repeated for each of the VOIs in the ordered determined in operation 610.

[0089] Box 615 includes operations 620, 630, 640 and 650. Operation 620 describes initializing a cutting parameter optimization algorithm. In the example implementation of Figure 6, the cutting parameter optimization algorithm is a genetic algorithm. The initialization step included in operation 620 involves setting up candidate parameter sets with different values for the cutting parameters to be optimized. In the implementation shown in Figure 6, the cutting parameters to be optimized are the depth of a contour plane zcand the angles 04, 02, 03, and 04of the cutting planes that pass through the edges of a quadrilateral inscribed in the outline of the VOI where zcintersects it. The edges ei, e2, e3, and e4are determined by the value of zcand thus may be optimized as part of selecting an optimal value for zc. A generation counter Gen is also initialized at an initial value of 0.

[0090] The candidate sets are initialized with a range of possible values for zcand each of the angles 0, (i = 1 to 4). The range of initial values for zcis given by equation 1, below:22 4898-7105-8594 1zc~clip(U (zmedianzmedian izbottom’ztop')Eqn. 1

[0091] Where zmedian is the vertical midpoint of the VOI, A = 387pm, clip(x, a, b) denotes constraining x to the interval [a, b], and U(a, b) represents a continuous uniform probability distribution over the interval [a, b], Different values of A may be used in other example embodiments. The range of values for each angle 0, is generated by equations 2 and 3, below:di = 0 + EL, Ei~ U (-30°, 30°), 0° < 0L< 60°Eqn. 20 = tan’1f - )'^medianzbottom'Eqn. 3

[0092] Where daVg is the average lateral distance from the centroid of the fitted quadrilateral to its edges ei (i = 1 to 4) at the depth of the plane zmedian.

[0093] Operation 620 is generally followed by operation 630 that includes performing a simulated 3D dissection based on each of the candidate parameter sets and evaluating one or more metrics based on the simulated dissection. The candidate sets may be referred to as individuals. In an example implementation of the simulated dissection, for each individual, a simulated 3D microdissection is performed: in the first step, contour milling is simulated by removing all off-target material within a vertical cylinder centered on the 2D cross section of the VOI at plane zc, using the minimum enclosing circle plus a 0.4 mm radial buffer (total radius rt); in the second step, four angled cuts are applied. Each cutting edge is extended beyond its original lateral endpoints by at least 50% of its length on both sides, providing a buffer to ensure full detachment of the VOI from its base.

[0094] Several metrics are generated based on the simulated dissection for each individual.Candidates that fail to completely detach the VOI are assigned zero recall and precision. Successful candidates are further evaluated to assess the maximum damage they create to the remaining, unextracted VOIs. A fitness function is defined based on the recall of the extracted VOI on the condition that it meets a set of criterial for both precision and damage as shown in equation 4, below:23 4898-7105-8594 1(Recall, if Precision > 85% and Max Damage < 15% Fitness — ( „ ,( 0, otherwiseEqn. 4

[0095] Recall and precision are computed based on the volumetric overlap between the target VOI and the successfully detached tissue volume as given by equation 5, below:Recall = precision =voKyor) voi(ydetachecpEqn. 5

[0096] Where Vol() denotes a 3D volume, and Vdetached refers to the subset of the simulated dissection region that is successfully separated from the base. The maximum collateral damage Max Damage is the damage to the remaining VOIs and is defined by equation 6, below: / Vol(i'0 / jnvCMt)\Max Damage = maxjej \ voKyoh-) IEqn. 6

[0097] Where Vcut includes all of the material removed by the simulated milling and angled cuts, and J denotes the set of VOIs that have not yet undergone simulated dissection. It is also evaluated whether or not Vcut creates multiple disconnected components within each VOIj. If so, then it is assumed that only the largest component is preserved, and that all remaining fragments may be treated as damage.

[0098] Operation 630 is generally followed by operation 640 that includes determining the best parameter set based on the simulated dissections across the individuals. The best performing individual is selected based on the fitness given by equation 4. If the fitness of the best performing individual has a recall that exceeds 30%, then the optimization for the cutting parameters for this VOI is finished and the steps of box 615 are repeated for each remaining VOI (if any). Similarly, a generation counter is checked. If the generation counter has reached 5, then enough optimization steps have been performed and box 615 is repeated. If the generation counter has not reached 5, and the best performing cutting parameters do not have a sufficient recall, then the method of box 615 proceeds to operation 650.24 4898-7105-8594 1

[0099] Operation 650 includes reproduction and mutation to generate new parameter sets. The best performing individual is reproduced, and a new set of candidate parameters are generated based on the best performing individual, random change, or random variation applied to the best performing sets. Operation 650 also includes incrementing the generation counter Gen. Operation 650 is then followed by operation 630 to repeat the steps of the box 615.

[0100] If the method of box 615 has completed for each of the VOIs (e.g., each VOI has a set of optimized cutting parameters) then the method of box 615 completes and the method 600 has generated a set of optimized cutting parameters. The optimized cutting parameters are used to generate cutting path instructions 604 that the tissue extraction system can use.

[0101] Figure 7 is a flow chart of a method of generating cutting path instructions based, in part on optimized cutting parameters according to some embodiments of the present disclosure. The method 700 may, in some embodiments, be implemented by a 3DPM system such as 100 of Figure 1. For example, the method 700 may represent instructions (e.g., 150 of Figure 1) loaded on a computing system such as 130 of Figure 1.

[0102] The method 700 accepts various inputs 702, 704 and 706 and uses those inputs to generate cutting path instructions 708 that can be used by the tissue extraction system (e.g., 120 of Figure 1 and / or 262 of Figure 2). In particular, Figure 7 shows the annotated coordinates of the VOIs 704, such as generated from operation 240 of Figure 2, a 3D digital model of the sample holder 702 (e.g., sample holder 102 of Figure 1, 204 of Figure 2, and / or 302 of Figure 3), and the optimized cutting parameters 706 (e.g., from operation 250 of Figure 2 and / or the output 604 of Figure 6) as inputs. Figure 7 shows a first inset 760 and a second inset 740 that show diagrams representing certain substeps of the method 700 as described in more detail herein.

[0103] The method 700 generally begins with operation 710 that includes generating mesh bodies for the VOIs. These mesh bodies are generated based on the coordinates of the borders of the VOIs generated by the annotation 704. Operation 710 is generally followed by operation 720 that includes generating computer assisted design (CAD) models of the VOIs. The CAD models may be generated from the mesh bodies of the VOIs generated in operation 710. Operation 720 is generally followed by operation 730 that includes generating CAD models for dissection. Operation 730 takes the tissue holder model 702 and the optimized cutting parameters 706 as inputs.25 4898-7105-8594 1

[0104] Inset 760 shows some of the steps performed as part of operation 730. The inset 760 shows a first CAD model 762 that represents a full bounding box model extending from the base to the top of the sample's surface. A second CAD model 764 is generated from the first CAD model by adding a vertical cylindrical cut (radius = rt) from the surface of the bounding box to the top plane of the VOI (e.g., a plane tangent to the highest point on the boundary). The inset 760 also shows a third CAD model 766 that is the first CAD model 762 with a cylindrical cut (radius = rt) down to the level of the plane zcfrom the optimized parameters 706. The third model 766 also includes a model of the portion of the VOI that is within the cylinder and extends above the plane zc. The inset 760 also shows a fourth CAD model 768 that shows a set of four bodies, each defined by subtracting a tilted cutting-plane box from the first CAD model 762 at an optimized angle 0 / from the optimized parameters 706. The thickness of each tilted box matches a diameter of the cutting element (e.g., 122 of Figure 1). In this example, a 200pm endmill was used for all 3D microdissections and so 200pm thickness is used.

[0105] The work coordinate system (WCS) origin is set to the registration mark or a coregistered 0 coordinate origin based therefrom. The four CAD models 762-768 are used for generating the cutting path instructions. Inset 740 shows the process of using the CAD models 762-768 to generate the cutting path instructions. Box 742 shows using the second CAD model 764 to define a 2D Pocket path to remove material above the VOI. Box 744 shows using the third CAD model 766 to generate a 3D Flow operation to engrave the portion of the VOI above plane zc. Box 746 shows applying several 2D contour paths to the cross-section of the VOI at the plane zc, with radial stock-to-leave offsets. The stock-to- leave offsets may vary from 0mm to 0.3mm in some embodiments. Box 746 sows using the fourth CAD model 768 to define the angled cuts for VOI detachment. For example, 2D Slot operations are applied. To minimize tissue deformation, the four angled cuts are performed sequentially and cyclically, where each cut removes a small portion of the tissue. The first round of cuts reaches near the zcplane. Subsequent paths incrementally deepen the cuts by some amount (e.g., 0.1 mm) one edge at a time (cyclically) until the bottom of all four tilted planes are reached.

[0106] The final cutting instructions generated by the boxes of inset 740 may be postprocessed and output as cutting path instructions usable by the tissue extraction system. For example, the cutting path instructions 708 may be output as a machine-specific G-code.

[0107] Figure 8 is flow chart of a method of using a 3DPM system according to some embodiments of the present disclosure. The method 800 of Figure 8 may be implemented26 4898-7105-8594 1by one or more of the systems, apparatuses, or portions thereof discussed herein. For example, the method 800 may be performed by the 3DPM system 100 of Figure 1. The method 800 may, in some embodiments, include all or a portion of the method 200 of Figure 2, the method 500 of Figure 5, 600 of Figure 6, and / or 700 of Figure 7.

[0108] The method 800 may generally begin with operation 810 that includes imaging a sample with a 3D imaging system (e.g., 100 of Figure 1, 222 of Figure 2, and / or 310 of Figure 3). The imaging operation 810 may be generally similar to the instructions 154 of Figure 1 or the operation 220 of Figure 2 in some embodiments. In some embodiments, operation 810 may include collecting a plurality of images in a 'stack' and forming the plurality of images into a 3D image. In some embodiments, operation 810 may include applying a false color to the 3D image. In some embodiments, operation 810 may include generating a 3D pathology dataset (e.g., 232 of Figure 2) as the 3D image.

[0109] Operation 810 is generally followed by operation 820 that includes identifying 3D coordinates of a VOI (e.g., 242 of Figure 2 and / or 504 of Figure 5) with a computing system (e.g., 130 of Figure 1). Operation 820 may be generally similar to operation 240 of Figure 2 in some embodiments. In some embodiments operation 820 may include annotating one or more VOIs in the 3D image. The annotation may be manual, automatic, or combinations thereof. The 3D coordinates of the VOI may be in a coordinate system of the 3D imaging system.

[0110] Operation 820 may generally be followed by operation 830 that includes translating the 3D coordinates of the VOI from a coordinate system of the 3D imaging system to a coordinate system of a tissue extraction system (e.g., 120 of Figure 1, 262 of Figure 2, and / or 350 of Figure 3) with the computing system. The method 800 may include registering the coordinate system of the 3D imaging system to the coordinate system of the tissue extraction system. The method 800 may include making a registration mark (e.g., 402 of Figure 4) on the sample holder with the tissue extraction system, imaging the registration mark on the sample holder with the 3D imaging system, and registering the coordinate system of the 3D imaging system to the coordinate system of a tissue extraction system based on a registration mark. In some embodiments, the registering may be performed before the imaging of operation 810 is performed.

[0111] Operation 830 may generally be followed by operation 840 that includes generating cutting path instructions for the VOI based on the translated 3D coordinates of the VOI with the computing system. The operation 840 may be generally similar to the operation 250 of Figure 2 and may include steps such as the methods 500 of Figure 5, 600 of Figure 6, and / or27 4898-7105-8594 1700 of Figure 7 in some embodiments. Operation 840 may include generating cutting path instructions with the computing system based on the translated 3D coordinates in some embodiments.

[0112] Operation 840 may include generating the cutting path instructions by optimizing one or more cutting parameters based on one or more metrics using an iterative optimization method. The one or more cutting parameters may include an initial milling step, angles of 4-quadrant detachment cuts or combinations thereof in some embodiments. The one or more metrics include recall, precision, amount of damage to surrounding tissue, or combinations thereof in some embodiments. The iterative optimization method is a genetic algorithm, swarm-based algorithms, reinforcement learning, or combinations thereof in some embodiments.

[0113] Operation 840 may generally be followed by operation 850 that includes cutting at least a portion of the VOI from the sample with a tissue extraction system using the cutting path instructions. Operation 850 may be generally similar to operation 260 of Figure 2 and / or operation 270 of Figure 2 in some embodiments. Operation 850 may include operating the tissue extraction system with the computing system based on the cutting path instructions in some embodiments.

[0114] In some embodiments, the method 800 includes mounting the sample to a sample holder (e.g., 102 of Figure 1, 204 of Figure 2, and / or 302 of Figure 3). The method 800 may include positioning the sample holder in a stage (e.g., 117 of Figure 1, 224 of Figure 2, and / or 312 of Figure 3) of the 3D imaging system for the imaging and positioning the sample holder in a stage (e.g., 126 of Figure 1, 264 of Figure 2, and / or 352 of Figure 3) of the tissue extraction system for the cutting. The method 800 may include positioning the sample holder in the stage of the 3D imaging system in an inverted position and positioning the sample holder in the stage of the tissue extraction system in an upright position. The metho d800 may include mounting the sample to the sample holder by gluing the sample to the sample holder.

[0115] In some embodiments, the method 800 may include identifying respective 3D coordinates of a plurality of VOIs in the sample based on the 3D image, generating cutting path instructions with the computing system for each of the plurality of VOIs by sequentially optimizing the cutting instructions for each of the plurality of VOIs based on the respective 3D coordinates, and cutting a plurality of volumes from the sample based on the cutting instructions with the tissue extraction system, wherein each of the plurality of volumes is associated with one of the plurality of VOIs. The method 800 may include performing an28 4898-7105-8594 1omics analysis on each of the plurality of volumes and relating omics information associated with each of the plurality of volumes to a 3D relationship between the plurality of VOIs in the sample.EXAMPLE

[0116] An example application that uses a 3DPM system such as 100 of Figure 1 is discussed. The example application involves taking a tissue biopsy that includes a tumor (or portion thereof) and using the 3DPM system, for example with the method 200 of Figure 2 to identify VOIs of interest in the tumor. Phylogenetic analysis is performed on the extracted VOIs to help determine tumor progression.

[0117] In particular, the 3DPM was used to study a prostate cancer case involving the bladder with a highly branched architecture. A biopsy was collected and optically cleared and stained. The tissue sample was stained with fluorescent dyes TO-PRO3 and Eosin Y and attached to the sample holder (e.g., 102 of Figure 1, 204 of Figure 2, and / or 302 of Figure 3) as part of an operation such as 210 of Figure 2. The sample holder was inverted and positioned within a matching slot (e.g., 318 of Figure 3) on an imaging chamber (e.g., 314 of Figure 3), with ethyl cinnamate (Eci) used as the index-matching medium (refractive index = 1.56) in the chamber (e.g., 316 of Figure 3). TO-PRO-3 fluorescence was excited at 638 nm and collected using a long-pass emission filter, while Eosin fluorescence was excited at 561 nm and collected using a band-pass emission filter on an OTLS microscope (e.g., 110 of Figure 1, 222 of Figure 2, and / or 310 of Figure 3) as part of 3D imaging such as operation 220 of Figure 2. Volumetric images collected by the OTLS were downsampled to to 2.9 pm / voxeL Image tile stitching and fusion were performed. A previously established H&E false-coloring algorithm, based on the Beer-Lambert law of absorption, was applied to the two-channel fluorescence images to simulate the appearance of standard H&E images to generate a dataset as in operation 230 of Figure 2. 3D annotations (e.g., as part of operation 230 of Figure 2) were generated by manually labeling image sections at 50-pm intervals on the false-colored 2D image stack, followed by 3D upsampling to generate continuous annotated volumes. 3D skeletonization of the tumor mask was performed.

[0118] Tumor segmentation was manually performed under the guidance of an expert clinician such as a board-certified anatomic pathologist subspecialized in GU Pathology as part of an operation such as 240 of Figure 2. A 3D skeletonization approach was used to extract the main structural trajectory of the tumor. To investigate the clonal relationships29 4898-7105-8594 1between distinct spatial branches within the tumor, seven VOIs were selected for microdissection and whole-genome sequencing.

[0119] Prior to co-registration (e.g., instruction 152 of Figure 1), the "zero working coordinate" of the CNC was manually set to one corner of the tissue holder. Co-registration was performed by engraving a target pattern on the tissue holder using a 500-pm diameter endmill as the cutting tool (e.g., 122 of Figure 1). After engraving, the holder was imaged in ECi, with fluorescein dye added, to visualize the engraved pattern. The acquired OTLS image was then aligned to the G-code toolpath, allowing for the derivation of a formula to transform the OTLS stage coordinates to the CNC working coordinates.

[0120] Machining parameters— including spindle speed, cutting feed rate, and plunge feed rate— were experimentally optimized to minimize tissue deformation with the final values being 24,000 rpm, 50 mm / min, and 30 mm / min, respectively. Final toolpaths (e.g., 604 of Figure 6) are generated from operations such as 250 of Figure 2 and / or 600 of Figure 6, and are post-processed and exported as machine-specific G-code. This two-step workflow— CAD generation followed by CAM programming— is potentially useful compared to directly generating G-code. First, toolpath generation for 5-axis CNC milling is relatively complex and may generally use professional CAM software to handle variable tool orientation, cutting logic, collision avoidance, and safety constraints. Second, G-code is a low-level machine language that depends heavily on the specific hardware and controller, making it difficult to generalize or maintain across systems.

[0121] Once optimized, the G-code files are executed on the 5-axis CNC system to perform microdissection (e.g., operation 260 of Figure 2). At the end of each operation, the endmill is cleaned with a brush to remove residual tissue debris, ensuring precise cutting.Microdissected tissues are manually collected upon visual confirmation of successful detachment of the VOIs from their tissue base. To prevent cross-VOI contamination, the endmill is rinsed with 100% ethanol between VOIs.

[0122] The 3D microdissected tissues extracted from the VOIs were washed sequentially in 100% and 70% ethanol, each for one day. The tissues were then rehydrated and washed in phosphate-buffered saline (PBS). DNA extraction was performed. After genomic DNA extraction, library preparation and whole-genome sequencing were performed. The genome sequences were analyzed for single nucleotide variants (SNVs) and copy number alterations (CNA).

[0123] Figure 9 is a figure showing an image of phylogenetic information determined from using a 3DPM on a tumor biopsy according to some embodiments of the present disclosure.30 4898-7105-8594 1The figure 900 was processed according to the steps described above with respect to the example application.

[0124] The figure 900 has several different elements. An image 910 is shown representing a view of the sample (e.g., biopsy) with labels for the different VOIs. A phylogenetic tree is shown as part of the image 910. The phylogenetic tree was reconstructed from somatic SNVs determined from processing the extracted VOIs, adjusting for local copy-number states. These sites encompassed the skeleton root near the site where the tumor enters the bladder, major branching nodes, and terminal leaf nodes forming multiple branches that extend in divergent 3D directions. Additionally, a region was demarcated with distinct pathological features (VOI 16; Fig. 2b) located near a highly branching node.

[0125] The figure 900 includes a table 920 that summarizes different properties determined from the phylogenetic analysis of the extracted VOIs.

[0126] The figure 900 also includes a 3D spatial tumor phylogenetic tree 930 constructed based on inferring possible dissemination routes of subclones between VOIs based on their cancer cell fractions (CCFs, i.e., the proportion of cancer cells within a sample that harbors a given set of mutations), and overlaid these routes on a 3D tumor skeleton generated from the 3D pathology dataset. The branch lengths reflect physical distances between regions in the tissue. In the 3D spatial tumor phylogenetic tree rings are shown representing the phylogenetic analysis at the different VOIs with the center ring corresponding to the ancestral clone and each outward ring representing a subsequent clone in the phylogenetic tree. The thickness (3D tree) or the angular span (2D plot) of each colored band represents the cancer cell fraction for that clone (360°: 1).

[0127] The 3D spatial phylogeny analysis reveals a most likely and spatially coherent dissemination path of the tumor through the tissue. In addition, comparison of the spatial branching and phylogenetic trajectories reveals diverse relationships: in this case, a single site acts as a node in both trees, whereas other nodes show divergence only in 3D spatial trajectories or phylogeny. These results indicate that spatial branching and clonal emergence are not always coupled, highlighting the spatial and evolutionary complexity of tumors and the necessity of integrating spatial and phylogenetic analyses for a full understanding of tumor evolution.

[0128] Of course, it is to be appreciated that any one of the examples, embodiments or processes described herein may be combined with one or more other examples,31 4898-7105-8594 1embodiments and / or processes or be separated and / or performed amongst separate devices or device portions in accordance with the present systems, devices and methods.

[0129] Finally, the above-discussion is intended to be merely illustrative of the present system and should not be construed as limiting the appended claims to any particular embodiment or group of embodiments. Thus, while the present system has been described in particular detail with reference to exemplary embodiments, it should also be appreciated that numerous modifications and alternative embodiments may be devised by those having ordinary skill in the art without departing from the broader and intended spirit and scope of the present system as set forth in the claims that follow. Accordingly, the specification and drawings are to be regarded in an illustrative manner and are not intended to limit the scope of the appended claims.32 4898-7105-8594 1

Claims

ClaimsWhat is claimed is:

1. A system comprising:a 3D imaging system configured to generate a 3D image of at least a portion of a sample; a computing system comprising a processor and a memory encoded with non-transitory instructions which, when executed by the processor cause the system to:generate 3D coordinates of a boundary of a volume of interest (VOI) identified in the 3D image of the at least a portion of the sample;translate the 3D coordinates of the VOI from an imaging coordinate system of the 3D image to an extraction coordinate system; andgenerate cutting path instructions in the extraction coordinate system for the VOI based on the registered 3D coordinates of the VOI; anda tissue extraction system configured to extract at least a portion of the VOI from the sample using the cutting path instructions.

2. The system of claim 1, further comprising:a sample holder comprising at least one registration mark, wherein the sample holder is configured to hold the sample, wherein coordinates of the at least one registration mark are coregistered to both the imaging coordinate system and the cutting coordinate system, and wherein the computing system performs the translation of the 3D coordinates of the VOI based, at least in part, on the co-registered coordinates of the at least one registration mark.

3. The system of claim 2,wherein the 3D imaging system includes an imaging stage with a recess configured to hold at least a portion of the sample holder, andwherein the tissue extraction system includes an extraction stage with a recess configured to hold at least a portion of the sample holder.33 4898-7105-8594 14. The system of claim 3, wherein the sample holder is configured to rest in the recess of the imaging stage in an inverted position and wherein the sample holder is configured to rest in the recess of the extraction stage in an upright position.

5. The system of claim 2, wherein instructions when executed by the processor cause the system to:form the registration mark on the sample holder with the tissue extraction system with first coordinates in the extraction coordinate system;image the registration mark with the 3D imaging system to record second coordinates of the registration mark in the imaging coordinate system; andco-register the imaging coordinate system to the extraction coordinate system based on the first coordinates and the second coordinates of the registration mark.

6. The system of claim 1, wherein the 3D imaging system includes an open-top light sheet microscope.

7. The system of claim 1, wherein the 3D imaging system includes a two-photon microscope, a confocal microscope, a light-field microscope, a wide field deconvolution system, an optical coherence tomography system, a reflectance OCT system, a reflectance confocal system, a nonlinear microscopy system, a phase-contrast imaging system, a holotomography system, a microCT imaging system, other forms of non-contact 3D imaging, or combinations thereof.

8. The system of claim 1, wherein the tissue extraction system includes a 5-axis computer numerical control (CNC) mill, one or more robotic arms, or combinations thereof.

9. The system of claim 1, wherein the instructions, when executed by the processor, cause the computing system to collect a sequence of images from the 3D imaging system in a stack and generate the 3D image from the stack.34 4898-7105-8594 110. The system of claim 1, wherein the 3D image includes a fluorescent image, a reflectance optical coherence tomography image, a reflectance confocal image, a non-linear microscopy image, a phase-contrast image, holotomography, a microCT image, or combinations thereof.

11. The system of claim 1, wherein the instructions, when executed by the processor, cause the system to:generate and optimize a set of cutting parameters for the VOI based on the 3D coordinates of the boundary of the VOI; andgenerate the cutting path instructions based on the optimized set of cutting parameters.

12. The system of claim 11, wherein the instructions, when executed by the processor, cause the system to:generate a plurality of sets of cutting parameters;perform a simulated tissue extraction based on each of the plurality of sets of cutting parameters;select a best one of the plurality of sets of cutting parameters based on the simulated tissue extraction;determine if the best one of the plurality of sets of cutting parameters meets one or more goals and (1) if the best one of the plurality of sets of cutting parameters does not meet the one or more goals generate a new plurality of sets of cutting parameters, based at least in part on the best one of the plurality of sets of cutting parameters and iteratively repeat the simulated tissue extraction with each of the new plurality of sets of cutting parameters, or (2) if the best one of the plurality of sets of cutting parameters does meet the one or more goals provide the best one as the optimized set of cutting parameters.

13. The system of claim 11, wherein the instructions, when executed by the processor, cause the system to generate and optimize the set of cutting parameters for the VOI with a genetic algorithm.35 4898-7105-8594 114. A method comprising:imaging a sample with a 3D imaging system to generate a 3D image of the sample; identifying 3D coordinates of a volume of interest (VOI) based on the 3D image with a computing system;translating the 3D coordinates of the VOI from a coordinate system of the 3D imaging system to a coordinate system of a tissue extraction system with the computing system;generating cutting path instructions for the VOI based on the translated 3D coordinates of the VOI with the computing system; andcutting at least a portion of the VOI from the sample with a tissue extraction system using the cutting path instructions.

15. The method of claim 14, further comprising:mounting the sample to a sample holder¬ positioning the sample holder in a stage of the 3D imaging system for the imaging; and positioning the sample holder in a stage of the tissue extraction system for the cutting.

16. The method of claim 15, further comprising:positioning the sample holder in the stage of the 3D imaging system in an inverted position; andpositioning the sample holder in the stage of the tissue extraction system in an upright position.

17. The method of claim 15, further comprising mounting the sample to the sample holder by gluing the sample to the sample holder.

18. The method of claim 14, further comprising:36 4898-7105-8594 1making a registration mark on the sample holder with the tissue extraction system; imaging the registration mark on the sample holder with the 3D imaging system; and registering the coordinate system of the 3D imaging system to the coordinate system of a tissue extraction system based on a registration mark.

19. The method of claim 14, further comprising operating the tissue extraction system with the computing system based on the cutting path instructions.

20. The method of claim 14, further comprising generating the cutting path instructions by optimizing one or more cutting parameters based on one or more metrics using an iterative optimization method.

21. The method of claim 20, wherein the one or more parameters include a depth of a plane, angles of one or more cuts that pass through an edge inscribed in the plane, or combinations thereof, and wherein the one or more metrics include recall, precision, amount of damage to surrounding tissue, or combinations thereof.

22. The method of claim 20, wherein the iterative optimization method is a genetic algorithm, swarm-based algorithms, reinforcement learning, or combinations thereof.

23. The method of claim 14, further comprising:identifying respective 3D coordinates of a plurality of VOIs in the sample based on the 3D image;generating cutting path instructions with the computing system for the plurality of VOIs by sequentially optimizing a set of cutting parameters for each of the plurality of VOIs based on the respective 3D coordinates; andcutting a plurality of volumes from the sample based on the cutting instructions with the tissue extraction system, wherein each of the plurality of volumes is associated with one of the plurality of VOIs.37 4898-7105-8594424. The method of claim 23, further comprising:performing an omics analysis on each of the plurality of volumes; andrelating omics information associated with each of the plurality of volumes to a 3D relationship between the plurality of VOIs in the sample.38 4898-7105-85941

Images