Search

CN-121999995-A - Tumor early screening method and system based on multi-mode medical image fusion

CN121999995ACN 121999995 ACN121999995 ACN 121999995ACN-121999995-A

Abstract

The invention discloses a tumor early screening method and a tumor early screening system based on multi-mode medical image fusion, which relate to the technical field of medical image analysis and comprise the following steps: and acquiring a multi-mode medical image sequence formed by the same suspected region in a continuous time period, synchronously recording body position change information under a corresponding time scale, and establishing a unified spatial position mark in the original image data. The method and the device have the advantages that the uniform spatial position mark is established, the organ boundary moving track is drawn, the boundary overlapping area generated in the fusion process is identified, the gray level interruption abnormal section is accurately positioned, the shielding risk of the tiny focus is reduced, meanwhile, the spatial micrometric compensation is carried out by combining the breathing rhythm through time range positioning and dislocation rearrangement treatment, the real development of the abnormal section is recovered, and the accuracy and the stability of early screening of tumors are improved.

Inventors

  • Dou Peining
  • WU ZUOHUI

Assignees

  • 南方医科大学

Dates

Publication Date
20260508
Application Date
20260408

Claims (10)

  1. 1. The early tumor screening method based on the multi-mode medical image fusion is characterized by comprising the following steps of: Acquiring a multi-mode medical image sequence formed by the same suspected region in a continuous time period, synchronously recording body position change information under a corresponding time scale, and establishing a unified spatial position mark in original image data; Carrying out spatial position change analysis on the multi-mode medical image sequence in the continuous time period by utilizing the uniform spatial position mark, and drawing the moving track of the organ boundary under different time scales to form an organ contour moving result; Carrying out local overlapping region scanning on the multi-mode medical image fusion result according to the organ contour movement result, identifying boundary compression fragments, extracting a gray level continuity interruption region, and determining an abnormal section with a lump shielding risk; Taking a multi-mode medical image sequence under the corresponding time scale around the abnormal section, comparing and analyzing the development differences under different body position states, and positioning the time range of boundary dislocation; and (3) performing dislocation rearrangement processing aiming at the positioned time range, introducing adjacent time scale images to perform local replacement in a fusion stage, performing spatial micropipetting compensation by combining respiratory rhythm information, removing a boundary covering state from a time dimension, and recovering the real development of an abnormal section.
  2. 2. The method for early tumor screening based on multi-modality medical image fusion according to claim 1, wherein the unified spatial location identification establishment step is as follows: Making a continuous time period acquisition plan around a preset anatomical positioning range, sequentially arranging time windows of the multi-mode medical image sequence, recording time scale information, synchronously recording body position and posture data and breathing state data, and associating the time scale information, the body position and posture data and the breathing state data with original image data formed under corresponding time scales; The method comprises the steps of taking time scale information as a main index to extract an anatomical reference position in original image data, constructing a unified space reference coordinate starting point, and mapping the original image data formed by different imaging modes under the same time scale to the unified space reference coordinate starting point; Performing spatial position identification coding around a unified spatial reference coordinate starting point, embedding time scale information into a spatial coordinate system, and associating posture data and breathing state data to the spatial coordinate system to form a unified spatial position identification system; and carrying out space mapping integration on the multi-mode medical image sequences in the continuous time period, and embedding time scale information, posture data and breathing state data into a unified space position identification system.
  3. 3. The method for early tumor screening based on multi-modal medical image fusion according to claim 2, wherein the unified spatial reference coordinate starting point selects an anatomical reference position which is kept stable and developed in a continuous time period, and keeps a consistent spatial coordinate definition in all time scale images, the posture data and the breathing state data are embedded into a spatial coordinate system in the form of spatial displacement reference parameters and are synchronously associated with time scale information, and the unified spatial position identifier runs through all original image data in the continuous time period.
  4. 4. The method for early tumor screening based on multi-modal medical image fusion according to claim 2, wherein the step of analyzing spatial position changes of the multi-modal medical image sequence in the continuous time period by using the unified spatial position identifier is as follows: Mapping the multi-mode medical image sequence in the continuous time period to a space coordinate frame corresponding to the unified space position mark, extracting organ boundary information corresponding to each time scale in the suspected region, and carrying out space positioning record according to the unified space position mark to construct a time sequence boundary data set; performing scale-by-scale comparison on adjacent time scale organ boundary space positions in the time sequence boundary data set around the unified space position mark to generate a space displacement data sequence among time scales; Integrating the space displacement data corresponding to each time scale in the continuous time period, and establishing a track expression structure combining the time dimension and the space displacement dimension to form an organ contour dynamic moving map surrounding the suspected region; And carrying out result expression processing on the organ contour dynamic moving map, and uniformly outputting the spatial position of the organ boundary, time scale information and spatial displacement data to form an organ contour moving result.
  5. 5. The method for early tumor screening based on multi-modality medical image fusion according to claim 4, wherein the step of scanning the local overlapping area of the multi-modality medical image fusion result using the organ contour movement result is as follows: mapping the organ contour movement result into a unified spatial position identification frame of a multi-modal medical image fusion result, extracting a projection area of an organ boundary in the fusion result, and constructing a boundary activity coverage band corresponding to the organ contour movement track; Performing intensive scanning on the adjacent boundary areas along the moving track of the organ outline, identifying boundary compression segments, marking the boundary compression segments and correlating with time scales and space positions; Carrying out continuity analysis on gray distribution in the fusion result around the boundary compression segment, searching a gray continuity interruption region, carrying out space overlapping analysis on the boundary compression segment, and identifying a potential shielding region; And comparing the change condition of the gray continuity interrupt area in different time scales by comprehensive space displacement data, marking abnormal sections with bump shielding risks, and recording the space coordinate range, the time scale interval and the boundary compression fragment number.
  6. 6. The method for early tumor screening based on multi-modal medical image fusion according to claim 5, wherein the step of locating the time scale range corresponding to the abnormal section by using the unified spatial location identifier is as follows: Extracting a space coordinate range corresponding to the abnormal section from the multi-mode medical image fusion result, reversely indexing the multi-mode medical image sequence according to the space coordinate range, and calling all time scale image data corresponding to the space coordinate range of the abnormal section; Synchronously associating the time scale information with the posture data recorded under the corresponding time scale, so that the image data under each time scale contains space coordinate information and posture state information, and arranging the development data scale by scale to form a continuous development sequence; Performing space comparison analysis around the development change condition of the abnormal section in the continuous time scale, comparing the boundary position of the abnormal section with the organ contour movement result, marking potential boundary dislocation occurrence nodes, and positioning the time range of boundary dislocation; and carrying out association record on the boundary dislocation occurrence time range and the posture data under the corresponding time scale to form a boundary dislocation occurrence time range result.
  7. 7. The method for early tumor screening based on multi-modal medical image fusion according to claim 6 is characterized by comprising the steps of comparing abnormal section development data under each time scale with organ contour movement results point by point, identifying and marking boundary compression fragments, precisely positioning starting and ending time intervals when boundary dislocation occurs through spatial comparison with displacement data in the organ contour movement results, synchronously recording the time intervals when the boundary dislocation occurs and body position change information, and ensuring direct correlation between development states of the abnormal sections under different time scales and body position change.
  8. 8. The method for early tumor screening based on multi-modality medical image fusion according to claim 6, wherein the dislocation rearrangement processing step is performed for an abnormal section within a boundary dislocation occurrence time range as follows: After the boundary dislocation occurrence time range is positioned, arranging the multi-mode medical image sequence according to time scales, and defining the development state of the abnormal section under each time scale to form a comparison basis of the state before and after dislocation occurrence; Carrying out local rearrangement processing on the abnormal section in the fusion stage, temporarily separating the abnormal section in the boundary coverage state, introducing adjacent time scale images for local replacement, and ensuring that the replacement area is aligned with the original abnormal section in a space coordinate system; Carrying out space micro-motion compensation on the replacement area by combining the breathing rhythm information, and adjusting the position of the replacement area in a space coordinate system according to the breathing rhythm change so as to form a continuous connection relationship between the replacement area and an adjacent area in the fusion image in the space position; After the local replacement and the spatial micro-motion compensation are completed, the abnormal section is integrally fused and reconstructed, so that the abnormal section is in a shielding state in the time dimension and the real development is restored.
  9. 9. The method for early tumor screening based on multi-modal medical image fusion according to claim 8, wherein the unified spatial position identification is kept unchanged in the partial replacement process, and time scale information marks are kept in the replacement area, the spatial micropipes are performed around the spatial coordinate ranges corresponding to the abnormal sections, and the abnormal sections are in a boundary separation state in the fusion stage and keep gray level distribution continuous.
  10. 10. The tumor early screening system based on multi-mode medical image fusion, which is used for realizing the tumor early screening method based on multi-mode medical image fusion as set forth in any one of claims 1-9, is characterized by comprising a unified space identification construction module, a time sequence displacement modeling module, a boundary overlapping detection module, a time sequence dislocation positioning module and a dynamic rearrangement compensation module: The unified space identifier construction module is used for acquiring a multi-mode medical image sequence formed by the same suspected region in a continuous time period, synchronously recording body position change information under a corresponding time scale, and establishing a unified space position identifier in original image data; the time sequence displacement modeling module is used for carrying out space position change analysis on the multi-mode medical image sequences in the continuous time period by utilizing the unified space position mark, describing the moving track of the organ boundary under different time scales and forming an organ contour moving result; The boundary overlapping detection module is used for carrying out local overlapping region scanning on the multi-mode medical image fusion result according to the organ contour movement result, identifying boundary compression fragments, extracting a gray continuity interruption region and determining an abnormal section with a lump shielding risk; The time sequence dislocation positioning module is used for calling a multi-mode medical image sequence under the corresponding time scale around the abnormal section, comparing and analyzing the development differences under different body position states and positioning the time range of boundary dislocation; And the dynamic rearrangement compensation module is used for executing dislocation rearrangement processing aiming at the positioned time range, introducing adjacent time scale images to perform local replacement in the fusion stage, performing space micro-motion compensation by combining the breathing rhythm information, removing the boundary coverage state from the time dimension, and recovering the real development of the abnormal section.

Description

Tumor early screening method and system based on multi-mode medical image fusion Technical Field The invention relates to the technical field of medical image analysis, in particular to a tumor early screening method and system based on multi-mode medical image fusion. Background The early tumor screening based on multi-mode medical image fusion refers to that when tumors are still in the initial stage of volume microminiature, atypical morphology or metabolic abnormality, multiple imaging results such as CT, MRI, PET-CT and the like of the same suspected region are collected simultaneously, anatomical structure information, soft tissue contrast information and metabolic function information in different mode images are automatically identified and extracted by means of a computer vision technology, and accurate registration and standardization processing are completed under a unified space coordinate system. On the basis, a multidimensional tumor feature expression map comprising structural features, functional features and texture features is constructed through multi-level feature expression and cross-modal depth fusion. The method further combines clinical labeling data and pathological knowledge to automatically position, divide the outline and classify the benign and malignant risks of the suspicious lesion areas, so that the identification precision of the micro lesions is improved by means of multisource information complementation and visual intelligent analysis capability in the early stage that the traditional single image is difficult to identify, and an intelligent screening process of earlier discovery and more accurate judgment is realized. The prior art has the following defects: In the prior art, in the process of multi-mode medical image acquisition, as different mode scanning is often completed at different times, the body position of a patient slightly changes, the breathing amplitude is different, and the spatial position of an organ slightly shifts. In the subsequent image registration and fusion process, if the registration accuracy is insufficient, the phenomenon that the organ boundaries are partially overlapped or mutually overlapped easily occurs, so that the micro tumor originally positioned in the organ edge area is covered by the adjacent organ image part. The overlapping condition is often mistakenly considered as normal tissue structure extension in vision, and the system can easily judge the normal tissue structure extension as background tissue in the characteristic extraction stage, so that a tiny early focus is covered, and the risk of missed diagnosis exists. The above information disclosed in the background section is only for enhancement of understanding of the background of the disclosure and therefore it may include information that does not form the prior art that is already known to a person of ordinary skill in the art. Disclosure of Invention The invention aims to provide a tumor early screening method and system based on multi-mode medical image fusion, so as to solve the problems in the background technology. In order to achieve the above purpose, the invention provides the following technical scheme that the tumor early screening method based on multi-mode medical image fusion comprises the following steps: Acquiring a multi-mode medical image sequence formed by the same suspected region in a continuous time period, synchronously recording body position change information under a corresponding time scale, and establishing a unified spatial position mark in original image data; Carrying out spatial position change analysis on the multi-mode medical image sequence in the continuous time period by utilizing the uniform spatial position mark, and drawing the moving track of the organ boundary under different time scales to form an organ contour moving result; Carrying out local overlapping region scanning on the multi-mode medical image fusion result according to the organ contour movement result, identifying boundary compression fragments, extracting a gray level continuity interruption region, and determining an abnormal section with a lump shielding risk; Taking a multi-mode medical image sequence under the corresponding time scale around the abnormal section, comparing and analyzing the development differences under different body position states, and positioning the time range of boundary dislocation; and (3) performing dislocation rearrangement processing aiming at the positioned time range, introducing adjacent time scale images to perform local replacement in a fusion stage, performing spatial micropipetting compensation by combining respiratory rhythm information, removing a boundary covering state from a time dimension, and recovering the real development of an abnormal section. Preferably, the step of establishing the unified spatial location identifier is as follows: Making a continuous time period acquisition plan around a preset anatomical positionin