CN-121981927-A - Cross-layer annotation correction and real-time rendering method and application thereof

CN121981927ACN 121981927 ACN121981927 ACN 121981927ACN-121981927-A

Abstract

The invention provides a cross-layer annotation correction and real-time rendering method and application thereof, and belongs to the technical field of digital pathological image processing. The method solves the problems of label dislocation, editing distortion and large point set rendering and blocking existing in the switching of the multi-focal plane slice. The technical scheme is characterized by comprising the steps of determining interlayer offset parameters of each focal plane and storing the interlayer offset parameters in association with focal plane identifiers, anchoring an original focal plane and storing original geometric data when labels are created, performing cross-layer mapping according to relative offset parameters when a target focal plane is switched, keeping geometric semantics stable for different label types, calculating a rendering complexity index, performing hierarchical simplified rendering in a continuous interaction stage, recovering high-precision display after interaction stability, and reversely writing the target geometric data back to an original focal plane coordinate system after editing. The method realizes the accurate correction of the annotation cross-layer and the real-time smooth interaction, and effectively assists in pathological diagnosis.

Inventors

HUANG QIANG
WANG ZIHAN
CHEN BAICHUAN
CHEN SHIJIE
KUANG GUOTAO
JIN JIE

Assignees

深圳市生强科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260409

Claims (10)

1. A cross-layer annotation correction and real-time rendering method is characterized by comprising the following steps: determining interlayer offset parameters of each focal plane relative to a reference focal plane, and establishing association storage of the interlayer offset parameters and focal plane identifiers; When the label is created, recording the current focal plane identifier as an original focal plane identifier, and storing original geometric data by using a coordinate system of the current focal plane; When switching to a target focal plane, calculating a relative interlayer offset parameter according to interlayer offset parameters corresponding to the original focal plane identifier and the target focal plane identifier, and mapping the original geometric data into target geometric data according to the relative interlayer offset parameter; In the annotation rendering process, calculating a rendering complexity index, executing simplified rendering in a continuous interaction stage according to the rendering complexity index, and recovering high-precision rendering after interaction is stable; When the editing operation occurs on the target focal plane, the edited target geometric data is reversely written back to the original focal plane coordinate system according to the relative interlayer offset parameter, and the original geometric data is updated and persisted.
2. The cross-layer annotation correction and real-time rendering method of claim 1, wherein determining the interlayer offset parameter for each focal plane relative to a reference focal plane comprises: acquiring at least one of a mechanical side offset estimator and an image side offset estimator, wherein the mechanical side offset estimator is obtained by calculation based on mechanical motion records and splicing correction records of scanning equipment, and the image side offset estimator is obtained by calculation based on image registration of overlapping image blocks of the reference focal plane and a target focal plane; When the mechanical side deviation estimation amount and the image side deviation estimation amount exist at the same time, weighting and fusing the mechanical side deviation estimation amount and the image side deviation estimation amount according to a preset confidence level to obtain the interlayer deviation parameter; when only one of the estimators exists, the existing estimators are directly used as the interlayer offset parameter.
3. The method of claim 1, wherein the inter-layer offset parameters further comprise an affine compensation matrix, wherein the mapping the raw geometry data to target geometry data according to the relative inter-layer offset parameters further comprises: Calculating a relative affine compensation matrix according to the affine compensation matrix of the original focal plane and the affine compensation matrix of the target focal plane; And carrying out affine transformation on the original geometric data by combining the relative interlayer offset parameter and the relative affine compensation matrix to generate the target geometric data.
4. The cross-layer annotation correction and real-time rendering method of claim 1, wherein the original geometry data is mapped to target geometry data according to the relative inter-layer offset parameters, and corresponding geometry semantic preserving mapping is performed for different types of annotations: For region type labels containing rectangles or ellipses, performing translation transformation on the center point, and keeping the width, height and rotation angle unchanged; for line type marking comprising a measuring rule line, performing transformation on a starting point and an ending point, updating a direction angle, storing an original physical measured value as an authoritative semantic attribute, and preferentially displaying the original physical measured value when displaying; For point set labeling including painting brushes or polygons, the point sequence and the closing relation are kept unchanged, and transformation is performed point by point on each control point.
5. The cross-layer annotation correction and real-time rendering method of claim 1, wherein calculating a rendering complexity index, performing simplified rendering in successive interaction stages in accordance with the rendering complexity index, comprises: Calculating the rendering complexity index according to the control point number of the current annotation, the current zoom level, the control point proportion in the visible region, the interactive triggering times in unit time and the continuous interactive state quantity; when the rendering complexity index is in a first preset interval, directly drawing high-precision target geometric data; when the rendering complexity index is in a second preset interval, generating a simplified expression by adopting a path simplification or sampling algorithm to draw; And when the rendering complexity index is in a third preset interval, only drawing a key skeleton, a bounding box or a visible region segment.
6. The cross-layer annotation correction and real-time rendering method of claim 5, wherein generating the reduced representation using a path reduction or sampling algorithm comprises: the method comprises the steps of performing sparsification processing on a point set by adopting a Douglas-Peucker algorithm or a fixed step sampling algorithm; the reduction tolerance or sampling step size is dynamically adjusted according to the current zoom level.
7. The method of cross-layer annotation correction and real-time rendering of claim 1, wherein reversely writing back the edited target geometry data to the original focal plane coordinate system according to the relative interlayer offset parameter, updating the original geometry data and persisting, comprising: generating an editing increment record, and back-calculating the edited target geometric data back to the original focal plane coordinate system to obtain new original geometric data; performing consistency check on the new original geometric data, wherein the consistency check comprises calculating the cross ratio of regional labels or round-trip mapping root mean square error of point set labels; if the consistency check is not passed, the original geometric data of the old version is reserved, the editing result is temporarily stored, and manual confirmation is requested.
8. The cross-layer annotation correction and real-time rendering method of claim 1, further comprising: If the target focal plane is switched to the third focal plane after editing without waiting for interaction to finish, the latest editing increment record in the to-be-written-back queue is read, and edited target geometric data is directly converted to the third focal plane for temporary display according to the relative offset parameters between the current target focal plane and the third focal plane.
9. A cross-layer annotation correction and real-time rendering device, comprising: The offset parameter processing module is used for determining interlayer offset parameters of each focal plane relative to a reference focal plane and establishing association storage of the interlayer offset parameters and focal plane identifiers; the labeling attribution recording module is used for recording the current focal plane identifier as an original focal plane identifier when the labeling is created, and storing original geometric data by using a coordinate system of the current focal plane; The cross-layer mapping module is used for calculating relative interlayer offset parameters according to interlayer offset parameters corresponding to the original focal plane identification and the target focal plane identification when switching to a target focal plane, and mapping the original geometric data into target geometric data according to the relative interlayer offset parameters; The hierarchical rendering module is used for calculating a rendering complexity index in the annotation rendering process, performing simplified rendering in a continuous interaction stage according to the rendering complexity index, and recovering high-precision rendering after interaction is stable; and the reverse write-back module is used for reversely writing back the edited target geometric data to the original focal plane coordinate system according to the relative interlayer offset parameter when the editing operation occurs on the target focal plane, updating the original geometric data and lasting the original geometric data.
10. A readable storage medium, characterized in that the readable storage medium has stored therein a computer program comprising program code for controlling a process to execute a process comprising the cross-layer annotation correction and real-time rendering method according to any of claims 1 to 8.

Description

Cross-layer annotation correction and real-time rendering method and application thereof Technical Field The invention relates to the technical field of digital pathological image processing, in particular to a multi-focal plane digital pathological section-oriented cross-layer annotation correction and real-time rendering method and application thereof. Background Multiple focal plane digital pathological slices typically acquire multiple images of different focal depths at the scanning device side for the reader to switch views along the focal depth direction. Due to the actual errors in the scanning stage during mechanical motion, optical path switching, focus compensation and image stitching, a stable but not negligible lateral offset is typically formed between the different focal planes. When processing such cross-slice labels, the existing common scheme mainly comprises that the labels are moved integrally only by performing one-time experience translation correction on a display end, or labels of different focal planes are physically isolated from each other and are not displayed or inhibited from being edited during switching, or only simple object copying and local coordinate conversion are performed. These conventional methods do not generate and multiplex interlayer offset parameters at the bottom layer, do not build a data model with the original focal plane as a unified authoritative reference, and lack a reverse write-back closed loop for editing operations and a hierarchical rendering control mechanism for dealing with complex annotation high-frequency interactions. Disclosure of Invention The embodiment of the invention provides a cross-layer annotation correction and real-time rendering method and application thereof, aiming at the problems that the attribution consistency and geometric semantics of annotations cannot be maintained during multi-focal plane switching in the prior art, accumulated drift errors are easy to generate during cross-layer repeated editing, real-time rendering of large point set annotations during dynamic interaction is serious in clamping and the like. The core technology of the invention mainly comprises the steps of constructing a data model with an original focal plane as an absolute anchor, realizing cross-layer geometric semantic retention mapping and persistent write-back of an editing result by utilizing a positive inverse transformation closed loop, and coupling a dynamic hierarchical rendering mechanism based on rendering complexity to balance display precision and interaction delay. In a first aspect, the present invention provides a multi-focal plane digital pathological section-oriented cross-layer labeling correction and real-time rendering method, the method comprising the steps of: determining interlayer offset parameters of each focal plane relative to a reference focal plane, and establishing association storage of the interlayer offset parameters and focal plane identifiers; When the label is created, recording the current focal plane identifier as an original focal plane identifier, and storing original geometric data by using a coordinate system of the current focal plane; when switching to the target focal plane, calculating a relative interlayer offset parameter according to interlayer offset parameters corresponding to the original focal plane identification and the target focal plane identification, and mapping the original geometric data into target geometric data according to the relative interlayer offset parameter; In the annotation rendering process, calculating a rendering complexity index, executing simplified rendering in a continuous interaction stage according to the rendering complexity index, and recovering high-precision rendering after interaction is stable; when the editing operation occurs on the target focal plane, the edited target geometric data is reversely written back to the original focal plane coordinate system according to the relative interlayer offset parameter, and the original geometric data is updated and persisted. Further, determining an interlayer offset parameter for each focal plane relative to a reference focal plane includes: acquiring at least one of a mechanical side deviation estimator and an image side deviation estimator, wherein the mechanical side deviation estimator is obtained by calculation based on mechanical motion records and splicing correction records of scanning equipment, and the image side deviation estimator is obtained by image registration calculation on overlapping image blocks of a reference focal plane and a target focal plane; When the mechanical side deviation estimated quantity and the image side deviation estimated quantity exist at the same time, weighting and fusing the two estimated quantities according to the preset confidence coefficient to obtain an interlayer deviation parameter, and when only one estimated quantity exists, directly taking the existing estimated quantit