CN-121986355-A - Determining image registration quality in an automated tissue dissection procedure

CN121986355ACN 121986355 ACN121986355 ACN 121986355ACN-121986355-A

Abstract

In one aspect, the invention relates to a method of quality control in an image registration process, the results of which are used in an automated dissection process of dissecting tissue from an anatomical slide on which tissue slices cut from a tissue sample are placed. The method (10) comprises the steps of: obtaining (11) a first image of a reference slide containing a reference sample cut from the same tissue sample; Obtaining (12) a second image of the anatomical slide; Registering (20) the first image onto the second image, wherein a transformation matrix is calculated as part of the image registration process to produce a best fit between the first image and the second image; calculating (30) a first quality measure of the image registration process based on the degree of overlap between the first image and the second image; calculating (40) a second quality measure of the image registration process based on parameters indicative of the skewness of the transformation matrix and the asymmetric scaling; Comparing (50) each value of the calculated first and second quality metrics with a corresponding threshold to determine whether each quality metric is met; If either the first or second quality metric is not satisfied, determining the image registration quality is unsatisfactory, and reporting (60) the determination to a user.

Inventors

S. K. Visser
R. J. Gwent
R. Weinberg-Friedel
H. Van Wayne harden

Assignees

XYALL私人有限公司

Dates

Publication Date: 20260505
Application Date: 20240725
Priority Date: 20230803

Claims (15)

1. A method (10) of quality control in an image registration process, the method comprising the steps of: obtaining (11) a first image (305) of a reference slide (1), the reference slide (1) comprising a reference sample (5) cut from a tissue sample; Obtaining (12) a second image (310) of an anatomical slide (130), the anatomical slide (130) comprising tissue slices cut from the same sample; registering (20) the first image onto the second image, wherein in sub-step (23) a transformation matrix is calculated ) Thereby producing a best fit between the first image and the second image; Calculating (30) a first quality measure of the image registration process based on a degree of overlap between the first image and the second image; Calculating (40) a second quality metric of the image registration process based on parameters indicative of the degree of skewness and scaling of the transformation matrix; comparing (50) each value of the calculated first and second quality metrics with a corresponding threshold to determine whether each quality metric is met; If either the first quality measure or the second quality measure is not satisfied, determining the image registration quality is unsatisfactory, and reporting (60) the determination to a user.
2. The method of claim 1, wherein the first quality metric is an intersection ratio (IoU) of a first image (305) and a second image (310), and wherein the method further comprises the sub-steps of: Segmenting the first image from the second image to identify tissue pixels in each image; Calculating the number of overlapping tissue pixels in two images And calculates the total number of tissue pixels in the combined two images ; Wherein the step (30) of calculating the first quality metric includes calculating IoU as Divided by 。
3. The method of claim 2, wherein the calculated IoU value is compared to a IoU threshold, and wherein the method comprises the further step of: A IoU threshold is set based on a polynomial function that depends on a tissue area, wherein the tissue area is obtained from a number of identified tissue pixels in the first image.
4. The method of any preceding claim, wherein the step of calculating (40) the second quality metric comprises decomposing a subset of the transformation matrix into a first matrix representing rotation and a second matrix representing skewness and scaling, the subset comprising coefficients defining rotation, scaling and skewness, and wherein the parameter of the calculated second quality metric is a condition number of the second matrix.
5. The method according to claim 4, comprising a further step (70) of calculating a third quality measure of the image registration process, wherein: The third quality metric is a parameter indicative of symmetric scaling of the second matrix; assuming that the skewness is negligible, step (70) includes computing a 2-norm of the second matrix, and If the calculated 2-norm value is not within the predetermined lower and upper thresholds, then determining the quality of the image registration process is unsatisfactory.
6. The method according to any preceding claim, comprising a further step (80) of calculating a fourth quality measure of the image registration process, wherein the fourth measure is a quantized uncertainty parameter providing a measure of spatial accuracy of image alignment in the registration process (20).
7. The method of claim 6, further comprising the step of: Marking the boundary of a region of interest (ROI) to be dissected on the first image, and Using the transformation matrix calculated in sub-step (23) of image registration ) Transferring the ROI annotation to the second image, Wherein the translated ROI-annotated coordinates are used to control an anatomical tool (120) in an automated anatomical process, and wherein: the calculated value of the adventitious parameter is compared to a threshold value derived from typical accuracy of an anatomical tool used in the automated anatomical procedure, and wherein the quality of the image registration is determined to be unsatisfactory if the adventitious parameter exceeds the threshold value.
8. The method according to claim 6 or 7 when dependent on claim 2, the method further comprising determining an average diameter of the tissue object detected during the sub-step of identifying tissue pixels in the second image, and wherein the step of calculating (80) the quantified uncertainty parameter comprises multiplying the calculated value IoU by the determined average diameter.
9. The method according to claim 6 or 7, wherein the step of image registration (20) comprises detecting (21) pairs of features in the first image and the second image, and calculating (22) an individual transformation matrix of each corresponding pair of features ) And wherein calculating (80) the indefinite parameter comprises: Identifying key points for each feature in the first image (305) ) Each feature in the first image (305) has been detected and matched to a corresponding feature in a second image (310); using the calculated transformation matrix of the image registration process ) And using an individual transformation matrix associated with the feature ) For each identified key point ) Transform and Based on Determining local spatial differences between two transforms 。
10. The method of claim 9, wherein the uncertainty parameter is calculated based on an average of the determined local spatial differences.
11. A method according to claim 9 when dependent on claim 7, the method further comprising the steps of: Each calculated spatial difference Mapped onto the second image, and Each difference is set Points marked by the ROI on the first image (305) are associated to calculate local values of the indefinite parameters of all marked points.
12. The method according to any one of claims 6 to 11, further comprising the step of: Assigning a quality score to the calculated value of each quality metric, and Based on each assigned quality score, an overall quality rating of the image registration process is calculated.
13. A quality control system (90) for an image registration process, the system comprising an imaging module (150) and a display (145), the imaging module (150) having an image sensor (155) and a processor (153), the processor (153) being programmed with an image registration algorithm and image processing software, wherein the processor (153) is further programmed with instructions that cause the system to perform the steps of the method according to any preceding claim.
14. An apparatus for automated dissection, the apparatus comprising the quality control system (90) of claim 13, the apparatus (100) further comprising a movable dissection tool (110) and a controller (140) for controlling movement of the tool based on coordinates received from the imaging module (150), wherein the received coordinates indicate a marked boundary of a region of interest (ROI) that has been transferred from a first image (305) of a reference slide (1) to a second image of an anatomical slide (130).
15. The apparatus of claim 14, wherein the processor (153) is programmed for performing the method of claim 12 and is further programmed for evaluating the quality of the achieved anatomical result using the calculated overall quality rating of the image registration process.

Description

Determining image registration quality in an automated tissue dissection procedure Technical Field The present invention relates to digital pathology, and more particularly to a method and system for quality control in an automated anatomy process to determine if reliable automated anatomy can be achieved. Background For molecular testing purposes in oncology, multiple portions of tissue sections are dissected from glass slides. In general, tissue is dissected from several slides of a single instance to have sufficient material for molecular testing. There is an increasing demand for automated anatomic processes, including the application of digital pathology techniques. A region of interest (ROI) from which tissue is to be collected may be indicated by a pathologist on a reference image, which is captured for a particular tissue slice. The pathologist draws lines for identifying the ROI boundaries, which are typically referred to as labeling (registration). An anatomical image of each slide containing another tissue slice from the same sample to be dissected is also captured. During image registration, a reference image is mapped onto each anatomical image, wherein the annotations of the ROI boundaries are also transferred to the anatomical image. After image registration, the dissection tool is controlled based on the transferred labeled coordinates to mechanically remove biological material within the labeled boundary from each dissection slide. An example of a method for determining tissue anatomical regions in a series of pathology slides is disclosed in WO 2021/001564. First and second annotations are obtained for corresponding first and second reference images in the series, the first and second reference images being separated by a number of intermediate images. Then, based on bi-directional image registration, annotations are generated for intermediate images between the reference images, and then both annotations are transferred to each intermediate image, which are then combined to obtain a combined annotation for each intermediate image. The method comprises determining a first set of registration parameters for registering the first reference image onto the intermediate image and a second set of registration parameters for registering the second image onto the intermediate image, wherein the first set of registration parameters and the second set of registration parameters are used to propagate the first annotation and the second annotation, respectively, onto the intermediate image. The method may further include determining a quality of fit of the annotations in the intermediate image and prompting the user to provide a third annotation in a third image from the series of intermediate images between the first image and the second image if the quality of fit does not meet the quality of fit criterion. The quality of fit may be determined based on the shape differences or surface area overlaps of the propagated first annotation and the propagated second annotation, or based on the magnitude of deformation of the propagated first annotation and/or the magnitude of deformation of the propagated second annotation. However, it may be that the quality of the fit of the labels appears to be satisfactory, while the quality of the image registration is unsatisfactory. Thus, the result of the determination may produce false positives, which, of course, are undesirable. As will be appreciated, for the quality of the anatomical results achieved, it is important to remove the tissue of interest from the anatomical slide. The overall anatomical result may be affected by several parameters associated with the equipment used in the automated anatomical procedure. For example, misalignment between the imaging module and the dissection tool (misalignment) and the accuracy of movement of the dissection tool can affect the quality achieved. However, even if there is no misalignment and the accuracy of movement is accurate, the anatomical results achieved can be unsatisfactory if the transfer of labels to the anatomical slide is not accurate due to insufficient quality of the image registration process. There is still room for improvement. Disclosure of Invention In a first aspect, the invention relates to a method of quality control in an image registration process, the results of which are used in an automated dissection process in which tissue is dissected from an anatomical slide on which tissue slices cut from a tissue sample are placed. The method comprises the following steps: obtaining a first image of a reference slide, the reference slide comprising a reference sample cut from the same tissue sample; Obtaining a second image of the anatomical slide; registering the first image onto the second image, wherein a transformation matrix is calculated as part of an image registration process, thereby producing a best fit between the first image and the second image; Calculating a first quality metric (metric) o