Tumor early screening method and system based on multi-modal medical image fusion

Abstract

The application discloses a tumor early screening method and system based on multi-modal medical image fusion, relates to the technical field of medical image analysis, and comprises the following steps: collecting a multi-modal medical image sequence formed in a continuous time period in a same suspected area, synchronously recording body position change information under a corresponding time scale, and establishing a unified spatial position identifier in original image data. The application establishes the unified spatial position identifier and depicts organ boundary moving tracks, identifies boundary overlapping areas generated in the fusion process, accurately positions gray interruption abnormal sections, and reduces the risk that a micro lesion is shielded. Meanwhile, through time range positioning and dislocation rearrangement processing, spatial micro movement compensation is combined with a respiratory rhythm, the real development of the abnormal sections is restored, and the accuracy and stability of tumor early screening are improved.

Description

Technical Field

[0001] This invention relates to the field of medical image analysis technology, specifically to a method and system for early tumor screening based on multimodal medical image fusion. Background Technology

[0002] Early tumor screening based on multimodal medical image fusion refers to the simultaneous acquisition of multiple imaging results, such as CT, MRI, and PET-CT, of the same suspected area when the tumor is still in its initial stage, characterized by small size, atypical morphology, or metabolic abnormalities. Utilizing computer vision technology, anatomical structure information, soft tissue contrast information, and metabolic function information from different modalities of images are automatically identified and features extracted, and precise registration and standardization are completed within a unified spatial coordinate system. Based on this, a multidimensional tumor feature atlas is constructed, incorporating structural, functional, and textural features, through multi-level feature expression and cross-modal deep fusion. Further combining clinical annotation data and pathological knowledge, suspicious lesion areas are automatically located, segmented, and classified as benign or malignant. This improves the accuracy of identifying small lesions in early stages, where traditional single-image methods are insufficient, by leveraging multi-source information complementarity and visual intelligent analysis capabilities, achieving an intelligent screening process with earlier detection and more accurate judgment.

[0003] The existing technology has the following shortcomings:

[0004] In existing technologies, during multimodal medical image acquisition, different modal scans are often completed at different times, patient positions may change slightly, respiratory amplitude may vary, and organ spatial positions may shift slightly. During subsequent image registration and fusion, insufficient registration accuracy can easily lead to local overlap or superposition of organ boundaries, causing tiny masses originally located at the organ's edge to be partially covered by images of adjacent organs. This superposition is often visually misinterpreted as an extension of normal tissue structure, and the system may easily classify it as background tissue during feature extraction, thus masking small early lesions and posing a risk of missed diagnosis.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for early tumor screening based on multimodal medical image fusion, so as to solve the problems in the background art mentioned above.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an early tumor screening method based on multimodal medical image fusion, comprising the following steps:

[0008] Collect multimodal medical image sequences of the same suspected area within a continuous time period, synchronously record the body position change information at the corresponding time scale, and establish a unified spatial location identifier in the original image data;

[0009] Using a unified spatial location identifier, we can analyze the spatial location changes of multimodal medical image sequences over a continuous time period, depict the movement trajectory of organ boundaries at different time scales, and form the results of organ contour movement.

[0010] Based on the organ contour movement results, the local overlapping area of ​​the multimodal medical image fusion results is scanned to identify boundary compression segments, extract grayscale continuity interruption areas, and identify abnormal segments with the risk of mass occlusion.

[0011] Multimodal medical image sequences at corresponding time scales were retrieved around the abnormal segments, and comparative analysis was conducted on the imaging differences under different body positions to pinpoint the time range in which the boundary misalignment occurred.

[0012] For the time range of the location, a misalignment rearrangement process is performed. During the fusion stage, adjacent time scale images are introduced for local replacement, and spatial micro-shift compensation is performed in combination with respiratory rhythm information. From the time dimension, the boundary coverage state is removed, and the true imaging of abnormal segments is restored.

[0013] Preferably, the steps for establishing a unified spatial location identifier are as follows:

[0014] A continuous time period acquisition plan is formulated around the preset anatomical positioning range. The multimodal medical image sequence is arranged in time window order, and time scale information is recorded. Body position and respiratory status data are recorded simultaneously. The time scale information, body position and respiratory status data are associated with the original image data formed under the corresponding time scale.

[0015] Using time scale information as the main index, anatomical reference positions are extracted from the original image data, a unified spatial reference coordinate starting point is constructed, and the original image data formed by different imaging methods at the same time scale are mapped to the unified spatial reference coordinate starting point.

[0016] Spatial location identification and coding are performed around a unified spatial reference coordinate starting point. Time scale information is embedded into the spatial coordinate system, and body posture data and respiratory status data are associated with the spatial coordinate system to form a unified spatial location identification system.

[0017] Spatial mapping and integration of multimodal medical image sequences within a continuous time period is performed, embedding time scale information, body posture data, and respiratory status data into a unified spatial location identification system to achieve spatial location comparison across time periods.

[0018] Preferably, the starting point of the unified spatial reference coordinate is an anatomical reference position that remains stable in imaging over a continuous time period, and the spatial coordinate definition is kept consistent in all time-scale images. Body position data and respiratory status data are embedded in the spatial coordinate system in the form of spatial displacement reference parameters and are synchronously associated with time-scale information. The unified spatial position identifier runs through all original image data over a continuous time period.

[0019] Preferably, the steps for analyzing spatial location changes in multimodal medical image sequences over a continuous time period using a unified spatial location identifier are as follows:

[0020] Multimodal medical image sequences within a continuous time period are mapped to a spatial coordinate frame corresponding to a unified spatial location identifier. Organ boundary information corresponding to each time scale is extracted in the suspected area, and spatial positioning is recorded according to the unified spatial location identifier to construct a time series boundary data set.

[0021] By comparing the spatial locations of organs at adjacent time scales in the time series boundary data set with a unified spatial location identifier, a scale-by-scale spatial displacement data sequence between time scales is generated.

[0022] By integrating the spatial displacement data corresponding to each time scale within a continuous time period, a trajectory expression structure combining the time dimension and the spatial displacement dimension is established, forming a dynamic movement map of the organ outline around the suspected area.

[0023] The results of the dynamic movement map of organ contours are processed and expressed in a unified manner, and the spatial position, time scale information and spatial displacement data of organ boundaries are uniformly output to form the organ contour movement results.

[0024] Preferably, the steps for scanning the local overlapping region of the multimodal medical image fusion results using organ contour movement results are as follows:

[0025] The organ contour movement results are mapped to a unified spatial location identification framework of the multimodal medical image fusion results, the projection area of ​​the organ boundary in the fusion results is extracted, and the boundary activity coverage zone corresponding to the organ contour movement trajectory is constructed.

[0026] Dense scanning of adjacent regions along the movement trajectory of organ contours is performed to identify boundary compression fragments, label boundary compression fragments, and associate them with time scale and spatial location;

[0027] A continuity analysis of the grayscale distribution in the fusion result is performed around the boundary compression segment to find regions where grayscale continuity is interrupted, and a spatial overlap analysis is performed with the boundary compression segment to identify potential occlusion regions.

[0028] By comparing the changes in grayscale continuity interruption areas across different time scales using comprehensive spatial displacement data, abnormal segments with the risk of tumor occlusion are marked, and the spatial coordinate range, time scale interval, and boundary compressed segment number are recorded.

[0029] Preferably, the steps for locating the time scale range corresponding to the abnormal segment using a unified spatial location identifier are as follows:

[0030] Extract the spatial coordinate range corresponding to the abnormal segment from the multimodal medical image fusion results, and use the spatial coordinate range to back-index the multimodal medical image sequence to retrieve all time-scale image data corresponding to the spatial coordinate range of the abnormal segment.

[0031] Synchronize and associate the time scale information with the body posture data recorded at the corresponding time scale, so that the image data at each time scale contains spatial coordinate information and body posture information, and arrange the development data step by step to form a continuous development sequence.

[0032] Spatial comparative analysis was conducted on the changes in the imaging of abnormal segments over a continuous time scale. The boundary position of the abnormal segment was compared with the results of organ contour movement to mark potential boundary misalignment nodes and locate the time range of boundary misalignment.

[0033] The time range of boundary misalignment occurrence is associated with the body posture data at the corresponding time scale to form a boundary misalignment occurrence time range result that includes the time start point, time end point, spatial coordinate range, and body posture information.

[0034] Preferably, the imaging data of abnormal segments at each time scale is compared point by point with the organ contour movement results to identify and mark the boundary compression segments; by spatially comparing with the displacement data in the organ contour movement results, the start and end time intervals of boundary misalignment are accurately located; the time intervals of boundary misalignment are recorded synchronously with the body position change information to ensure a direct correlation between the imaging status of abnormal segments at different time scales and body position changes.

[0035] Preferably, the misalignment rearrangement processing steps for abnormal segments within the time range of boundary misalignment occurrence are as follows:

[0036] After identifying the time range of the boundary misalignment, the multimodal medical image sequences are arranged according to the time scale, and the imaging status of the abnormal segment at each time scale is clarified, forming a basis for comparison of the state before and after the misalignment.

[0037] During the fusion phase, the abnormal segments are locally rearranged, the abnormal segments with boundary coverage are temporarily separated, and adjacent time scale images are introduced for local replacement to ensure that the replacement area is aligned with the original abnormal segment in the spatial coordinate system.

[0038] Spatial micro-shift compensation is performed on the replacement area by combining respiratory rhythm information. The position of the replacement area in the spatial coordinate system is adjusted according to the changes in respiratory rhythm, so that the replacement area and the adjacent area in the fused image form a continuous connection relationship in spatial position.

[0039] After completing local replacement and spatial micro-shift compensation, the abnormal section is fused and reconstructed as a whole, so that the abnormal section is unmasked in the time dimension and restored to true development.

[0040] Preferably, during the local replacement process, the unified spatial location identifier remains unchanged, and the time scale information marker is retained in the replacement area. Spatial micro-movement compensation is performed around the spatial coordinate range corresponding to the abnormal segment. The abnormal segment presents a boundary separation state and maintains continuous grayscale distribution during the fusion stage.

[0041] An early tumor screening system based on multimodal medical image fusion includes a unified spatial identifier construction module, a temporal displacement modeling module, a boundary overlap detection module, a temporal misalignment localization module, and a dynamic rearrangement compensation module.

[0042] The unified spatial identifier construction module collects multimodal medical image sequences formed in the same suspected area within a continuous time period, synchronously records the body position change information at the corresponding time scale, and establishes a unified spatial location identifier in the original image data.

[0043] The temporal displacement modeling module uses a unified spatial location identifier to analyze the spatial location changes of multimodal medical image sequences within a continuous time period, depicts the movement trajectory of organ boundaries at different time scales, and forms the result of organ contour movement.

[0044] The boundary overlap detection module scans the local overlapping area of ​​the multimodal medical image fusion results based on the organ contour movement results, identifies boundary compression segments, extracts grayscale continuity interruption areas, and identifies abnormal segments with the risk of mass occlusion.

[0045] The temporal misalignment localization module retrieves multimodal medical image sequences at corresponding time scales around the abnormal segment, compares and analyzes the imaging differences under different body positions, and locates the time range in which the boundary misalignment occurs.

[0046] The dynamic rearrangement compensation module performs misalignment rearrangement processing for the time range of the location. During the fusion stage, it introduces adjacent time scale images for local replacement and combines respiratory rhythm information for spatial micro-shift compensation. From the time dimension, it removes the boundary coverage state and restores the true imaging of abnormal segments.

[0047] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0048] This invention establishes a unified spatial location marker over a continuous time period and dynamically depicts the movement trajectory of organ boundaries, enabling multimodal medical images to form a correlated expression in both the temporal and spatial dimensions. This allows for the identification of boundary overlap regions caused by changes in body position and respiratory differences during the fusion stage. By precisely locating boundary compression fragments and areas of grayscale discontinuity, small lesions that are easily obscured gain independent imaging support, effectively reducing the risk of occlusion due to boundary overlap and improving the identification ability of small masses in peripheral areas.

[0049] This invention identifies the time range of boundary misalignment and introduces adjacent time-scale images for local replacement during the fusion stage. It also combines respiratory rhythm information to perform spatial micro-shift compensation, removing the coverage of abnormal segments from a temporal perspective and restoring them to their true imaging morphology. By performing temporal backtracking and rearrangement of the misalignment process, the fusion result maintains spatial continuity and imaging integrity, thereby improving the accuracy and stability of early tumor screening and enhancing the ability to detect early, small lesions. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0051] Figure 1 This is a flowchart of the method for early tumor screening based on multimodal medical image fusion according to the present invention.

[0052] Figure 2 This is a schematic diagram of the modules of the early tumor screening system based on multimodal medical image fusion of the present invention. Detailed Implementation

[0053] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0054] This invention provides, for example Figure 1 The illustrated method for early tumor screening based on multimodal medical image fusion includes the following steps:

[0055] Collect multimodal medical image sequences of the same suspected area within a continuous time period, synchronously record the body position change information at the corresponding time scale, and establish a unified spatial location identifier in the original image data;

[0056] To acquire multimodal medical image sequences over a continuous time period and establish a unified spatial location identifier, the specific implementation steps are as follows:

[0057] During the imaging preparation phase for the same suspected area, a continuous time-segment acquisition plan is formulated around the preset anatomical positioning range. The acquisition time windows of the multimodal medical image sequence are sequentially arranged so that different imaging methods form a set of image data with temporal continuity around the same suspected area on a continuous time axis. Before each imaging begins, the current time scale information is recorded, and the subject's posture and respiratory status data at that time scale are recorded simultaneously. The time scale information, posture data, and respiratory status data are correlated one-to-one with the original image data formed at the corresponding time scale, so that the multimodal medical image sequence formed within the continuous time period has a complete identification in the time dimension, thus providing a complete temporal reference basis for the subsequent establishment of a unified spatial location identification.

[0058] After acquiring multimodal medical image sequences over a continuous time period and establishing temporal correlations, spatial reference points are extracted from the original image data corresponding to each time scale, using time scale information as the primary index. Anatomical reference positions that can be repeatedly identified at each time scale are selected within the suspected area, and a unified spatial reference coordinate starting point is constructed around these anatomical reference positions. Based on this, the original image data formed by different imaging methods at the same time scale are mapped to this spatial reference coordinate starting point, so that the multimodal medical image sequences at each time scale form a preliminary unified relationship in the spatial dimension. At the same time, the body posture data and respiratory status data between each time scale are preserved, so that the unified spatial reference coordinate starting point not only has spatial positioning significance but also contains temporal scale correlation attributes, thus forming a spatial reference basis with temporal identification attributes.

[0059] After establishing a spatial reference base with time-scaled correlation attributes, spatial location identification encoding is performed on all raw image data within a continuous time period around this spatial reference base. The starting point of the spatial reference coordinates is taken as the origin, and the time scale information is embedded in the spatial coordinate system, so that the pixel position in each raw image data can be mapped to a unified spatial location identifier. In this process, body posture data and respiratory state data are converted into spatial displacement reference parameters and associated with the spatial coordinate system, so that the multimodal medical image sequences formed at different time scales are all within the same spatial location identifier framework, thereby establishing a unified spatial location identifier system that spans a continuous time period, so that the time dimension and the spatial dimension form a linkage relationship.

[0060] After the unified spatial location identification system is constructed, multimodal medical image sequences within a continuous time period are spatially mapped and integrated, enabling the original image data at each time scale to be queried and accessed within the unified spatial location identification system. Simultaneously, time scale information, body posture data, and respiratory status data are embedded as auxiliary identification information into the unified spatial location identification system, allowing subsequent spatial alignment processing to directly access the spatial location identification and body posture data at the corresponding time scale, achieving spatial location comparison across time periods. Through these processes, multimodal medical image sequences formed within a continuous time period possess unified spatial location identification at the original image data level, providing a stable reference for subsequent spatial location change analysis and organ contour movement trajectory depiction, and ensuring a traceable spatial correspondence between multimodal medical image sequences formed at different time scales.

[0061] Using a unified spatial location identifier, we can analyze the spatial location changes of multimodal medical image sequences over a continuous time period, depict the movement trajectory of organ boundaries at different time scales, and form the results of organ contour movement.

[0062] To achieve spatial location change analysis and organ boundary movement trajectory depiction of multimodal medical image sequences over a continuous time period, the specific implementation steps are as follows:

[0063] With a unified spatial location identification system already established, all multimodal medical image sequences within a continuous time period are mapped to the spatial coordinate framework corresponding to the unified spatial location identification, ensuring that the original image data generated at different time scales are all under the same spatial reference. Within this unified spatial location identification framework, with the suspected area as the analysis scope, organ boundary information corresponding to each time scale is extracted one by one. The organ boundary at each time scale is spatially located and recorded according to the unified spatial location identification, so that the spatial location data of organ boundaries at different time scales have a unified coordinate expression form. At the same time, the time scale information is associated with the corresponding organ boundary spatial location data and stored, thereby constructing a time series boundary data set with the unified spatial location identification as the core, laying the foundation for subsequent spatial location change analysis.

[0064] After forming a time-series boundary data set, the spatial position data of organ boundaries between adjacent time scales are compared scale by scale, based on a unified spatial location identifier. By continuously recording the changes in the boundary position of the same spatial coordinate region at different time scales, a spatial displacement data sequence between corresponding time scales is generated. In the process of generating the spatial displacement data sequence, the time scale order is used as the main clue for change analysis, so that the organ boundary position at each time scale can be directly correlated with the organ boundary position at the previous time scale. Thus, within the framework of a unified spatial location identifier, the dynamic change path of organ boundaries in continuous time periods is depicted, making the spatial position change analysis results continuous and traceable.

[0065] After generating the spatial displacement data sequence between adjacent time scales, the spatial displacement data corresponding to all time scales within a continuous time period are integrated, and a trajectory expression structure combining the time dimension and the spatial displacement dimension is established around a unified spatial location identifier. In this trajectory expression structure, the organ boundary position corresponding to each time scale is marked in coordinate form under the unified spatial location identifier, and connected in chronological order to form a continuous trajectory line, so that the movement trajectory of the organ boundary at different time scales is completely presented in the form of a spatial path. At the same time, the range of the movement trajectory is limited within the unified spatial location identifier framework, so that the trajectory always revolves around the suspected area, avoiding cross-regional interference, thereby forming a complete dynamic movement map of the organ outline.

[0066] After the dynamic movement map of organ contours is constructed, the map is processed to express the results. The spatial position of the organ boundary, the time scale information, and the spatial displacement data corresponding to each time scale are output in a unified manner to form the organ contour movement result. The organ contour movement result not only includes the specific spatial position of the organ boundary at different time scales, but also includes the spatial displacement relationship between adjacent time scales and the overall movement trajectory shape. This allows the multimodal medical image sequence in a continuous time period to present a complete dynamic change state of the organ boundary under a unified spatial position identification framework. This provides a clear trajectory reference for subsequent scanning of local overlapping areas of fused images and provides a direct spatial comparison basis for identifying boundary overlap phenomena caused by changes in body position and respiratory state.

[0067] Based on the organ contour movement results, the local overlapping area of ​​the multimodal medical image fusion results is scanned to identify boundary compression segments, extract grayscale continuity interruption areas, and identify abnormal segments with the risk of mass occlusion.

[0068] To achieve precise identification of locally overlapping regions in multimodal medical image fusion results and determine abnormal segments at risk of mass occlusion, the specific implementation steps are as follows:

[0069] Based on the established organ contour movement results, the organ contour movement results are mapped as a whole to the unified spatial location identification framework corresponding to the multimodal medical image fusion results, so that the organ contour movement trajectory and the multimodal medical image fusion results form a spatial overlay relationship under the same spatial coordinate system. Around the continuous time scale trajectory range recorded in the organ contour movement results, the projection area of ​​the organ boundary in the fusion results is extracted, and a boundary activity coverage zone corresponding to the organ contour movement trajectory is constructed. This allows the subsequent scanning process to focus on the spatial area covered by the organ contour movement trajectory, thereby limiting the scanning range of the local overlapping area and ensuring that the scanning process always revolves around the area where the organ boundary undergoes spatial changes.

[0070] After constructing the boundary activity coverage zone, the adjacent regions of the boundary are densely scanned segment by segment along the spatial distribution path of the organ contour movement trajectory in the multimodal medical image fusion results. The changes in the spatial spacing between adjacent organ boundaries in the fused image are continuously recorded. During this scanning process, the focus is on the segments marked with frequent spatial displacement changes in the organ contour movement results. The boundary morphology of these segments in the fused image is compared point by point to identify segments where the boundary spacing undergoes compression changes in continuous spatial positions. These compressed boundary segments are numbered and labeled according to a unified spatial position identifier, so that the compressed boundary segments form a one-to-one correspondence with the corresponding time scale and spatial position, thereby completing the systematic identification of compressed boundary segments.

[0071] After the boundary compression segments are labeled, a continuity analysis of the grayscale distribution in the multimodal medical image fusion results is performed around the spatial location of each boundary compression segment. Grayscale transition sequences are extracted along the normal direction of the boundary compression segments, and the spatial trend of grayscale value changes is continuously depicted to find spatial regions where grayscale continuity is interrupted within the boundary compression segment area. When the grayscale transition sequence shows discontinuous changes within the normal movement trajectory marked by the organ contour movement results, a spatial overlap analysis is performed on the grayscale continuity interruption region and the boundary compression segment. The overlapping part of the two under a unified spatial location identifier is extracted, so that the grayscale continuity interruption region and the boundary compression segment form a spatial correlation relationship, thereby identifying potential occlusion regions caused by boundary compression during the fusion process.

[0072] After extracting the grayscale continuity interruption region and establishing a spatial association with the boundary compressed segment, the spatial displacement data at the corresponding time scale in the organ contour movement results are integrated to compare the changes of the grayscale continuity interruption region at continuous time scales. This observes whether there is positional shift and morphological recovery phenomenon in the region at different time scales. When the grayscale continuity interruption region is covered within a certain time scale range but shows independent imaging at adjacent time scales, the spatial region is marked as an abnormal segment with the risk of mass occlusion. The spatial coordinate range, corresponding time scale interval, and boundary compressed segment number of the abnormal segment are uniformly recorded to form a complete abnormal segment identification result. This allows the potential mass occlusion region caused by organ boundary overlap in the multimodal medical image fusion results to be clearly located, thus providing an accurate spatial reference for subsequent time backtracking and misalignment rearrangement processing.

[0073] Multimodal medical image sequences at corresponding time scales were retrieved around the abnormal segments, and comparative analysis was conducted on the imaging differences under different body positions to pinpoint the time range in which the boundary misalignment occurred.

[0074] To achieve precise location of the time scale range corresponding to the abnormal segment, the specific implementation steps are as follows:

[0075] After identifying abnormal segments with the risk of tumor occlusion, the spatial coordinate range corresponding to the abnormal segment is extracted from the multimodal medical image fusion results using a unified spatial location identifier as a reference. Based on this spatial coordinate range, the multimodal medical image sequences formed within a continuous time period are indexed in reverse to retrieve all time-scale image data corresponding to the spatial coordinate range of the abnormal segment. During the retrieval process, the spatial coordinates of the abnormal segment under the unified spatial location identifier framework are used as query conditions, so that the original image data formed at different time scales can be presented at the same spatial coordinate location. This constructs a time-series image set around the abnormal segment, laying a spatial consistency foundation for subsequent comparative analysis of imaging differences under different body positions.

[0076] After forming a time-series image set around the anomalous segment, the time scale information is synchronously associated with the body posture data recorded at the corresponding time scale, so that the image data at each time scale includes not only spatial coordinate information but also body posture information. Around the fixed spatial position of the anomalous segment in a unified spatial location identification framework, the development morphology at different time scales is arranged scale by scale, so that the image performance of the anomalous segment in different body postures forms a continuous development sequence in chronological order. By observing the boundary morphology, grayscale distribution, and adjacent tissue relationships of the anomalous segment in different body postures at the same spatial coordinate position, the influence of different body postures on the development results can be intuitively presented in the time dimension, thus providing a complete temporal background for the formation process of boundary misalignment.

[0077] After constructing the imaging sequence, a spatial comparative analysis was performed on the imaging changes of abnormal segments across continuous time scales. The boundary position of the abnormal segment at each time scale was compared with the corresponding time scale boundary position recorded in the organ contour movement results, establishing a correspondence between the boundary appearance of the abnormal segment in the fused image and the boundary appearance in the original multimodal medical image sequence. When an abnormal segment exhibits boundary overlap or gray-scale occlusion in the fused image at a certain time scale, but shows separation at adjacent time scales, that time scale is marked as a potential boundary misalignment node. Through segment-by-segment comparison of continuous time scales, the time interval for the abnormal segment to transition from a boundary separation state to a boundary overlap state is clearly defined, thereby locating the start and end time scales of boundary misalignment, ensuring that the time range of boundary misalignment has a clear temporal boundary within a unified spatial location identification framework.

[0078] After locating the time range of boundary misalignment occurrence, this time range is associated with the body posture data at the corresponding time scale, so that the time range of boundary misalignment occurrence simultaneously includes spatial coordinate information and body posture information. By continuously arranging the imaging states of each time scale within this time range, the dynamic process of the abnormal segment changing from an unmasked state to a masked state can be clearly presented, thus forming a boundary misalignment occurrence time range result that includes time start, time end, spatial coordinate range, and body posture information. This result not only reflects the spatial performance of the abnormal segment in the multimodal medical image fusion result, but also reveals the influence path of different body postures on the formation of boundary misalignment, providing a clear temporal basis and spatial reference for subsequent misalignment rearrangement processing and spatial micro-shift compensation.

[0079] For the time range of the location, a misalignment rearrangement process is performed. During the fusion stage, adjacent time scale images are introduced for local replacement, and spatial micro-shift compensation is performed in combination with respiratory rhythm information. From the time dimension, the boundary coverage state is removed, and the true imaging of abnormal segments is restored.

[0080] To achieve misalignment rearrangement of abnormal segments within the time range of boundary misalignment occurrence and to remove the boundary coverage state during the fusion phase, the specific implementation steps are as follows:

[0081] After identifying the time range of boundary misalignment, multimodal medical image sequences corresponding to all time scales within this time range are extracted using a unified spatial location identifier as a reference. These sequences are then arranged scale by scale around the spatial coordinate range of the abnormal segment within the unified spatial location identifier framework, forming a continuous time series of images at each time scale within the time range of boundary misalignment. In this continuous time series, the imaging state of the abnormal segment at each time scale is clearly defined. Images at time scales showing boundary coverage are arranged side by side with adjacent images at time scales showing boundary separation, forming a comparative basis for the state before and after misalignment from a temporal perspective. This provides a complete time series basis for subsequent misalignment rearrangement processing.

[0082] After forming a continuous time series arrangement, a local rearrangement process is performed on the spatial coordinate range of the abnormal segment during the fusion stage. The abnormal segment region in the boundary coverage time scale image is temporarily separated from the fused image, and the development content of the corresponding spatial coordinate range in the adjacent time scale image is introduced for local replacement. During the local replacement process, the unified spatial location identifier is kept unchanged, so that the replacement area is completely aligned with the original abnormal segment in the spatial coordinate system. At the same time, the time scale information mark is retained, so that the local replacement behavior has a clear time source identifier. By introducing the adjacent time scale image for local replacement during the fusion stage, the coverage state originally formed by boundary misalignment is removed in spatial representation, thereby reconstructing the separation state development basis of the abnormal segment in the fused image.

[0083] After completing the local replacement process, the respiratory rhythm information recorded within the time range of the boundary misalignment occurrence is associated with the corresponding time-scale images. Using a unified spatial location identifier as a reference, spatial micro-shift compensation is performed on the replacement area. During the spatial micro-shift compensation process, based on the displacement trend of organ spatial position caused by respiratory rhythm changes at different time scales, the replacement area is slightly adjusted within the unified spatial location identifier framework, so that the replacement area and other areas in the current fused image form a continuous connection in spatial position. By converting respiratory rhythm information into a basis for spatial displacement adjustment, the local replacement area maintains consistency in both the time and spatial dimensions, thereby avoiding the generation of new boundary misalignment phenomena due to differences in respiratory rhythm.

[0084] After completing local replacement and spatial micro-shift compensation, the spatial coordinate range of the abnormal segment is fused and reconstructed as a whole, so that the replacement area and the surrounding tissue imaging form a continuous expression in terms of grayscale distribution and boundary transition. Under the unified spatial location identification framework, the imaging results of the abnormal segment after local replacement and spatial micro-shift compensation are integrated with the original fused image, so that the abnormal segment that was originally in a boundary coverage state is removed in the time dimension and restored to an unmasked true imaging state in the spatial coordinate system. Through the above misalignment rearrangement process, the boundary coverage state formed within the time range of the boundary misalignment is effectively removed, and the true imaging of the abnormal segment is restored in the fusion stage, thus providing a complete, continuous and unmasked image expression basis for subsequent early tumor screening.

[0085] This invention establishes a unified spatial location marker over a continuous time period and dynamically depicts the movement trajectory of organ boundaries, enabling multimodal medical images to form a correlated expression in both the temporal and spatial dimensions. This allows for the identification of boundary overlap regions caused by changes in body position and respiratory differences during the fusion stage. By precisely locating boundary compression fragments and areas of grayscale discontinuity, small lesions that are easily obscured gain independent imaging support, effectively reducing the risk of occlusion due to boundary overlap and improving the identification ability of small masses in peripheral areas.

[0086] This invention identifies the time range of boundary misalignment and introduces adjacent time-scale images for local replacement during the fusion stage. It also combines respiratory rhythm information to perform spatial micro-shift compensation, removing the coverage of abnormal segments from a temporal perspective and restoring them to their true imaging morphology. By performing temporal backtracking and rearrangement of the misalignment process, the fusion result maintains spatial continuity and imaging integrity, thereby improving the accuracy and stability of early tumor screening and enhancing the ability to detect early, small lesions.

[0087] This invention provides, for example Figure 2 The tumor early screening system based on multimodal medical image fusion shown includes a unified spatial identifier construction module, a temporal displacement modeling module, a boundary overlap detection module, a temporal misalignment localization module, and a dynamic rearrangement compensation module.

[0088] The unified spatial identifier construction module collects multimodal medical image sequences formed in the same suspected area within a continuous time period, synchronously records the body position change information at the corresponding time scale, and establishes a unified spatial location identifier in the original image data.

[0089] The temporal displacement modeling module uses a unified spatial location identifier to analyze the spatial location changes of multimodal medical image sequences within a continuous time period, depicts the movement trajectory of organ boundaries at different time scales, and forms the result of organ contour movement.

[0090] The boundary overlap detection module scans the local overlapping area of ​​the multimodal medical image fusion results based on the organ contour movement results, identifies boundary compression segments, extracts grayscale continuity interruption areas, and identifies abnormal segments with the risk of mass occlusion.

[0091] The temporal misalignment localization module retrieves multimodal medical image sequences at corresponding time scales around the abnormal segment, compares and analyzes the imaging differences under different body positions, and locates the time range in which the boundary misalignment occurs.

[0092] The dynamic rearrangement compensation module performs misalignment rearrangement processing for the time range of the location. During the fusion stage, it introduces adjacent time scale images for local replacement and combines respiratory rhythm information for spatial micro-shift compensation. From the time dimension, it removes the boundary coverage state and restores the true imaging of abnormal segments.

[0093] The tumor early screening method based on multimodal medical image fusion provided in this embodiment of the invention is implemented through the aforementioned tumor early screening system based on multimodal medical image fusion. For details of the specific methods and procedures of the tumor early screening system based on multimodal medical image fusion, please refer to the embodiments of the tumor early screening method based on multimodal medical image fusion described above, which will not be repeated here.

[0094] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for early tumor screening based on multimodal medical image fusion, characterized in that, Includes the following steps: Collect multimodal medical image sequences of the same suspected area within a continuous time period, synchronously record the body position change information at the corresponding time scale, and establish a unified spatial location identifier in the original image data; Using a unified spatial location identifier, we can analyze the spatial location changes of multimodal medical image sequences over a continuous time period, depict the movement trajectory of organ boundaries at different time scales, and form the results of organ contour movement. Based on the organ contour movement results, the local overlapping area of ​​the multimodal medical image fusion results is scanned to identify boundary compression segments, extract grayscale continuity interruption areas, and identify abnormal segments with the risk of mass occlusion. Multimodal medical image sequences at corresponding time scales were retrieved around the abnormal segments, and comparative analysis was conducted on the imaging differences under different body positions to pinpoint the time range in which the boundary misalignment occurred. For the time range of the location, a misalignment rearrangement process is performed. During the fusion stage, adjacent time scale images are introduced for local replacement, and spatial micro-shift compensation is performed in combination with respiratory rhythm information. From the time dimension, the boundary coverage state is removed, and the true imaging of abnormal segments is restored.

2. The method for early tumor screening based on multimodal medical image fusion according to claim 1, characterized in that, The steps for establishing a unified spatial location identifier are as follows: A continuous time period acquisition plan is formulated around the preset anatomical positioning range. The multimodal medical image sequence is arranged in time window order, and time scale information is recorded. Body position and respiratory status data are recorded simultaneously. The time scale information, body position and respiratory status data are associated with the original image data formed under the corresponding time scale. Using time scale information as the main index, anatomical reference positions are extracted from the original image data, a unified spatial reference coordinate starting point is constructed, and the original image data formed by different imaging methods at the same time scale are mapped to the unified spatial reference coordinate starting point. Spatial location identification and coding are performed around a unified spatial reference coordinate starting point. Time scale information is embedded into the spatial coordinate system, and body posture data and respiratory status data are associated with the spatial coordinate system to form a unified spatial location identification system. Spatial mapping and integration of multimodal medical image sequences over a continuous time period is performed, embedding time scale information, body posture data, and respiratory status data into a unified spatial location identification system.

3. The method for early tumor screening based on multimodal medical image fusion according to claim 2, characterized in that, The starting point of the unified spatial reference coordinate is an anatomical reference position that remains stable and visible over a continuous time period. The spatial coordinate definition is maintained consistently across all time-scale images. Body position data and respiratory status data are embedded into the spatial coordinate system as spatial displacement reference parameters and are synchronously associated with time-scale information. The unified spatial position identifier is used throughout all original image data over a continuous time period.

4. The method for early tumor screening based on multimodal medical image fusion according to claim 2, characterized in that, The steps for analyzing spatial location changes in multimodal medical image sequences over a continuous time period using a unified spatial location identifier are as follows: Multimodal medical image sequences within a continuous time period are mapped to a spatial coordinate frame corresponding to a unified spatial location identifier. Organ boundary information corresponding to each time scale is extracted in the suspected area, and spatial positioning is recorded according to the unified spatial location identifier to construct a time series boundary data set. By comparing the spatial locations of organs at adjacent time scales in the time series boundary data set with a unified spatial location identifier, a spatial displacement data sequence between time scales is generated. By integrating the spatial displacement data corresponding to each time scale within a continuous time period, a trajectory expression structure combining the time dimension and the spatial displacement dimension is established, forming a dynamic movement map of the organ outline around the suspected area. The results of the dynamic movement map of organ contours are processed and expressed in a unified manner, and the spatial position, time scale information and spatial displacement data of organ boundaries are uniformly output to form the organ contour movement results.

5. The method for early tumor screening based on multimodal medical image fusion according to claim 4, characterized in that, The steps for scanning local overlapping areas using organ contour movement results from multimodal medical image fusion are as follows: The organ contour movement results are mapped to a unified spatial location identification framework of the multimodal medical image fusion results, the projection area of ​​the organ boundary in the fusion results is extracted, and the boundary activity coverage zone corresponding to the organ contour movement trajectory is constructed. Dense scanning of adjacent regions along the movement trajectory of organ contours is performed to identify boundary compression fragments, label boundary compression fragments, and associate them with time scale and spatial location; A continuity analysis of the grayscale distribution in the fusion result is performed around the boundary compression segment to find regions where grayscale continuity is interrupted, and a spatial overlap analysis is performed with the boundary compression segment to identify potential occlusion regions. By comparing the changes in grayscale continuity interruption areas across different time scales using comprehensive spatial displacement data, abnormal segments with the risk of tumor occlusion are marked, and the spatial coordinate range, time scale interval, and boundary compressed segment number are recorded.

6. The method for early tumor screening based on multimodal medical image fusion according to claim 5, characterized in that, The steps for locating the time scale range corresponding to the anomaly segment using a unified spatial location identifier are as follows: Extract the spatial coordinate range corresponding to the abnormal segment from the multimodal medical image fusion results, and use the spatial coordinate range to back-index the multimodal medical image sequence to retrieve all time-scale image data corresponding to the spatial coordinate range of the abnormal segment. Synchronize and associate the time scale information with the body posture data recorded at the corresponding time scale, so that the image data at each time scale contains spatial coordinate information and body posture information, and arrange the development data step by step to form a continuous development sequence. Spatial comparative analysis was conducted on the changes in the imaging of abnormal segments over a continuous time scale. The boundary position of the abnormal segment was compared with the results of organ contour movement to mark potential boundary misalignment nodes and locate the time range of boundary misalignment. The time range of boundary misalignment occurrence is associated with the body posture data at the corresponding time scale to form the result of the time range of boundary misalignment occurrence.

7. The method for early tumor screening based on multimodal medical image fusion according to claim 6, characterized in that, The imaging data of abnormal segments at each time scale are compared point by point with the organ contour movement results to identify and mark the boundary compression segments; by spatially comparing with the displacement data in the organ contour movement results, the start and end time intervals of boundary misalignment are accurately located; the time intervals of boundary misalignment are recorded synchronously with the body position change information to ensure the direct correlation between the imaging status of abnormal segments at different time scales and body position changes.

8. The method for early tumor screening based on multimodal medical image fusion according to claim 6, characterized in that, The steps for handling misalignment rearrangement in abnormal sections within the time range of boundary misalignment occurrence are as follows: After identifying the time range of the boundary misalignment, the multimodal medical image sequences are arranged according to the time scale, and the imaging status of the abnormal segment at each time scale is clarified, forming a basis for comparison of the state before and after the misalignment. During the fusion phase, the abnormal segments are locally rearranged, the abnormal segments with boundary coverage are temporarily separated, and adjacent time scale images are introduced for local replacement to ensure that the replacement area is aligned with the original abnormal segment in the spatial coordinate system. Spatial micro-shift compensation is performed on the replacement area by combining respiratory rhythm information. The position of the replacement area in the spatial coordinate system is adjusted according to the changes in respiratory rhythm, so that the replacement area and the adjacent area in the fused image form a continuous connection relationship in spatial position. After completing local replacement and spatial micro-shift compensation, the abnormal section is fused and reconstructed as a whole, so that the abnormal section is unmasked in the time dimension and restored to true development.

9. The method for early tumor screening based on multimodal medical image fusion according to claim 8, characterized in that, During the local replacement process, the unified spatial location identifier remains unchanged, and the time scale information is retained in the replacement area. Spatial micro-movement compensation is performed around the spatial coordinate range corresponding to the abnormal segment. The abnormal segment presents a boundary separation state and maintains continuous grayscale distribution during the fusion stage.

10. A tumor early screening system based on multimodal medical image fusion, used to implement the tumor early screening method based on multimodal medical image fusion as described in any one of claims 1-9, characterized in that, It includes a unified spatial identifier construction module, a temporal displacement modeling module, a boundary superposition detection module, a temporal misalignment localization module, and a dynamic rearrangement compensation module: The unified spatial identifier construction module collects multimodal medical image sequences formed in the same suspected area within a continuous time period, synchronously records the body position change information at the corresponding time scale, and establishes a unified spatial location identifier in the original image data. The temporal displacement modeling module uses a unified spatial location identifier to analyze the spatial location changes of multimodal medical image sequences within a continuous time period, depicts the movement trajectory of organ boundaries at different time scales, and forms the result of organ contour movement. The boundary overlap detection module scans the local overlapping areas of the multimodal medical image fusion results based on the organ contour movement results, identifies boundary compression segments, extracts grayscale continuity interruption areas, and identifies abnormal segments with the risk of mass occlusion. The temporal misalignment localization module retrieves multimodal medical image sequences at corresponding time scales around the abnormal segment, compares and analyzes the imaging differences under different body positions, and locates the time range in which the boundary misalignment occurs. The dynamic rearrangement compensation module performs misalignment rearrangement processing for the time range of the location. During the fusion stage, it introduces adjacent time scale images for local replacement and combines respiratory rhythm information for spatial micro-shift compensation. From the time dimension, it removes the boundary coverage state and restores the true imaging of abnormal segments.

Images