JP4653041B2 - System and method for synthesizing image blocks and creating a seamless enlarged image of a microscope slide - Google Patents

Description

Related applications

  This application relates to US provisional applications 60 / 605,835 and 60 / 609,961 filed on August 31, 2004 and September 15, 2004, respectively. Accompanied by a priority claim. This application is also related to the "system and method for creating an enlarged image of a microscope slide" which is ongoing and is assigned to the same assignee as this application. Incorporate.

  The present invention relates generally to systems and methods for creating virtual microscope slides, and more particularly to methods and systems for creating seamless images of virtual microscope slides.

  Virtual microscope slides typically contain digital data representing an enlarged image of the microscope slide. Since the virtual slide has a digital format, it can be stored in a medium such as a computer memory, and can be transmitted to a viewer at a remote place via a communication network such as the Internet or an intranet.

  Virtual slides produce a significant effect compared to traditional microscope slides. In some cases, virtual slides allow physicians to make diagnoses more quickly, conveniently, and economically than was possible with traditional microscope slides. For example, a virtual slide can also be used by a remote user, such as a remote expert over a communication link, so that the doctor can ask for the expert's opinion and make a diagnosis without delay . Alternatively, virtual slides can be stored in digital form and considered at a time convenient for physicians and specialists.

  Typically, the virtual slide is one or more covering a microscope slide (including the sample whose image is desired to be magnified) positioned under the microscope objective and covering all or part of the slide. It is obtained by capturing images and combining the multiple images to create a single integrated digital image of the slide. It may be preferable to divide a slide into multiple regions and generate separate images for each region. The reason is that, in many cases, the entire slide is larger than the field of view of a high performance objective lens (eg, 10x, 20x, 40x objective lens) and captures the entire slide image at the desired magnification. This is because a plurality of images must be acquired for this purpose. Furthermore, many tissue type surfaces are non-uniform in height and contain local variations that are difficult to capture a focused image.

  When a sample is larger than one field of view, the imaged area must be combined in some way to create a single image. These image blocks need to be synthesized so that the created image does not suffer from defects due to imperfect alignment of the image. This can be achieved in several ways. According to some systems, using a very accurate xy stage, the images are produced in the correct position so that multiple images can be simply contiguous to obtain a properly matched set of images. Would be able to. Other methods also rely on software algorithms to determine the best match for the set of images. This process is often referred to in the art as stitching. Sometimes caused by mosaicing. These stitching algorithms attempt to optimally position these image blocks using a large number of images, resulting in a single image that is seamless and well formed.

  Some algorithms use image modification, such as image warping, to produce a more optimal alignment between image blocks. However, this process has a number of drawbacks, especially for imaging systems used in medicine. This is because accuracy is very important for the person making the diagnosis. Image modification can cause undesirable phenomena such as reduced resolution and aliasing and is undesirable. Also, image correction can be expensive and time consuming from a computational point of view. Furthermore, creating an optimal image position in the global coordinate system can result in multiple solutions. Which solution to choose from and the reproducibility of the process exist as problems to be solved.

  Certain features, aspects, and advantages of the present invention relate to a type of microscope system in which a physical material sample slide that is movable in a microscope optical field of view is supported for viewing and image capture. The present invention is a method for observing a digital virtual slide having seamless edge quality characteristics, comprising defining a plurality of digital region images for each physical slide, and according to the defined plurality of digital region images. Scanning the physical slide.

  The scanned image is stored according to a position index metric associated with the position of each image in the entire mosaic representation of the physical slide, and a normalized correlation search is performed on the next adjacent region image block. Executed.

  A set of relative position offset values and correlation coefficients is determined for the region image block and the next adjacent region image block. A portion of the region image block is viewed as a field of view of the display, and a composite of the portion of the region image block is synthesized according to the set of relative position offset values and correlation coefficients to form the portion Only is synthesized.

  Effectively, by moving the field of view of the display, additional region image blocks are displayed, and the compositing step is performed only with respect to the additional region image blocks brought into the new field of view. Is done.

  In an additional aspect of the invention, for each physical slide, a plurality of regions are defined, each region having an area characterized by at least a width dimension W. An image of a first defined area is captured, where the microscope is positioned at a first position relative to the physical slide. The position of the microscope is adjusted to a second position along a dimension characterized by W and an overlap value (OverlapX), and an image of a second defined region is captured, The second area and the first area share the image data included in the overlapping area defined by OverlapX. A particular feature of the present invention is that the determining step is performed on image data included in an overlapping region between adjacent regions.

  In another aspect of the invention, for each physical slide, a plurality of regions are defined, each region having an area characterized by at least a height dimension H. An image of a first defined area is captured, where the microscope is positioned at a first position relative to the physical slide. Next, the position of the microscope is adjusted to a second position characterized by H and a second overlap value (OverlapY), and an image of a second defined region is captured, and the second The area and the first area share image data included in the overlapping area defined by OverlapY.

  Effectively, the determining step of the method is performed on image data contained in any of the overlapping areas between adjacent areas. Alternatively, the determining step of the method is performed on image data included in both overlapping regions between adjacent regions.

  The set of relative position offset values and correlation coefficients for a region image block and the next adjacent region image block are determined by a normalized correlation search and are relative positions associated with each region image block. The set of offset values and correlation coefficients is stored as a data set corresponding to the region image block. A subset of the region image blocks is provided for viewing on the display and positions each block relative to its next adjacent block according to the set of relative position offset values determined relative to the block. Thus, overlapping portions of the next adjacent region image block are combined.

  In another aspect, the present invention is an observation system that images digital data representing a virtual microscope slide and creates a composite image that defines the field of view of the observation system by seamlessly combining region images with each other. It can be characterized as a method. In this respect, the method according to the invention comprises the step of dividing the physical microscope slide into a matrix of a plurality of composite area images, each area image having a row and column index position in said matrix. Scanning the physical microscope slide with the selected objective lens such that the field of view of the objective lens is equal to the size of one composite area image, the current area image, and the next adjacent Overlapping a specified portion of a region image to be overlapped such that the two images share image data in the specified portion, and an offset to the current image and its next adjacent image Determine a set of values, and the overlap for the current image is the next adjacent Making the image highly correlated with overlapping portions of the image; displaying the current image synthesized with the next adjacent image without considering the offset value of the currently displayed image; including. The method further includes generating a correlation coefficient for the overlapping portion based on the determined offset value.

The features that produce the effect of the present invention include the step of evaluating which particular image in the composite region image is required to represent the currently used field of view of the virtual microscope slide; Searching the particular image in with a correlation coefficient associated with the corresponding set of offset values and finding the searched region image in the viewer according to its nominal index position in the matrix Comparing a correlation coefficient value and a threshold value of a next adjacent image, and defining a positive connection criterion between a plurality of next adjacent images having a correlation coefficient value equal to or greater than the threshold. And the overlap of the next adjacent image defines a seamless edge between the two next adjacent images So that the further comprises, placing said retrieved images with each other.

  A connected component labeling algorithm is applied to the plurality of region images, and a common algorithmic label is assigned to a region image block having an associated positive connection criterion. An image block with a common algorithmic label is formed on the connected components. From one of the current image block and the next adjacent image block, one overlapping area is selected to represent the digital image data included in the overlapping area, and the retrieved image is Positioned relative to each other to define a seamless edge between two adjacent images.

  The present invention can also be implemented as a type of microscope system in which a physical material sample slide that is movable in the optical field of view of the microscope is supported for viewing and image capture. The system includes a microscope that includes a rotatable turret that holds a plurality of microscope objectives each having a corresponding magnification. The system further includes a sample stage that is movable along orthogonal plane axes and is robot controlled and means for determining the current position of the stage relative to the original index position along each of the axes. Including.

  A processor coupled to the stage controls stage movement such that the physical slide is scanned as digital data into a memory data store, the physical slide comprising a matrix of a plurality of composite image blocks Each image block has a corresponding row and column position in the matrix. Each image block is characterized by dimensions in the first and second directions, and the processor allows the next adjacent image block to be identified between those image blocks and to have a reproducible overlap. Scanned to occur.

  The system further comprises a correlation processor that determines an offset value along the first and second directions during the next adjacent image block, the offset value being Define a relative positioning vector required to position the next adjacent image blocks relative to each other such that overlapping portions of these image blocks appear as seamless edge regions.

A viewer suitable for displaying a portion of the matrix of the composite image block as a high resolution partial image of the physical slide is an image compositing means under microprocessor operation control, comprising: In order to provide a composite image block for this purpose, it can include image composition means for compositing the next adjacent blocks with each other according to their defined relative positioning vectors. The image synthesizing means synthesizes only the image blocks necessary for constructing a desired field of view in the viewer, and further causes the user to move to a new field of view along the first and second directions. When doing so, it operates to combine the resulting image blocks into a single field of view.

  The above and other features, aspects and advantages of the present invention will become more apparent upon consideration of the following detailed description, the appended claims and the accompanying drawings.

  The various embodiments described herein provide systems and methods for creating well stitched virtual microscope slides. In certain aspects of an embodiment, an optimal relative alignment between each image and the next image is calculated. The calculated optimal relative alignment is then used by a dynamic stitching system to provide a plurality of properly stitched images, for example in an image viewer. This dynamic synthesis system can be implemented in either software or hardware, or a combination of the two, as will be appreciated by those skilled in the software and firmware arts. it can.

  During the viewing process, this dynamic composition system determines the optimum alignment for the various blocks of the image that contribute to the currently viewed image. The relative alignment for each of the image blocks that contribute to the area being observed is examined in comparison with the relative alignment for the other blocks. By ignoring image blocks that do not contribute to areas not currently observed, the synthesis system can be more flexible when aligning blocks. An optimized set of positioning offsets is determined for the viewing area, and the blocks of the image that contribute to the current viewing area are optimally stitched together (stitched, synthesized). As the current field of view changes, it will be necessary to provide additional blocks of images to fill the new field of view, and the dynamic compositing system will be called again, and new blocks for new blocks that contribute to the new viewing area. The optimal position offset is determined.

  By making optimal alignment decisions limited to the area currently observed, the ability to create well stitched virtual slides is significantly improved in both speed and performance. Prior systems sought to create a single globally synthesized image on an a priori basis in a process referred to herein as “global synthesis”. In fact, before attempting to display such an image, all image blocks were processed, all offsets were determined, and a “perfect” image was envisioned. However, an attempt to create a single globally synthesized image is a very difficult problem to solve, even with a very accurate mechanical system. Due to distortion characteristics, which are some well-known features of microscope optical systems, such as non-telecentricity and non-flatness, a change in magnification that results in spatial distortion of the image results.

  The absolute frequency of optical distortion of a single image is often very small, for example on the order of one pixel, but the optical distortion is not linear in nature. When trying to synthesize multiple image blocks (often hundreds and sometimes thousands) together using a global synthesis system, this type of error is cumulative and does not distort the image. A global synthesis is not possible without introducing. Resolving image distortion has its own problems. For example, the computational burden is high, the resolution of the image is reduced, and it is very difficult to make the resulting image parallel to the scanning process.

  Fortunately, most imaging applications generally do not require the creation of a globally synthesized “master image”. However, what is needed is that the particular field of view that a human or imaging program is viewing must be stitched together. By changing the observation paradigm to one that creates a single field of view provided in a well-stitched manner when a new field of view is required by the viewer user, this problem changes dramatically. . Unlike global processes, which have a large a priori computational burden, the method according to the present invention considers that only images that contribute to the current field of view must be stitched together.

  The various embodiments described and shown in this application have many advantages over existing systems that have been employed by the prior art. For example, “imaging” is performed on a “local” basis, thus eliminating cumulative errors that impair global synthesis. Since global sewing (synthesizing) is not required, many of the processing techniques that are computationally burdensome in the prior art become unnecessary. Thus, image distortion correction or image projection techniques need only be used in severe situations.

  In one embodiment of the invention, the image block is obtained from a virtual slide imager (when the image is scanned), so that the image block is in a processor provided as part of a microscope scanning system. Are sent simultaneously to the compression system and the relative image matching system. The compression system and the relative matching system are realized in software, firmware, hardware, or a combination thereof. Although described as part of a microscope system, the image matching system can be easily implemented in a processor provided as part of an image viewing system (also referred to as a “viewer”). Can do.

  In a manner well known to those skilled in the art, the compression system compresses the contents of the received image block and writes the compressed data to a virtual slide file. The image matching system then receives the image block and calculates an optimal relative offset for the two images, the left image and the upper image of the selected image. Once these relative offsets are calculated, these offsets are stored in volatile or non-volatile memory, such as a disk drive, DVD, CD-ROM, or other storage device.

  At the end of the scanning process, the compressed image data and the relative composite offset are written to memory storage. During the viewing process, when the user is manipulating the image by identifying a specific viewing area, the compositing system (possibly in the viewer's processor) is called to dynamically and automatically scan the process. The relative offset calculated during is used to determine the optimal alignment for the contributing image. This allows the user to obtain a seamless or nearly seamless image viewing experience. As mentioned above, with respect to compression and matching systems, the synthesis (stitching) system can be implemented as hardware, software or firmware in a processor located in or associated with the observation system. it can.

  As will be described in more detail below, a physical slide must be examined as a series of images, particularly when the physical slide must be evaluated at a high magnification. A single slide expressed as 40x or higher magnification maintenance is composed of multiple (possibly hundreds) individual field scans arranged in a matrix format. A set of scans moving from left to right and rastering for each row is indexed based on position, and the entire data set of physical slides is defined. When such an image is captured and associated with a relative position in the matrix (mosaic), a “virtual slide” is created. In this virtual slide, the digital data represents such optical data, replacing analog optical data observed by a clinician looking through the microscope eyepiece. In this way, each field of view of the microscope defines one image (also referred to as “region” in this application), and during each scan, an image for each defined region is acquired.

  FIG. 1 is a simplified, generally schematic block diagram of an exemplary imaging system 100 used to obtain a magnified image of a microscope slide, according to an embodiment of the present invention. The imaging system 100 suitably includes a microscope system 10, a camera (or imaging) system 11, and a control processor 20. The microscope system 10 includes at least one rotatable objective turret 18 of the type capable of supporting two, three, four or five optical objective lenses or lenses of varying magnification. The microscope system 10 further includes a computer controlled microscope stage 14. The microscope stage 14 is automatically movable in the x, y and z directions and moves the x, y and z translation motors to the stage platform via a control system called x, y control 16 and z control 17. By mechanically coupling, it can be controlled robotically. A suitable light source (not shown) is placed under the microscope stage 14 and a translational movement under the stage is possible to shift the apparent light source for the sample on the microscope stage 14.

  The translational movement of the microscope stage 14 and the intensity of the light source can be controlled, for example, by a software program running as software or a firmware program application on the control processor 20. One skilled in the art will appreciate that a capacitor can be provided that collects the light produced by the light source and directs it in the direction of the sample.

  The rotation of the objective turret 18 moves the desired objective lens into the optical path of the microscope, producing an image of the sample having the characteristic magnification of the objective lens. According to the present invention, for example, one or a plurality of objective lenses held by the turret 18 including a 4 × objective lens and a 20 × objective lens are provided. While not meant to be limiting in any way, examples of robotic controlled microscope systems suitable for use with the present invention include the Olympus BX microscope system with Prior H101. The Olympus BX microscope system is manufactured and sold by Olympus America, Inc., Melville, New York. The Praia H101 stage is manufactured and sold by Praia Scientific, Inc., Rockland, Massachusetts. Other similar computerized stages can be used, examples of which are manufactured and sold by Ludl Electronics Products, Inc., Hawthorne, New York.

  The microscope objective turret 18 is controlled by a focus / objective control system 15 which has a combination of a robot controllable motor and motor driver and is coupled to the objective turret to rotate the turret to various desired objective lenses. Can be moved into the optical path of the microscope. Upon receipt of the appropriate move command signal, the focus / objective control system 15 moves the different objective lens 19 into the optical path of the microscope system by commanding the motor to rotate the rotatable turret 18. Can do.

  In one example, the X, Y control 16 includes a motor that controls the stage 14 in the x, y, and z directions and a suitable motor driver circuit that energizes the motor. In this application, the x and y directions refer to vectors in the plane in which the stage 14 resides, and the z direction refers to the direction perpendicular to both x and y and the vertical or in-focus direction. Preferably, mechanical devices and electronic control circuits that cause stage movement are implemented, including some form of open or closed loop motor positioning servo system, and stage 14 is positioned with great accuracy. Or its translational motion can be determined very accurately in the x, y and z directions. Instead of controlling the movement of the stage 14 in the xy plane, or in addition to controlling such movement, the microscope 12 itself can move in the xy plane. In this alternative embodiment, the translation of the microscope is controlled or its position can be determined with the same accuracy as the stage 14 and using a positioning device substantially the same as the stage 14.

  When the X and Y control 16 is configured to operate in a closed loop manner, the position feedback information is either from the motor itself or, if higher accuracy is desired, an optical position encoder or laser interferometer position. It is possible to recover from the encoder. Closed-loop servo control of stage motion allows the position of the stage to be determined very accurately and ensures that translation commands are responded with very high accuracy. This is well understood by those skilled in the art. Thus, a command to translate the stage in the positive x direction by 50 microns results in the stage 14 being moved exactly 50 microns in the positive x direction at least to the limit of the mechanical resolution of the motor system. It will be.

  When the microscope system is configured to operate in a semi-closed loop or open loop manner, stage control does not depend on the feedback itself, and it is accurate where the motor controlling the stage is commanded to go. It is at least necessary to define For example, a typical stepper motor provides translational motion as a series of “steps” that depend on “direction” and “number of steps” commands. As long as the translational motion per “step” is calibrated (or determined differently or known), the motion (movement) command results in a known (or calculated) stage. Will be moved in the commanded direction by the distance. All that remains is to save a record of the motion commands along with the starting point and determine the “current position” of the microscope stage.

  The camera system 11 includes a main camera 22 and control electronics 24 as shown in the embodiment of FIG. Although a single main camera 22 is shown, multiple main cameras may be provided as appropriate, as will be described in detail later. A position encoder may be coupled to the stage motor or the stage itself, and is configured to transmit a signal indicating the position of the stage 14 to the main camera 22 or its control electronics 24. With such a configuration, the camera can capture an image at a desired predetermined position even while the stage 14 is continuously moving. For example, the position encoder monitors the distance traveled by the stage 14 (or a motion command in the case of an open loop or semi-closed loop system) and sends a predetermined signal every time the stage moves several microns. Send. The main camera 22 captures images in response to a set or subset of such electrical signals received from a positioning feedback device, such as a rotary or linear scale encoder, for example, at regular intervals. Configured to produce an image.

  In one particular example, a linear encoder located along the scan axis of the slide provides absolute position feedback to the control system to generate an accurate periodic signal for image capture. These periodic signals act as external triggers to the camera for fast and consistent cross-sectional image capture. This embodiment solves a number of positioning error problems, for example, relating to the true conversion of electrical control signals to the actual mechanical position of the slide with respect to the image plane of the camera as follows: Error (the difference in position from the electrically commanded position to the actual mechanical response of the positioning system to that commanded position). This embodiment also provides protection against periodic degradation of mechanical hardware caused by repetitive use of lead screws, loose coupling, friction, environmental issues, and the like.

  Alternatively, the main camera 22 can be configured to capture images at regular time intervals or based on pulses sent to the motor. For example, a control pulse sent to a stepper or linear motor can be used. These are, for example, raw transistor-transistor logic (TTL) signal pulses that are provided via an electronic counter circuit that generates absolute or relative output pulses and triggers the camera for image capture, or is amplified. Or a control pulse. The TTL step and command signal generated via the stepper controller pulse generator as described above can be fed back to the controller via the encoder feedback channel. In this configuration, an integrated real-time “pulse counter” counts the pulses to produce a periodic pulsed output for the camera. This technique is used in conjunction with a motor directional signal output as an input to a controller for bidirectional or unidirectional output trigger pulse control that captures an image based on the direction of motion. Alternatively, a clockwise or counterclockwise mode of operation can be used for motor control, and directional pulses can be returned to the controller for periodic camera triggers synchronized with motion.

  In one aspect of the invention, focusing is performed by causing the stage 14 to slightly shift in the z direction under the control of a corresponding z control circuit 17. Since the relative momentum during focusing is significantly less than the relative momentum during net z translation, the focus circuit 17 is a micro controlled by a suitable motor driver circuit operating in parallel with the z-axis stage translation motor. Includes stepper motor. Thus, the z-axis stage translation motor is provided with a greater net response characteristic, and thus the motor is capable of optical sectioning of the sample in the vertical direction. That is, the sample can be viewed on various horizontal planes that are arranged to pass vertically to the sample, and at the same time, the focus motor properly focuses each image plane. It can provide the micromotion necessary to do this.

  In another embodiment, focusing is performed by slightly deflecting the objective turret 18 in the z direction under the control of the focus / objective control system 15. For example, the piezoelectric transducer can perform focusing by slightly deflecting the objective turret 18 and its corresponding objective lens 19 in response to a signal received from the piezoelectric amplifier.

An illumination circuit (not shown) controls the intensity of the illumination light source in a conventional manner.
The main camera 22 is preferably a high resolution color digital camera that operates at high resolution and high data rate. In this embodiment, it is assumed that a JAICV-M7CL + camera is used, but other cameras having image quality and resolution comparable to this can be used within the scope of the present invention. Images captured by the main camera 22 are sent to the control processor 20 via control electronics 24 such as a camera link card. As is well known to those skilled in the art, a camera link card interfaces with a digital camera that supports a specific protocol and physical interface. Other protocols and physical interfaces are possible with the present invention, and the specific interfaces described herein are not meant to be limiting in any way.

  The control processor 20 can be implemented as a small platform computer system such as an IBM type x86 personal computer system, but is suitable for generating the commands and control signals necessary to operate the microscope system. • Data processing and platform performance to host software programs. The control processor 20 includes special software or circuitry that can perform image processing functions. For example, the control processor 20 performs image analysis and obtains measurements such as contrast, entropy and sharpness. The control processor 20 also includes special software or circuitry that can manipulate and synthesize digital images. The control processor 20 can receive and interpret commands generated by a system user on a conventional input device such as a mouse or keyboard, and can further manipulate user-defined commands to operate various components of the microscope system. It can be converted into a suitable signal.

  The control processor 20 typically includes a serial interface, a peripheral component interconnect (PCI) interface, and a number of other couplings that define the system interface to which the various control electronics that operate the microscope system are connected. The microscope system 100 is coupled via an arbitrary interface among the interfaces.

  Embodiments of the present invention provide an improved system and method for constructing virtual slides that are extremely independent of accurate mechanical positioning and whose computational requirements are much smaller compared to the prior art. provide.

  FIG. 2 is a flowchart 300 of an example method for generating an image of a microscope slide in accordance with the present invention. In this regard, FIG. 3 is an example of a region of interest (ROI) of the microscope slide S, in which four images 501, 502, 503 and 504 arranged in two rows (row I and row II) are shown. For example, for the user, the image provided on the viewer is constructed. The position of these images is determined by the X and Y controls 16 of the stage 14.

  Referring to FIG. 2 in conjunction with FIG. 3, multiple regions of the slide are defined during scanning. Each region has a width W and a height H approximately equal to the width and height of the field of view of the objective lens of the microscope. The objective of the microscope is initially positioned so that it is located at a point P1 (0, y0) where the center of the field of view is above one edge of the slide. Point P1 is in this case the original position of row I. In step 312 (FIG. 2), an image corresponding to region 501 is acquired in row I, which has a width W and a height H. In step 313, the X and Y control 16 moves the stage 14 relative to the objective lens, and the point P1 moves to P2, which is separated by a distance equal to (W-OverlapX). As will be apparent to those skilled in the art, to adjust point P1 to point P2, objective turret 18 is also moved relative to the slide, as well as other movements as described below.

  In the context of the present invention, “OverlapX” is a predetermined numerical value that almost specifies an overlap (overlap) that occurs in the x-axis direction between a plurality of adjacent images. This number should preferably be as small as possible, generally at least the minimum area required for a meaningful match between the search radius in the x direction (herein referred to as “search RX”) and the image region edge. Must be greater than the sum of The size of the search radius is a function of the mechanical accuracy of the system. The minimum matching area depends on the resolution of the optical system and the size of the outer shape of the tissue that constitutes the content of the slide being viewed. Conversely, the mechanical accuracy of the system depends on cumulative errors that impair the positional accuracy, such as vibrations, computer controlled stage 14 tolerances, and other mechanical system discontinuities. . An example of a suitable value for the search RX is 16 pixels, and a suitable value for the exemplary OverlapX is 48 pixels.

  Returning again to FIG. 2, in step 314, an image corresponding to region 502 is acquired at point P2. In step 315, the system determines whether the end of the scan row (row I in this case) has been reached by examining whether the position of the microscope objective relative to the sample is at the end of the scan row. If not, the system returns to step 313 to move and acquire steps 313 and 314 until the end of the current scan row is reached (ie, the answer to the query in step 315 is yes). Iterate. Once the end of the scan row is reached, the system determines in step 316 whether there are more rows to scan. Since there is scan row II in this example, in step 317 the X and Y control system 16 moves the stage relative to the optical device in the y direction to a position P3 by a distance of (H-OverlapY).

  “OverlapY” is a predetermined numerical value that almost specifies an overlap (overlap) that occurs in the y-axis direction between a plurality of adjacent images. This numerical value is preferably as small as possible, and is generally at least a value obtained by adding a search radius in the y direction (herein referred to as “search RY”) and a minimum area necessary for meaningful matching. Must also be large. As described above for search RX, the size of the search radius is a function of the mechanical accuracy of the system. The minimum coincidence area depends on the resolution of the optical system and the size of the outer shape of the tissue. Similar to the example described above with respect to the x-axis, an example of a suitable value for the search RY is 16 pixels and a suitable value for the exemplary OverlapY is 48 pixels.

  When the system 100 is at position P3, an image corresponding to region 503 is obtained at step 314. The system 100 then repeats steps 313-315 in the x direction opposite the x direction of the preceding scan row (in this case, this opposite x direction is the negative x direction), Move to position P4 by a distance (W-OverlapX). An image corresponding to the next region 504 is then acquired in step 314. The result is a series of overlapping images. The edge region of each image overlaps with the edge region of the next image along both x and y edges.

  As those skilled in the art will appreciate, due to the lack of mechanical accuracy in the positioning system as described above, the actual captured position of the image is determined by the X and Y controls 16. Is not exactly the same as the position to be.

  In this regard, FIG. 4 shows an example of the actual position of the captured images 501 through 504, showing the position error obtained during the scan. These errors can be small or large, depending on the resolution of the image acquisition system and the mechanical accuracy of the motion control system. Note that the human eye can perceive even an offset error of one pixel. In an observation system capable of resolving an image with a resolution of 0.22 microns per pixel, even a quarter of a micron (250 nanometer) error can cause a perceptible error in positioning. . Since this system believes that the actual image position appears to be displayed in FIG. 3, if the actual positioning of FIG. 4 has not been determined, the image will always shift on the display. Will be (not aligned). As a result, the lack of proper image alignment results in image seams.

  The image scanning and capture methods described above can be used by those skilled in the art to image a region of interest of any size. As the various blocks of the image are captured, they are sent simultaneously to the compression subsystem and the matching subsystem. These subsystems are implemented in the control processor 20 or some other such processor. Each of these subsystems operates one or more separate execution units, such as threads or processes. This provides better system performance, stability and scalability. A separate process is used as an execution unit on an operating system that protects memory, such as Windows NT or any other UNIX-like system. In addition, a low overhead data sharing mechanism (eg, shared memory) can be used to store image data between different parts of the system.

  Stability is enhanced because it is unlikely that the entire system will go down if one of these subsystems fails. Scalability is enhanced because more processing units (eg, processors, distributed hosts, etc.) can be used in a parallel architecture.

  The compression subsystem may use various image compression systems that are intended to operatively compress the content data of the scanned image block. Suitable image compression systems are widely known in the art, such as JPEG, JPEG2000, TIFF, GeoTIFF, PNG. Alternatively, the block of images can be written to a disk or other storage device in its uncompressed format. The image is preferably stored in a format that allows the individual blocks to be retrieved individually.

  FIG. 10 shows an example of how an image block is stored in a row and column format that facilitates such direct retrieval. In the example of FIG. 10, the image block (1, 2) is the image area block starting from the lower left of the matrix, the row position is second (starting from modulo 0), and the column position is It is understood to represent the image area in the third position (again starting from modulo 0). Image compression and storage formats such as TIFF, GeoTIFF, JPEG, etc. allow such direct storage, as is widely known in the art.

  During compression, a lower resolution image can be captured and stored with a corresponding desired higher resolution image. Such multiple resolution capture and storage is known in the art as a “resolution pyramid” or “pyramiding”. This is because, for example, a wide field of view, which inevitably has a lower resolution, is often considered the desired initial field of view of the tissue slide. The clinician first looks at a wide area and then decides that only a limited portion of the entire image needs to be evaluated at high resolution. If an appropriate part of the entire image is available at a lower resolution, a larger size image (and consequently a larger file size) need not be transmitted.

  When storing higher resolutions, the area block here consisting of lower resolution images is considerably smaller. In this case, multiple region image blocks can be combined into a single image by allowing only one of the multiple images to contribute to the overlapping region. Since overlapping regions provide the same data in each image block, only one image block is needed to contribute data to the overlapping region. FIG. 11 shows an example in which a plurality of image blocks contributing to the overlapping region are deleted when four image blocks are combined to form one intermediate image M. In the example of FIG. 3, image blocks 501 and 504 have an overlap, whereas in the example of FIG. 11, image block 504 is the only one that contributes to the overlap to create a lower resolution image. Selected as the image block.

  Similarly, in the example of FIG. 3, all four image blocks contribute to the “four corners” overlap region X at the center where all four images converge. In the example of FIG. 11, in contrast, block 502 has been selected as the only contributing block to the central overlap region X ′. The selection of image blocks that are selected to contribute is determined using a number of methods. For example, there are methods such as arbitrarily selecting the block with the best focus or the best contrast, or simply selecting the first image block retrieved by software. Once the overlapping regions are removed, a single image can be created that removes the duplicated image that results from the overlap.

  As described above, image blocks sent to the compression subsystem are also sent to the matching subsystem. When the block of images arrives at the coincidence subsystem, it is evaluated for relative positioning. If a particular block exists immediately to the left of the current block (that is, if the current block is not the leftmost column), the normalized correlation coefficient is used exhaustively to the right of that block The relative positioning of the image is found such that the edge is matched to the left with respect to the left edge of the current block, resulting in the best match, i.e. minimizing seam appearance . Exhaustive normalized correlation is well known in the art and has been shown to be a reliable and robust method for image integration. Although computationally highly concentrated, normalized correlation (also referred to as NGC) is suitable for finding patterns in spatially transformed images.

  FIG. 8 is a flow diagram illustrating exemplary steps in an example method for applying such an exhaustive normalized correlation coefficient to two image blocks (eg, current block ( Match the left block (along the x-axis) or match the current block to the block just above (along the y-axis)). An image block (1M) that matches the left edge, such as image 501 in FIG. 9, is up to its start position (specified by coordinates CurX, CurY) relative to the current block, such as 502 in FIG. Within the range in the x direction, the search radius X (SearchRX) is moved sequentially in the left-right direction. Similarly, the matching image block is moved in the y direction (up and down) by the search radius Y (SearchRY), as shown in steps 810 and 815 of FIG.

For each of CurX and CurY, in step 825 of FIG. 8, a correlation algorithm is used to calculate a correlation coefficient. Initially, a maximum correlation (MC) value (which can have a value in the range of -1 to +1) is set. If the calculated correlation coefficient is greater than the value previously set for the maximum correlation (MC), then in step 830 the calculated value is returned as MC and the calculated x and y offset ( (Represented here as BestX and BestY) are returned as CurX and CurY, respectively. Furthermore, it should be noted that this is an iterative process and the returned values of CurX and CurY are the x and y search radii (SearchRX and SearchRY), as shown in steps 840 and 850. Is evaluated against.

  This iterative process is to ensure that all possible permutations are evaluated, and the resulting offset vector for the “synthesized whole” is It is to ensure that no deviation of any “open area” occurs when multiple blocks are combined. It is further noted that the possible size (area) of the matched image parts (overlapping regions) are considered equal. These inevitably define a rectangular part that has the same length and height, and is evaluated orderly within a defined positional boundary (a specific number of pixels in the x and y directions). .

Referring back to FIG. 5 , the resulting x and y offsets (BestX and BestY) along with the calculated correlation coefficient (MC in FIG. 8) are combined and expanded to an expanded horizontal offset vector (hOff and (Shown at 511 in FIG. 5). As those skilled in the art will appreciate by looking at FIG. 5, the horizontal offset vector hOff is the x and y offset along one edge of the image (here chosen as the horizontal edge). It is a vector sum of individual direction vectors. An offset vector and a correlation coefficient are related data values, and a particular MC value is associated with its corresponding offset vector.

In this example, the extended vertical offset vector, denoted vOff, causes block 503 to block 502 in relation to the “other” direction image edge (here vertical edge) to obtain optimal positioning. It represents a positioning vector that defines the direction and distance to be moved. It should be noted that the magnitudes of the expanded offset vectors hOff and vOff are not shown to scale in FIG. These extended offset vectors are then stored along with the corresponding correlation coefficients for later recall. If there is a block above the current block (ie if the current block is not in the top row), the lower edge of the upper block is listed in FIG. Is matched to the upper edge of the current block using the exhaustive normalized correlation coefficient by the same procedure. The top image is the new image to match ( IM ).

  In this example, the top image is block 503 in FIG. 9 and the current block remains the same block 502. The resulting x and y offset is recorded as an extended vertical offset vOff (512 in FIG. 5) along with the correlation coefficient. The vector vOff represents a positioning vector that defines the direction and distance by which the block 503 is moved relative to the block 502 in order to obtain optimal positioning, as in the case described above. By applying hOff and vOff to the original image positions of block 501 and block 503 in FIG. 9, respectively, the image blocks 501, 502 and 503 illustrated in FIG. Set) is obtained. Each extended offset consists of an offset vector represented by a pair of integers and a correlation coefficient represented by a floating number with a certain precision. In step 825 of the method 800, a correlation for a properly obtained matching area is calculated, where the size of the matching area remains constant for each pair of image blocks.

  After the entire scan is finished, each image block has an associated extended horizontal offset as well as an associated extended vertical offset. The exception to this is the image block in the top row and making up the leftmost column. Naturally, the top leftmost image block does not have an associated extended offset. The image block in the top row (except for the top leftmost image block described above) has only an extended vertical offset. These extended offsets can be in the same file as the associated metadata, along with the block of images, or in a separate metadata file in an XML file, binary file, or any other searchable file format. Memorized.

If the compression subsystem creates a multi-resolution image set, an extended offset for the lower resolution image is also created, and the calculated offset vector by a percentage equal to the ratio of the resolution (enlargement) values of the size it is saved by the scale down to Turkey. For example, if hOff has a size of 4 in the x direction and 6 in the y direction, the hOff for an image with 50% resolution is 2 in the x direction and 3 in the y direction. It is particularly beneficial to scale down higher resolution to lower resolution offsets. The reason is that the correlation coefficient (and offset) determined from the high resolution image is always more accurate than the correlation coefficient determined from the low resolution image. This is especially true when lower resolution images are created from higher resolution images.

  The user typically wants to see a virtual slide on the image viewer device. In accordance with the present invention, a block diagram of an image viewer suitable for use with the method described above is shown in FIG. The image viewer 900 of this example includes a processor 902 that operatively includes a compositing subsystem 904 that will be described in detail below. The image viewer further includes a display on which the image block is provided to the user. The image is typically displayed in a viewer window (formed by application software control) on the display. The image viewer 900 further includes a memory 908 coupled to the processor, which may be a non-volatile memory such as a disk drive or CD-ROM, or a volatile memory, Other memory suitable for storing blocks and other data described above may be used.

  The image viewer can be realized as a stand-alone device or as a client device in a network (Internet or Intranet). The image block and data can be obtained from a network source such as the Internet or Ethernet (registered trademark), or a storage (NAS) attached to the network. While viewing, the memory 908 is accessed by the viewer's processor 902, whether it is a disk or other memory. If it is desired to see an image, the viewer application is called. The viewer application is implemented as a software application such as compiled C ++, Java (registered trademark), Visual Basic, or other similar programs running on a PC-based operating system such as Windows or OS-X Is done. A synthesis subsystem that can be implemented as part of the viewer application program dynamically synthesizes the saved image as the user scans the entire virtual slide and the provided field of view changes. As discussed above, the synthesis subsystem can be implemented as firmware in “specific application” hardware instead of or in software.

  During the viewing process, the viewer application displays the field of view to the user. The image blocks required to represent the current field of view (typically 4 to 9 blocks are required to represent one field of view) correspond to these image blocks Searched with extended horizontal offset and extended vertical offset (and optionally cached in cache memory). To avoid submitting partial blocks and better process the data at the edges of the field of view, preferably padding is applied to provide at least one extra row or column around the four edges of the field of view. Here, the synthesis subsystem is called to position the retrieved image block in the viewer window. If the image block is from a lower resolution level created by merging multiple images and removing overlapping areas, the compositing subsystem is not called and the image is simply its Displayed in the original position.

  The synthesis subsystem sets a threshold on the correlation coefficient value to determine if there is a true match between the blocks. For example, if the correlation coefficient value is greater than 0.95, the corresponding offset positions the two blocks in a manner that is considered a true match. If the correlation coefficient is less than 0.95, the matching region is probably over white space and the correlation coefficient is set to -1. Connection criteria are defined as follows: Referring to FIG. 5, for any given block (such as 502), if a particular block truly matches the topmost block (503), these two blocks ( 502 and 503) are considered connected. If the block truly matches the left block (501), then these two blocks (502 and 501) are still considered connected.

  In one example of the present invention, the synthesis subsystem uses the connection criteria defined above and applies a standard coloring (labeling) algorithm known in the art to all blocks. The same color (label) is specified for a plurality of blocks connected to each other to form connected components. Accordingly, each connected component is composed of one or a plurality of blocks. Among the connected component blocks, one of the blocks is selected as the anchor block. For example, the top leftmost block (ie, the smallest block in lexicographic order) is selected as the anchor block and the remaining blocks are connected to that anchor block according to the present invention. It can be.

  In this regard, FIG. 6 illustrates a collection of two connected components, one of which includes four blocks 603, 602, 601 and 604 that are anchored to an anchor block 601 and the other is A block 606 is tethered to a block 605 including two blocks. Blocks with “A” (blocks 601 and 605) are anchor blocks. It should be noted here that in FIG. 6, not the block offset information but the block connection state is shown.

  The anchor block can be positioned at any location, but preferably its original location (the location returned when the block of the image is commanded to be captured, as shown in FIG. 3). ) Should be kept. As a result, the anchor block will have an effective offset of (0, 0) from the indicated position of its base, and thus an absolute position offset of x = 0 and y = 0. Any block in the same connected component can follow the anchor block through at least one connection path. For example, in the embodiment of FIG. 6, the connection path is established via connection paths from block 603 to anchor block 601, from 603 to 602, and from 602 to 601 (anchor A).

  Next, a position offset must be calculated for each retrieved non-anchor image block. The extended offset is accumulated along the path to the anchor block to determine the position offset for a given block. Adding the position offset to the anchor block offset determines the absolute position offset for the given block. For example, in FIG. 7, the extended vertical offset for block 704 is x = 4, y = 6, and correlation coefficient = 0.99. The expanded horizontal offset for block 702 is x = 1, y = -2, and the correlation coefficient = 0.98. Thus, the combined extended offset from paths 704 to 702 to 701 is a combination of values for each path. That is, x = 4 + 1 = 5 and y = 6-2 = 4. Then, the combined correlation coefficient = 0.99 * 0.98 = 0.702. The offsets x = 5 and y = 4 are position offsets, and a correlation coefficient of 0.9702 is used as a score for the path. The absolute position offset can be determined by adding the position offsets x = 5 and y = 4 to the offset of the anchor block offset (typically x = 0 and y = 0). That is, the absolute position offset is x = 5 + 0 = 5 and y = 4 + 0 = 4.

  Furthermore, there can be more than one distinct path to the anchor block. For example, as shown in FIG. 7, block 704 connects to anchor block 701 via a path 701 via 704 to 702 and a path 701 via 704 to 703. be able to. In order to determine which of the two paths is the optimal path, a correlation coefficient is multiplied along each path to obtain a value for that path. The path with the highest correlation value is selected as the optimal path. The position offset and the absolute position offset calculated along the optimal path are used, and the other paths are discarded.

  Next, the retrieved block of image 704 is drawn in the viewer window, which is offset from its original position by the calculated absolute position offset described above. When the process described above (such as calculating correlation values and determining position and absolute position offsets) is repeated as necessary for all blocks, a seamless (or almost seamless) image of the region of interest is obtained. Occurs in the viewer window.

  The user then asks for a new field of view to be displayed. The process described above (ie, requesting an image block, setting a threshold, applying connection criteria to the labeling algorithm to determine connected components, selecting an anchor block, and position offset along the optimal path And the absolute position offset is determined and rendered) is repeated and a new field of view is provided to the viewer and user.

As is apparent, this method represents a rendering scheme that is much simpler and less computationally expensive than in the prior art. Certain calculations are performed a priori (offset and correlation coefficient), as a function of seeing the dynamic synthesis, and occurs only as a function of the actual field of view provided to the user. Only when the field of view changes, the system of the present invention performs a compositing operation, which is only for new content that appears on the display, or a new image that must be used to contribute to the field of view. Only about blocks.

  The imaging and viewer system of FIGS. 1 and 12 is disclosed in the form that various functions are performed by discrete functional block diagrams. However, many of these functions can also be implemented in a configuration in which one or more functions of these blocks are implemented, for example, by one or more appropriately programmed processors. In addition, various imaging systems and methodologies have been described in different ways to clearly illustrate their novel features. However, as those skilled in the art will appreciate, many of the techniques and apparatus can be combined with each other or substituted for one by one without affecting the usefulness of the present invention. is there. It should be further noted that although the various motions and axes described above have been described using a convenient axis concept, such descriptions do not imply any particular direction or feature in space. Translation of the objective lens can be achieved by stage movement, turret movement, or movement of the microscope head itself. As long as the motion occurs in relation to the index position on the slide and is characterizable, measurable and repeatable, it is not important how the motion is achieved.

  Furthermore, the principles of the present invention have been described above in the context of steps represented in various flowcharts. The steps of the present invention need not be performed in the particular order described and need not be performed at the particular time described. What is needed to implement the present invention is the determination of the relative offset value of the region image block, the calculation of the correlation value for the matching region, and the dynamic provision of multiple region image blocks as the virtual slide field of view. Yes, the displayed image is only evaluated and adjusted to eliminate the deviation based on the field of view.

  Accordingly, the foregoing has only described the principles of the present invention. Those skilled in the art will be able to create many other configurations that implement the principles of the invention without departing from the spirit and scope of the invention. The embodiments described in this application and shown in the drawings are not meant to be limiting, but merely to illustrate the invention, the scope of which is defined solely by the appended claims.

FIG. 2 is a simplified, generally schematic block diagram of an imaging system used to acquire an image of a microscope slide according to the present invention. 4 is a flow diagram illustrating a method for scanning and capturing an image of a microscope slide, according to an embodiment of the present invention. FIG. 4 is an example of a region of interest of a microscope slide including four images arranged in a manner based on the position of the microscope slide and having overlapping edge regions, according to an embodiment of the present invention. FIG. 4 is an example of a region of interest of a microscope slide that includes four images that have overlapping edges that are aligned after being synthesized. FIG. 6 is a simplified block diagram illustrating an example of the position of an image block after calculating and applying an offset vector according to the present invention. FIG. 2 is a diagram generally illustrating a connection path between an image block and an anchor block in accordance with the principles of the present invention. 4 is another embodiment of a connection path connecting an image block to an anchor block according to the present invention. 4 is a flowchart illustrating an exemplary method for an exhaustive correlation search between two overlapping images according to the present invention. Similar to the embodiment of FIG. 4, it is an example of an additional region of interest illustrating three images with overlapping image boundary edges that are positioned and properly synthesized according to the requirements of the positioning system. It is an example of a matrix sequence for storing image blocks in a disk file and illustrates a method for indexing individual area blocks. FIG. 4 is an example of four image blocks combined with each other by removing overlapping regions, similar to the embodiment of FIG. FIG. 2 is a schematic block diagram of an image viewer device that has competed to implement the principles of the present invention.

Claims (17)

A method of viewing a digital virtual slide having seamless edge image quality characteristics in a microscope system of the type supported for viewing and image capture of a movable physical material sample slide in a microscope optical field of view, comprising:
Defining a plurality of digital image areas for a physical slide;
Scanning the physical slide according to the defined plurality of digital image regions to capture a plurality of region image blocks;
Storing the plurality of regional image blocks according to a position index metric associated with a respective position of each of the plurality of image blocks in the overall mosaic representation of the physical slide;
Determining a set of relative position offset values for the area image block and each of the next plurality of adjacent area image blocks, and a correlation coefficient for an overlapping portion between the image blocks based on the relative position offset value; When,
Viewing a field of view comprising less than all of a plurality of digital region images and including a portion of the region image block and a portion of at least one region image block of the next adjacent plurality of region image blocks Steps,
Synthesizing a composite image of the portion of a region image block and the portion of at least one region image block of the next adjacent plurality of region image blocks, the combining comprising: The relative position offset value relative to the region image block included in the field of view and limited to the portion of the region image block included in the field of view and the portion of the at least one region image block. And combining only using the correlation coefficient;
A method comprising the steps of:   2. The method of claim 1, wherein an additional region image block is displayed by moving the field of view of the display, and the step of combining is performed only with respect to the additional region image block. Feature method. The method of claim 1, wherein
Defining a plurality of regions for each physical slide, wherein each region of the plurality of regions has an area characterized by at least a width dimension. Defining the plurality of regions including the step of:
Scanning the physical slide according to the defined plurality of digital image areas to capture a plurality of area image blocks;
The step of capturing a plurality of region image blocks comprises:
Capturing an image of a first defined region in which the microscope is positioned at a first position relative to the physical slide;
Adjusting the position of the microscope to a second position characterized by the width dimension and a first overlap value;
The second definition wherein the second defined area and the first defined area share image data contained in the first overlap area defined by a first overlap value Capturing an image of the rendered region.   4. The method of claim 3, wherein the determining step is performed on image data included in an overlapping region between a plurality of adjacent regions. The method of claim 3, wherein
Defining a height dimension for each of the plurality of regions;
Adjusting the position of the microscope to a third position characterized by the height dimension and a second overlap value;
Capturing an image of a third defined region;
The third defined area and the first defined area share image data contained in a second overlap area defined by a second overlap value. A method characterized by that.   6. The method of claim 5, wherein the determining step of claim 3 is for image data contained in either a first overlap region or a second overlap region between adjacent regions. A method characterized in that it is performed.   6. The method of claim 5, wherein the determining step of claim 3 is performed on image data contained in both first and second overlapping regions between adjacent regions. Feature method.   4. The method of claim 3, wherein the set of relative position offset values and a correlation coefficient for a region image block and a next adjacent region image block determine, at least in part, a best location match. In order to be determined by performing a normalized correlation analysis on the region image block and the next adjacent region image block.   9. The method of claim 8, wherein the set of relative position offset values and correlation coefficients associated with each region image block are stored as a data set corresponding to the region image block. how to. The method of claim 9, wherein
Displaying a subset of the region image blocks on a display;
Positioning each region image block with respect to its next adjacent region image block according to the set of relative position offset values determined for the region image block; Combining the overlapping parts of the blocks;
The method of further comprising.   The method of claim 10, wherein the set of relative position offset values is applied to the next adjacent region image block in relation to an anchor block. In an observation system for imaging digital data representing a virtual microscope slide, a method of creating a composite image that defines a field of view of the observation system by seamlessly combining region images with each other,
Dividing a physical microscope slide into a matrix of a plurality of composite image regions, each image region having a row and column index position in the matrix; and
Scanning the physical microscope slide with a selected objective lens so that the field of view of the objective lens is equal to the size of one composite image area to capture a plurality of area image blocks;
Overlapping a current region image block and a specified portion of a next adjacent region image block such that the two images share image data in the specified portion;
A set of offset values for the current region image block and the next adjacent region image block is represented by an overlapping portion for the current region image block and an overlapping portion of the next adjacent region image block. Determining to further correlate to, further comprising generating a correlation coefficient for the overlapping portion based on the determined offset value ;
Viewing a field of view comprising less than all of the region image block and including a portion of the region image block;
Combining only region image blocks contained within the field of view to form a composite image, wherein the composite uses the offset value and the correlation coefficient to form the composite image ;
Displaying the composite image including the current region image block synthesized with the next adjacent region image block without considering an offset value of the region image block not currently displayed;
A method comprising the steps of:   The method of claim 12, wherein the determining step further comprises generating a correlation coefficient for the overlapping portion based on the determined offset value. 14. The method of claim 13, wherein
Evaluating which particular region image block of the plurality of region image blocks is required to represent the currently used field of view of the virtual microscope slide;
Searching the particular region image block among a plurality of the region image blocks along with the corresponding set of offset values and associated correlation coefficients;
Finding the retrieved region image block in a viewer according to its nominal index position in the matrix;
Comparing the correlation coefficient value and threshold value of the next adjacent image;
Defining a positive connection criterion between next adjacent images having a correlation coefficient value equal to or greater than the threshold;
Arranging the searched images relative to each other such that overlapping portions of next adjacent images define a seamless edge between the two next adjacent images;
The method of further comprising. The method of claim 14, wherein
Applying a connected component labeling algorithm to the plurality of region image blocks;
Assigning a common algorithmic label to a region image block having an associated positive connection criterion;
Forming an image block with a common algorithmic label into connected components;
The method of further comprising.   16. The method of claim 15, wherein one overlapping region is representative of digital image data contained in the overlapping region from either the current region image block or the next adjacent region image block. And the retrieved images are positioned relative to each other such that the selected overlap region defines a seamless edge between the two adjacent images. A microscope system of the type in which a physical material sample slide movable in the optical field of the microscope is supported for observation and image capture,
A microscope including a rotatable turret that holds a plurality of microscope objectives each having a corresponding magnification;
A sample stage that can be moved along orthogonal plane axes and is robot controlled;
Means for determining a current position of the stage relative to an initial index position along each of the axes;
A processor coupled to the stage for controlling stage movement such that the physical slide is scanned as digital data into a memory data storage device, the physical slide comprising a plurality of composite image blocks The processor, each image block having a corresponding row and column position in the matrix;
Each image block is characterized by a dimension in a first and second direction, and by the processor, the next adjacent image block is identified and a reproducible overlap is Scanned to occur between adjacent image blocks,
A correlation processor for determining an offset value along said first and second directions between next adjacent image blocks, said offset value determining said next adjacent image block from these images The correlation processor defining relative positioning vectors necessary to position each other such that overlapping portions of blocks appear as seamless edge regions;
A viewer for displaying a portion of the matrix of the composite image block as a partial image of the physical slide;
Image synthesizing means for providing a field of view consisting only of a part of the matrix of composite image blocks for viewing and for synthesizing the next adjacent image blocks with each other, the composition of the image comprising: And further comprising the image synthesizing means that is limited to image blocks within and uses the offset value and the relative positioning vector ;
The image compositing means further selects a subsequent image block when the user seeks a movement along the first and second directions to a new field of view without taking into account the offset value of the area image not currently displayed. A microscope system that operates to synthesize in the field of view.

Images