WO2025230724A1

WO2025230724A1 - System and method for oct-based tissue screening

Info

Publication number: WO2025230724A1
Application number: PCT/US2025/024859
Authority: WO
Inventors: Shaohua Pi; Yuanyuan Chen; Jose Alain Sahel
Original assignee: University of Pittsburgh
Current assignee: University of Pittsburgh
Priority date: 2024-05-01
Filing date: 2025-04-16
Publication date: 2025-11-06
Anticipated expiration: 2026-11-01

Abstract

An OCT-based tissue screening system includes a sample arm for delivering a light signal for imaging, a camera positioned adjacent the sample arm, a motorized platform for enabling 3-D manipulation of a relative position between the sample arm and the tissue sample, and a processing apparatus including an object detection component configured to determine a location of the tissue sample based on an image of the tissue sample captured by the camera, a motor control component configured to operate the motorized platform, an imaging depth optimization component configured to determine an optimal imaging depth for the system based on a plurality of OCT intensities obtained at a plurality of depths, and a segmentation component configured to delineate a number of tissues regions within each of the number of B-scans at the optimal imaging depth.

Description

SYSTEM AND METHOD FOR OCT-BASED TISSUE SCREENING

CROSS REFERENCE TO RELATED APPLICATIONS:

[0001] This application claims priority to U.S. Provisional Patent Application Serial No. 63/640,931, filed on May 1, 2024, and titled ‘‘System and Method for OCT-Based Tissue Screening,"’ the disclosure of which is incorporated herein by reference.

STATEMENT OF GOVERNMENT INTEREST

[0002] This invention was made with government support under grant #EY030991 awarded by the National Institutes of Health (NIH). The government has certain rights in the invention.

FIELD OF THE INVENTION

[0003] The disclosed concept relates generally to tissue screening systems, and, in particular, to a fully automated OCT-based tissue screening system for research using ex vivo tissue cultures, especially for applications requiring high-throughput screening of therapeutic agents, potential therapies, and optimal protocols.

BACKGROUND OF THE INVENTION

[0004] Screening systems, particularly in the context of drug discovery, play a pivotal role in identifying potential therapeutic compounds from a large number of chemicals, significantly accelerating the discovery from hit identification to clinical trials. By offering a unique balance between the simplicity of cell-based assays and the complexity of in vivo animal models, tissue culture is emerging as an advantageous tool for screening to provide a more physiologically and practically relevant context in studying drug effects and disease mechanisms. However, the lack of non-invasive, quantitative ex vivo imaging tools has hindered the use of tissue cultures for high-throughput assessment of the biological activity, efficacy, and safety of substances.

[0005] Optical coherence tomography (OCT) has long been a cornerstone in ophthalmological diagnostics due to its non-invasive, high-resolution, and 3-D imaging capabilities. Unlike traditional histology, which is time-consuming, structurally disruptive, and often necessitates multiple samples, OCT offers a rapid, longitudinal, and unbiased approach to quantitatively characterize tissue responses to variables such as pharmacological agents. Despite its clinical prominence, however, the role of OCT in ex vivo applications has been understudied, largely due to the lack of dedicated customization and optimization for static and non-perfused living tissues.

SUMMARY OF THE INVENTION:

[0006] In one embodiment, the disclosed concept provides an OCT-based tissue screening system that includes a sample arm structured and configured to deliver a light signal for OCT imaging of a tissue sample, a camera positioned adjacent to the sample arm, a motorized platform structured and configured to enable 3-D manipulation of a relative position between the sample arm and the tissue sample, and a processing apparatus including an object detection component configured to determine a location of the tissue sample based on an image of the tissue sample captured by the camera, a motor control component configured to operate the motorized platform to cause the sample arm to be aligned with the tissue sample based on the determined location of the tissue sample such that the light signal will be directed into the tissue sample by the sample arm, an imaging depth optimization component configured to determine an optimal imaging depth for the system when the sample arm is aligned with the tissue sample based on a plurality of OCT intensities obtained using the sample arm at a plurality' of depths when the sample arm is aligned with the tissue sample, and a segmentation component configured to (i) receive a B-scan volume of the tissue sample captured at the optimal imaging depth when the sample arm is aligned with the tissue sample, wherein the B-scan volume includes a number of B-scans, and (ii) delineate a number of tissues regions within each of the number of B-scans.

[0007] In another embodiment, the disclosed concept provides an OCT-based tissue screening method that includes determining a location of a tissue sample based on an image of the tissue sample captured by a camera coupled to an OCT sample arm structured and configured to deliver a light signal for OCT imaging of the tissue sample, aligning the sample arm with the tissue sample based on the determined location of the tissue sample such that the light signal will be directed into the tissue sample by the sample arm. determining an optimal imaging depth when the sample arm is aligned with the tissue sample based on a plurality of OCT intensities obtained using the sample arm at a plurality of depths when the sample arm is aligned with the tissue sample, receiving a B-scan volume of the tissue sample captured at the optimal imaging depth when the sample arm is aligned with the tissue sample, wherein the B-scan volume includes a number of B-scans, and delineating a number of tissues regions within each of the number of B-scans. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] A full understanding of the invention can be gained from the following description of the preferred embodiments when read in conjunction with the accompanying drawings in which:

[0009] FIGS. 1 and 2 are schematic diagrams of an OCT-based tissue screening system according to an exemplary embodiment of the disclosed concept;

[0010] FIG. 3 is a schematic diagram of an operational pipeline of the fully automated OCT-based tissue screening system according to an exemplar}' embodiment of the disclosed concept;

[0011] FIG. 4 is a schematic diagram of a transformer-based segmentation method employing a hybrid architecture according to an exemplary embodiment of the disclosed concept;

[0012] FIG. 5 are exemplary en face images from performing the pipeline of the disclosed concept on mouse retinal explant cultures, including the structure and thickness heatmap; and

[0013] FIG. 6 is a schematic diagram of an exemplar} control system for implementing the displaced concept according to an exemplar}⁷ embodiment.

DETAILED DESCRIPTION OF THE INVENTION

[0014] As used herein, the singular form of “a”, “an”, and “the” include plural references unless the context clearly dictates otherwise.

[0015] As used herein, the statement that two or more parts or components are “coupled” shall mean that the parts are joined or operate together either directly or indirectly, i.e., through one or more intermediate parts or components, so long as a link occurs.

[0016] As used herein, “directly coupled” shall mean that two elements are directly in contact with each other.

[0017] As used herein, the term “number” shall mean one or an integer greater than one (i.e., a plurality).

[0018] As used herein, the terms “component” and “system” are intended to refer to a computer related entity, either hardware, a combination of hardware and software, software, or software in execution. For example, a component can be, but is not limited to being, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and/or a computer. By way of illustration, both an application running on a server and the server can be a component. One or more components can reside within a process and/or thread of execution, and a component can be localized on one computer and/or distributed between two or more computers.

[0019] As used herein, the term “residual neural network” or “residual network” shall mean a seminal deep learning model/ architecture in which the weight layers leam residual functions with reference to the layer inputs.

[0020] As used herein, the term “transformer neural network” or “transformer network” shall mean a deep learning model/architecture which is based on the multi-head attention mechanism.

[0021] Directional phrases used herein, such as, for example and without limitation, top, bottom, left, right, upper, lower, front, back, and derivatives thereof, relate to the orientation of the elements show n in the drawings and are not limiting upon the claims unless expressly recited therein.

[0022] The disclosed concept will now' be described, for purposes of explanation, in connection with numerous specific details in order to provide a thorough understanding of the disclosed concept. It will be evident, however, that the disclosed concept can be practiced without these specific details without departing from the spirit and scope of this innovation.

[0023] As described in detail herein, the disclosed concept provides a fully automated OCT-based tissue screening sy stem for research using ex vivo tissue cultures, especially for applications requiring high-throughput screening of therapeutic agents, potential therapies, and optimal protocols. The unique features of the system include automated and successive OCT scan acquisition, as w ell as automated and reliable parameter readout, which are empowered with both hardware integrations (motorized platform with object detection ability) and algorithm (e.g., vision transformer for segmentation) innovations.

[0024] FIGS. 1 and 2 are schematic diagrams of an OCT-based tissue screening system 5 according to an exemplary embodiment of the disclosed concept, and FIG. 3 is a schematic diagram of an operational pipeline of the fully automated OCT-based tissue screening system of the disclosed concept that, in the exemplary’ embodiment, may be implemented in screening system 5. As seen in FIG. 1, OCT-based tissue screening system 5 is made up of four “arms” comprising a laser arm 10, a reference arm 15, a sample arm 20, and a detection arm 25. Laser arm 10 is structured and configured to generate a light signal for delivery' to reference arm 15 and sample arm 20. and includes a light source 30, such as a laser or a diode, a collimating lens 35. and a beam splitter/fiber coupler 40. Reference arm 15 is structured and configured to provide a reference signal for OCT-based tissue screening system 5 wherein the optical length of the reference path matches the optical length of sample arm 20, and includes a collimating lens 45 and an axial (z) scanning dichroic mirror 50. Sample arm 20 is structured and configured to deliver the light into the sample for imaging, and includes a collimating lens 55, a lateral (x, y) scanning dichroic mirror 60, an objective lens 65, a camera 70 (such as, without limitation, a webcam or another type of digital imaging device), and a multi-well tissue culture plate 75 for holding multiple tissue cultures. Detection arm 25 is structured and configured to merge the two light beams from refence arm 15 and sample arm 20 so they may interfere and be collected for computer processing as described herein, and includes a collimating lens 80, a photodetector 85, and a control system 90.

[0025] OCT-based tissue screening system 5 also includes a motorized platform 150, which is shown in exemplary form FIG. 2, that is coupled to sample arm 20. Motorized platform 150 includes a slidable tray 95 for holding tissue culture plate 75, and three motors and a support structure for selectively moving tray 95 and OCT sample arm 20 in three dimensions under the control of control system 90. Specifically, motorized platform 150 includes a Y-motor 100 coupled to tray 95 by a first sliding structure for selectively sliding tray 95 in the y-direction under the control of control system 90, a Z-motor 105 coupled to sample arm 20 by a second sliding structure for selectively moving sample arm 20 in the z- direction under the control of control system 90, and an X-motor 110 coupled to sample arm 20 and Z-motor 105 by a third sliding structure for selectively moving sample arm 20 and Z- motor 105 in the x-direction under the control of control system 90.

[0026] Operation of the OCT-based tissue screening system 5 of the disclosed concept will now be described with reference to the operational pipeline of the disclosed concept shown in FIG. 3. It will be understood, ho ever, that this is not meant to be limiting and that the pipeline described herein may be used in connection with other similarly structured OCT systems without departing from the scope of the disclosed concept.

[0027] The first bottleneck challenge of tissue screening systems is to evaluate manytissue samples in a fast and robust manner, as conventional histological imaging is prohibitively complicated and time-consuming for a screening method. To solve this problem, the disclosed concept provides motorized platform 150 described above comprising X-motor 100, Y-motor 105, Z-motor 110 and the associated support/sliding structures in the illustrated exemplary embodiment. More specifically, X-motor 100 and the associated support/sliding structures form an X-linear stage, Y-motor 105 and the associated support/sliding structures form a Y- linear stage, and Z-motor 1 10 and the associated support/ sliding structures form a Z- linear stage. Tissue culture plate 75 sits in tray 95, which is provided with dimensions carefully customized for standard multi-well culture plates (e.g., 86x l28-mm standard plates or other standard plate of different dimensions) and for installation of the slider of the Y- linear stage of customized motorized platform 150. Motorized platform 150 thus allows for 3-D manipulation of the position of tray 95 and sample arm 20 for automated and successive OCT imaging across tissue samples held in tissue culture plate 75. Additionally, camera 70 is mounted parallel to sample arm 20, with its field of view (FOV) adjusted to cover an area slightly bigger than a well of tissue culture plate 75 held in tray 95. After being installed, the position of camera 20 is precisely measured and calibrated with sample arm 20 (X- and Y- distances) to calibrate the field of view from the OCT images and the camera images and guide OCT imaging.

[0028] Referring again to the pipeline summarized in FIG. 3 and using the exemplary screening system 5 shown in FIGS. 1 and 2 for illustrative purposes, reference arm 15 and the focus and the polarization of the OCT imaging are pre-optimized and set before imaging. For the commencement of imaging, the X- and Y- linear stages are initially set to place camera 70 at the center of the first well (Al: first row and first column) of tissue culture plate 75. After that, a picture of the well is taken by camera 70 and fed into an object detection algorithm implemented in control system 90 to identify the existence and calculate the precise location of the tissue sample, if any. The output of the object detection is a bounding box suggesting the target tissue area. The centroids of the sample, i.e., the X- and Y- coordinates, are then converted to the corresponding voltages to drive the adjustment of X motor 110 and Y motor 100 and bring the OCT light beams from light source 30 to the tissue area. In the exemplary embodiment, the centroid is the centroid of the rectangular bounding box that surrounds the tissue sample. Alternatively, the centroid can be the centroid of the tissue sample itself and not the bounding box.

[0029] After the OCT light beam is horizontally moved to the tissue area, the position of Z motor 105 is iterated to adjust the OCT light beam to search for the optimal imaging depth. The depth is determined by examining the intensify averaged across the entire B-scan. More specifically, this process includes (i) obtaining a B-scan of the tissue sample at each of multiple depths, (ii) determining an intensify for each B-scan/depth by averaging the intensify across the B-scan, and (iii) identifying/choosing the depth that yielded the highest average intensify/brightness as the optimal depth. The position of Z motor 105 with maximal brightness is noted/saved and referred to later to trigger a saving module for OCT scan recording. From the experimental experience of the inventors, the optimal position of Z motor 105 can be maintained and used among multiple samples. Therefore, the iteration process just described can performed once within one tissue culture plate 75 or limited within a narrow range during practice.

[0030] Once the optimal imaging depth is determined as just described, an OCT scan volume is acquired for the tissue sample. Specifically, in this step, an OCT scan is taken volumetrically in 3-D, which consists of hundreds of cross-sectional images (B-scans) at different tissue locations. After finishing the scan acquisition, each of the B-scan images of the volume is segmented by a pre-trained deep-leaming network to delineate the tissue regions. In the exemplary embodiment, after segmenting all B-scans within the volume, a 3-D binary' mask is created to delineate the tissue regions in 3-D. Then, the thickness (pm), area (mm²), and volume (mm³) are calculated as readouts from the segmented binary images. These quantitative readouts are stored and displayed for downstream statistical analysis. [0031] In the exemplary embodiment, the object detection is performed using a deep learning algorithm based on the well-known Single Shot MultiBox Detector (SSD), which is described in Liu, et al.. Ssd: Single shot multibox detector, Computer Vision-ECCV 2016: 14th European Conference, Amsterdam. The Netherlands, October 11-14, 2016. Proceedings, Part 1 14, pages 21-37. The advantage of SSD is that it maintains accuracy and significantly improves processing speed by eliminating proposal generation and subsequent pixel or feature resampling stages, as well as encapsulating all computation in a single network. In the exemplary embodiment, convolutional feature layers with progressively decreased size were added to the truncated base network, allowing prediction of detection at multiple scales. During experimentation performed by the present inventors, three hundred camera images of the wells with a FOV of 30x30-mm w ere acquired and manually labeled for training (240 images) and testing (60 images). A weighted sum of the localization loss and the confidence loss was utilized during training. A batch size of 4 and stochastic gradient descent (SGD) optimizer with an initial learning rate of 5e-4 were used. During this experimentation, the centroid of tissue samples was able to be determined within 7 milliseconds. In addition, the mean average precision (with Intersection Over Union threshold set at 0.5 to 0.95 with an interval of 0.05) was 0.889, indicating a precise location detection. The centroid shift between the predictions and the labels w as ~10 pm. The success rate of detection w as 100%, w ith void output for all empty w ells, indicating a high reliability' of this algorithm.

[0032] Moreover, the challenges during segmentation include frequent artifacts of specular reflection, inconsistent tissue reflectance, and strong interference from the tissue- supporting membrane. To surpass these challenges, the disclosed concept employs a transformer-based method employing a hybrid architecture of a number of residual neural network (ResNet) blocks (described in He et al.. Deep residual learning for image recognition, Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pages 770-778) and a number of multi-scale hierarchical transformer blocks (described in Lee et al., Multi-path vision transformer for dense prediction, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pages 7287- 7286) to capture both the local and global features efficiently. This is shown schematically in FIG. 4. As show n in FIG. 4, the backbone of the model is comprised of tw o independent networks including multipath transformer networks for extracting long-distance dependencies of explant features and a residual network for extracting local features. OCT B-scan images are first processed by these two backbones in multiple scales in a parallel fashion. Then, the local and global features provided by these two networks are concatenated and sent to the corresponding decoder block to produce the segmentation map.

[0033] During further experimentation performed by the present inventors, for tissue segmentation, 2150 manually labeled B-scan images were used for segmentation training. Data augmentation including translation, contrast adjustment, and vertical/horizontal flip was applied to enlarge the training set. The learning rate (le-2), batch size=2, epochs number=30, and AdamW optimizer were utilized. The recall, precision, and dice rates of the segmentation model were 0.8680±0.2043, 0.8447±0.2451, and 0.8926±0.2388. respectively. The processing time for each B-scan image was around 150 milliseconds. These parameters suggested a fast and accurate segmentation performance by the proposed algorithm.

[0034] FIG. 5 show s exemplary en face images obtained from performing the pipeline of the disclosed concept on mouse retinal explant cultures, including the structure and thickness heatmap. The images shown in FIG. 5 are at Baseline (P15) and Day 10 (P25) treated with negative and positive compounds, and the histology images were obtained after Day 10 screening for validation. Tissue area and volume are calculated by projecting and integrating all B-scans. To avoid bias from peripherals, average thickness is calculated only in the central tissue regions with thickness larger than the median value. For characterization, the system was programmed to take five volumes from each sample. Repeatability was calculated as pooled-standard deviation among the repetitions for all samples. The repeatability was high for all readouts (volume: 0.005 mm3, area: 0.039 mm2, thickness: 0.4 pm), indicating highly reliable measurements offered by the system of the disclosed concept. The reproducibility, calculated as the pooled-standard deviation among the samples at same condition, was 0.107 mm3 for volume, 0.755 mm2 for area, and 12.1 pm for thickness, all corresponding to ~5% variation of the measurements.

[0035] FIG. 6 is a schematic diagram of an exemplary control system 90 according to an exemplary embodiment of the disclosed concept. As seen in FIG. 6, control system 90 is a computing device structured and configured to receive the signals from photodetector 85 and process that data as described herein. Control system 90 may be, for example and without limitation, a PC, a laptop computer, or any other suitable device structured to perform the functionality described herein. Control system 90 includes an input apparatus 115 (such as a keyboard), a display 120 (such as an LCD), and a processing apparatus 125. A user is able to provide input into processing apparatus 125 using input apparatus 115, and processing apparatus 125 provides output signals to display 120 to enable display 120 to display information to the user (such as images generated from a tissue culture and readout data as described herein). Processing apparatus 125 comprises a processor and a memory. The processor may be, for example and without limitation, a microprocessor (pP), a microcontroller, or some other suitable processing device, that interfaces with the memory. The memory can be any one or more of a variety of types of internal and/or external storage media such as, without limitation, RAM, ROM, EPROM(s), EEPROM(s), FLASH, and the like that provide a storage register, i.e., a non-transitory' machine readable medium, for data storage such as in the fashion of an internal storage area of a computer, and can be volatile memory or nonvolatile memory. The memory has stored therein a number of routines (comprising computer executable instructions) that are executable by the processor, including routines for implementing the disclosed concept as described herein. In particular, processing apparatus 125 includes an object detection component 130 for performing object detection as described herein, a motor control component 135 for controlling X motor 110, Y motor 100, and Z motor 110 as described herein, an imaging depth optimization component 140 for determining the optimal imaging depth as described herein, and a segmentation component 145 for segmenting the B-scans as described herein.

[0036] Thus, in short, the disclosed concept provides for the first time an innovative OCT-based tissue screening system tailored for high-throughput applications in, for example, drug discovery and evaluation, providing multi-plexed readouts for unbiased tissue morphological description. Advantageous aspects include a custom-designed motorized platform for precise 3-D positioning of samples alongside sophisticated deep learning algorithms for automated tissue detection and segmentation. Validated through rigorous testing with retinal explant cultures, the disclosed concept is able to accurately measure drug efficacy, overcoming the limitations in the traditional histology methods.

[0037] While specific embodiments of the invention have been described in detail, it will be appreciated by those skilled in the art that various modifications and alternatives to those details could be developed in light of the overall teachings of the disclosure.

Accordingly, the particular arrangements disclosed are meant to be illustrative only and not limiting as to the scope of disclosed concept which is to be given the full breadth of the claims appended and any and all equivalents thereof.

Claims

What is claimed is:

1. An OCT-based tissue screening system, comprising: a sample arm structured and configured to deliver a light signal for OCT imaging of a tissue sample; a camera positioned adjacent to the sample arm; a motorized platform structured and configured to enable 3-D manipulation of a relative position between the sample arm and the tissue sample; and a processing apparatus including: an object detection component configured to determine a location of the tissue sample based on an image of the tissue sample captured by the camera; a motor control component configured to operate the motorized platform to cause the sample arm to be aligned with the tissue sample based on the determined location of the tissue sample such that the light signal will be directed into the tissue sample by the sample arm; an imaging depth optimization component configured to determine an optimal imaging depth for the system when the sample arm is aligned with the tissue sample based on a plurality of OCT intensities obtained using the sample arm at a plurality of depths when the sample arm is aligned with the tissue sample; and a segmentation component configured to (i) receive a B-scan volume of the tissue sample captured at the optimal imaging depth when the sample arm is aligned with the tissue sample, wherein the B-scan volume includes a number of B-scans, and (ii) delineate a number of tissues regions within each of the number of B-scans.

2. The tissue screening system according to claim 1, wherein the segmentation component configured to is further configured to calculate a number of readout parameters for each of the number of B-scans.

3. The tissue screening system according to claim 2, wherein the readout parameters include thickness, area and volume.

4. The tissue screening system according to claim 1, wherein the object detection component implements a neural network for determining the location of the tissue sample.

5. The tissue screening system according to claim 4, wherein the neural network comprises a deep learning algorithm.

6. The tissue screening system according to claim 5, wherein the deep learning algorithm is based on a Single Shot MultiBox Detector (SSD).

7. The tissue screening system according to claim 1, wherein the location of the tissue sample is based on a centroid of the tissue sample determined from the image of the tissue sample captured by the camera.

8. The tissue screening system according to claim 7, wherein the object detection component is configured to generate a bounding box surrounding the tissue sample from the image of the tissue sample captured by the camera, and wherein the centroid is a centroid of the bounding box.

9. The tissue screening system according to claim 1, wherein each of the plurality of intensities is a B-Scan based intensity.

10. The tissue screening system according to claim 9, wherein each B-scan based intensity is an average of intensities across a B-scan that corresponds to the B-Scan based intensity, and wherein the optimal imaging depth corresponds to the depth where the B-scan based intensity is a maximum of the B-scan based intensities.

11. The tissue screening system according to claim 1, wherein the segmentation component includes a plurality of transformer neural network blocks each for extracting global features comprising long-distance dependencies of explant features and a plurality of residual neural network blocks each for extracting local features, wherein the transformer neural network blocks and the residual neural network blocks are organized in a plurality of associated pairs of blocks each comprising one of the transformer neural network blocks and one of the residual neural network blocks.

12. The tissue screening system according to claim 10, wherein in the segmentation component each B-scan (i) is first processed by each of the associated pairs of blocks in multiple scales in a parallel fashion to produce the global features and the local features of each of the associated pairs of blocks, and (ii) the global features and the local features of each of the associated pairs of blocks are concatenated and sent to a corresponding decoder block of a decoder to produce a segmentation map.

13. The tissue screening system according to claim 10, wherein the motorized platform includes a first motor, a second motor and a third motor, wherein the sample arm is coupled to the first motor and the second motor, wherein the first motor is structured and configured to move the sample arm in a first direction and wherein the second motor is structured and configured to move the sample arm in a second direction that is perpendicular to the first direction, wherein the tissue screening system further comprises a tray coupled to the third motor for holding the tissue sample, and wherein the third motor is structured and configured to move the tray in a third direction that is perpendicular to the first direction and the second direction.

14. An OCT-based tissue screening method, comprising: determining a location of a tissue sample based on an image of the tissue sample captured by a camera coupled to an OCT sample arm structured and configured to deliver a light signal for OCT imaging of the tissue sample; aligning the sample arm with the tissue sample based on the determined location of the tissue sample such that the light signal will be directed into the tissue sample by the sample arm; determining an optimal imaging depth when the sample arm is aligned with the tissue sample based on a plurality of OCT intensities obtained using the sample arm at a plurality of depths when the sample arm is aligned with the tissue sample; receiving a B-scan volume of the tissue sample captured at the optimal imaging depth when the sample arm is aligned with the tissue sample, wherein the B-scan volume includes a number of B-scans; and delineating a number of tissues regions within each of the number of B-scans.

15. The tissue screening method according to claim 14, further comprising calculating a number of readout parameters for each of the number of B-scans.

16. The tissue screening method according to claim 15, wherein the readout parameters include thickness, area and volume.

17. The tissue screening method according to claim 14, wherein an object detection component implementing a neural network determines the location of the tissue sample.

18. The tissue screening method according to claim 17, wherein the neural network comprises a deep learning algorithm.

19. The tissue screening method according to claim 18, wherein the deep learning algorithm is based on a Single Shot MultiBox Detector (SSD).

20. The tissue screening method according to claim 17, wherein the location of the tissue sample is based on a centroid of the tissue sample determined from the image of the tissue sample captured by the camera.

21 . The tissue screening method according to claim 20, wherein the object detection component is configured to generate a bounding box surrounding the tissue sample from the image of the tissue sample captured by the camera, and wherein the centroid is a centroid of the bounding box.

22. The tissue screening method according to claim 14, wherein each of the plurality⁷ of intensities is a B-Scan based intensity.

23. The tissue screening method according to claim 22, wherein each B-scan based intensity is an average of intensities across a B-scan that corresponds to the B-Scan based intensity⁷, and wherein the optimal imaging depth corresponds to the depth where the B- scan based intensity is a maximum of the B-scan based intensities.

24. The tissue screening method according to claim 14, wherein the delineating and calculating are performed by a segmentation component that includes a plurality of transformer neural network blocks each for extracting global features comprising longdistance dependencies of explant features and a plurality of residual neural network blocks each for extracting local features, wherein the transformer neural network blocks and the residual neural network blocks are organized in a plurality of associated pairs of blocks each comprising one of the transformer neural network blocks and one of the residual neural network blocks.

25. The tissue screening method according to claim 24, wherein in the segmentation component each B-scan (i) is first processed by each of the associated pairs of blocks in multiple scales in a parallel fashion to produce the global features and the local features of each of the associated pairs of blocks, and (ii) the global features and the local features of each of the associated pairs of blocks are concatenated and sent to a corresponding decoder block of a decoder to produce a segmentation map.

26. A computer program product, comprising a non-transi lory computer usable medium having a computer readable program code embodied therein, the computer readable program code being adapted to be executed to implement a method of tissue screening as recited in claim 14.