4D-ct or 4d-cta cardiac chamber volume curve construction and cardiac function evaluation method

Abstract

The application discloses a 4D-CT or 4D-CTA heart cavity volume curve construction and heart function evaluation method, comprising the following steps: step one: acquiring 4D-CT or 4D-CTA time sequence images of the same subject, arranging phases according to acquisition time, and constructing a standardized cardiac cycle coordinate system; step two: performing heart cavity segmentation to obtain initial segmentation results of the heart cavity in each phase; step three: forming a unified time sequence heart cavity object through propagation, fusion and consistency constraint; step four: constructing a full-cycle volume curve; step five: identifying key phases; step six: calculating heart function parameters and performing quality control closed loop processing, and outputting heart function evaluation results. The application improves the continuity of heart cavity time sequence analysis, the stability of key phase identification and the reliability of heart function evaluation results.

Description

Technical Field

[0001] This invention relates to the fields of medical image processing, temporal medical image analysis, quantitative assessment of cardiac function, and computer-aided assessment, and particularly to a method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA. The method can be implemented using a software system, a medical image post-processing workstation, a server, a terminal device, or a storage medium storing program instructions. Background Technology

[0002] With the increasing demand for precise assessment of cardiovascular diseases and the continuous development of intelligent medical image analysis technology, the quantitative analysis of central cavity motion processes in 4D-CT or 4D-CTA time-series images and its auxiliary cardiac function assessment techniques have received widespread attention. Existing cardiac function assessment schemes mainly rely on phase-by-phase segmentation, phase-by-phase volume statistics, and extreme value extraction methods for volume measurement and parameter calculation.

[0003] However, in practical applications, the acquired 4D-CT and 4D-CTA multiphase images are prone to phase loss, phase duplication, uneven phase density, and inconsistent temporal organization under different equipment, different reconstruction protocols, and different cardiac cycles, leading to difficulties in phase alignment and instability in the basis of full-cycle analysis.

[0004] Meanwhile, different phases are often treated in isolation, and existing methods often struggle to establish continuous correspondences and stable mappings of the same cardiac chamber throughout the complete cardiac cycle, easily leading to boundary drift, regional jumps, and temporal mismatches. For cardiac chamber temporal objects with artifact interference, local missegmentation, and non-stationary volume changes, traditional curve construction and key phase identification methods are prone to false extrema, abnormal fluctuations, and parameter distortion, thus affecting the stability and reliability of cardiac function assessment results.

[0005] Therefore, it is necessary to provide a method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA, so that the cardiac chambers can form a unified temporal analysis object with cross-phase continuity and physiological rationality within a complete cardiac cycle, and on this object, complete stable volume curve construction, key phase identification, cardiac function parameter calculation, and quality control closed-loop output. Summary of the Invention

[0006] One objective of this invention is to propose a method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA. This invention does not simply segment the images of each phase and perform volume statistics, but rather constructs a unified temporal cardiac chamber object through an integrated mechanism of temporal standardization, cross-phase continuous correspondence, and consistency correction. This unified temporal cardiac chamber object serves as the sole analytical basis for subsequent volume curve construction, key phase identification, and calculation of cardiac function parameters.

[0007] The method for constructing cardiac chamber volume curves and assessing cardiac function according to embodiments of the present invention includes the following steps: Step 1: Acquire 4D-CT or 4D-CTA time-series images of the same subject, read at least one of the following: phase identification information, acquisition time information, ECG gating information, and reconstruction time information, organize the phases according to the acquisition time, and construct a standardized cardiac cycle coordinate system; Step 2: Perform cardiac chamber segmentation on each phase image in the standardized cardiac cycle coordinate system to obtain the initial segmentation results of each phase cardiac chamber; Step 3: Select a reference phase from the initial segmentation results of the cardiac chambers in each phase. Using the spatial transformation relationship or temporal correlation information of adjacent phases, propagate the initial segmentation results of the reference phase cardiac chambers to the other phases. Then, fuse the propagation results with the initial segmentation results of the current phase cardiac chambers. Under the conditions of forward and reverse consistency verification, long-range temporal consistency constraints and physiological consistency constraints, the establishment of cross-phase continuous correspondence and consistency correction are completed simultaneously to form a unified temporal cardiac chamber object. Step 4: Extract the volume values ​​of each phase from the unified temporal cavity object, and construct the full-cycle volume curve according to the time order in the standardized cardiac cycle coordinate system; Step 5: Search for peaks, valleys, and inflection points along the full-cycle volume curve, and identify key phases by combining the relationship between adjacent phase changes, local neighborhood stability, peak-valley stability, curve continuity, or inter-chamber consistency. Step Six: Calculate cardiac function parameters based on the corresponding volume values ​​of key phases in the full-cycle volume curve. During the construction of the unified temporal chamber object, generation of the full-cycle volume curve, identification of key phases, and calculation of cardiac function parameters, perform quality control and reliability assessment on abnormal phases, abnormal curves, and abnormal parameters. Based on the assessment results, perform closed-loop processing to generate at least one of the following: reliability score, abnormal cause code, and review prompt. Output the cardiac function assessment result.

[0008] Optionally, step one specifically includes: Acquire 4D-CT or 4D-CTA time-series images of the same subject, read at least one of the acquisition time information, phase identification information, ECG gating information, and reconstruction time information corresponding to each phase image, and establish the correspondence between the acquisition time information and each phase image. Based on the acquisition time information, phase identification information, or ECG gating information, the phase images are sorted by time, and the phase images are normalized according to the phase identification to form an initial phase sequence; An integrity check is performed on the initial phase sequence to identify phase missing, phase repetition, uneven phase density, and cross-cardiac cycle confounding conditions. Based on the identification results, one of the following processes is performed: deduplication, filtering, interpolation completion, or resampling to form a corrected phase sequence. Based on the time position of each phase image in the corrected phase sequence, each phase image is mapped to a unified cardiac cycle time axis to form a standardized cardiac cycle coordinate system.

[0009] Optionally, step two specifically involves: Arrange the phase images in the standardized cardiac cycle coordinate system according to their time position to form a phase image sequence; Heart chamber segmentation is performed phase by phase along the phase image sequence. In the phase image corresponding to each time position, the heart chamber region is determined by one of the following methods: boundary contour determination, region segmentation, model segmentation, and deep learning segmentation. The target heart chamber includes at least one or a combination of the left ventricle, right ventricle, left atrium, or right atrium. The cardiac chamber region is converted into a cardiac chamber segmentation labeling result corresponding to the time position, so that each time position corresponds to a cardiac chamber segmentation labeling result; The cardiac chamber segmentation and labeling results are matched with the phase image sequence according to the time position to form the initial segmentation results of the cardiac chamber for each phase.

[0010] Optionally, step three specifically includes: The initial segmentation result of the reference phase is selected from the initial segmentation results of the cardiac chambers in each phase according to the time position, and the temporal continuity relationship between the reference phase and the adjacent phase is determined according to the standardized cardiac cycle coordinate system. Based on the spatial transformation relationship or temporal correlation information of adjacent phases, the initial segmentation result of the reference phase is propagated phase by phase along the time position to each adjacent phase, forming the forward propagation result corresponding to each time position, and propagated phase by phase along the opposite time direction, forming the reverse propagation result corresponding to each time position. The forward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase of the heart cavity corresponding to the same time position, and the backward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase of the heart cavity corresponding to the same time position, to form the fused results corresponding to each time position. Perform forward and reverse consistency checks on the fusion results corresponding to each time position to determine the continuous mapping relationship between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position, forming a cross-phase continuous correspondence relationship; In the cross-phase continuous correspondence, the continuous mapping relationship between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position is extracted according to the time position order to form a time-series mapping sequence; Along the temporal mapping sequence, a long-range mapping consistency comparison is performed between the initial segmentation result of the cardiac cavity at the reference phase and the corresponding cardiac cavity region at non-adjacent time positions. Boundary deviation value, region deviation value, and mapping continuity deviation value are calculated to form a long-range temporal consistency constraint. Along the time-series mapping sequence, a physiological change consistency comparison is performed between the corresponding cardiac chamber regions at adjacent time positions. The direction of volume change, the magnitude of volume change, and the trend of boundary change are calculated to form physiological consistency constraints. The physiological consistency constraints include at least one of the following: logical constraints on volume change between chambers, approximately conservation constraints on left and right ventricular output, and reasonable constraints on volume change before and after key phases. Based on long-term temporal consistency constraints and physiological consistency constraints, boundary correction and region correction are performed on the corresponding cardiac cavity regions at each time position, and the corrected cardiac cavity regions at each time position are connected in chronological order to form a unified temporal cardiac cavity object.

[0011] Optionally, step four specifically includes: In a unified temporal cardiac cavity object, the cardiac cavity regions corresponding to each time position are extracted according to the time position order in the standardized cardiac cycle coordinate system to form a cardiac cavity region sequence. Volume statistics are performed on the cardiac cavity regions corresponding to each time position in the cardiac cavity region sequence. The number of voxels contained in the cardiac cavity region corresponding to each time position is calculated, and the volume value corresponding to each time position is determined based on the number of voxels and the spatial volume corresponding to each voxel. According to the time position in the standardized cardiac cycle coordinate system, the volume value corresponding to each time position and the corresponding cardiac chamber region are matched and organized to form a volume recording sequence; By sequentially connecting the volume values ​​corresponding to the previous time position with the volume values ​​corresponding to the next time position in the volume record sequence, a volume change trajectory is formed. The volume change trajectory is continuously unfolded at all time points in the standardized cardiac cycle coordinate system to form a full-cycle volume curve. The volume values ​​corresponding to each time position in the full-cycle volume curve are written into the curve record set in chronological order to form the full-cycle volume curve result of the target heart chamber within a complete cardiac cycle.

[0012] Optionally, step five specifically includes: The volume values ​​corresponding to each time position are extracted along the full-cycle volume curve in chronological order, and the volume change is calculated based on the difference between the volume value corresponding to the previous time position and the volume value corresponding to the next time position, thus forming a volume change sequence. Based on the direction of increase or decrease of volume change at each time position in the volume change sequence, divide the volume increase interval into volume increase interval and volume decrease interval, and determine the time position at which the direction of volume change changes. The peak position is determined at the time position where the volume increase interval intersects with the volume decrease interval; the trough position is determined at the time position where the volume decrease interval intersects with the volume increase interval; and the turning point is determined at the time position where the volume change changes from increasing to decreasing and from decreasing to increasing. The peak position, trough position, and turning point are checked for consistency with the changes in adjacent time positions, local neighborhood stability, cross-phase consistency, or inter-chamber physiological consistency. Peak positions, trough positions, and turning points that meet the condition of continuous change are retained, and the retained results are determined as key phases.

[0013] Optionally, step six specifically includes: Extract the corresponding volume value of each key phase in the full-cycle volume curve from the key phases, and record them according to the time position of the key phases in the standardized cardiac cycle coordinate system to form a key phase volume record. The maximum and minimum volume values ​​are determined in the key phase volume records. The stroke volume is calculated based on the difference between the maximum and minimum volume values. The ejection fraction is calculated based on the ratio of the stroke volume to the maximum volume value, thus forming a record of cardiac function parameters. Quality control and reliability assessment are performed based on key phase volume records, full-cycle volume curves, and cardiac function parameter records. The time position of key phases, continuity of volume changes, and range of cardiac function parameters are checked to identify abnormal phases, abnormal curves, and abnormal parameters, and to form quality control results. Based on the quality control results, closed-loop processing is performed on abnormal phases, abnormal curves, and abnormal parameters. The closed-loop processing includes one of the following: removing abnormal phases and interpolating compensation based on neighboring phases, reducing the weight of abnormal results, performing restrictive correction on local curves, retaining results and outputting warning information, marking abnormal areas, and prompting manual review. After the closed-loop processing is completed, the key phase volume records, cardiac function parameter records, and quality control results are collected according to time position to form at least one of the following: a confidence score, an abnormality cause code, and a suggested review information, and a cardiac function assessment result is formed.

[0014] Optionally, the physiological consistency constraints include at least one of the following: logical constraints on inter-chamber volume changes, approximately conservation constraints on left and right ventricular output, and reasonable constraints on volume changes before and after key phases.

[0015] Optionally, when there are multiple competing extreme points, local pseudo-extreme points, or phase conflicts, abnormal extreme points are excluded based on local neighborhood stability, cross-phase consistency, or inter-compartment physiological consistency, and unstable extreme points are downweighted or marked as pending review.

[0016] Optionally, the cardiac function assessment results include at least one of the following: cardiac function parameter values, full-cycle volume curve, key phase number or time position, quality control conclusion, confidence score, abnormality cause code, and phase number or heart chamber name recommended for review.

[0017] The beneficial effects of this invention are: Compared to existing technologies that isolate and process 4D-CT images of each phase before performing volume measurements and parameter calculations, this invention first performs temporal standardization processing on 4D-CT / 4D-CTA multi-phase images and constructs a standardized cardiac cycle coordinate system. Then, through an integrated mechanism of cross-phase continuous correspondence and consistency correction, a unified temporal cavity object is formed. This unified temporal cavity object unifies the anatomical structural changes of the same cavity within a complete cardiac cycle into a continuous temporal framework. This reduces the impact of phase loss, phase repetition, local boundary drift, temporal mismatch, and single-frame missegmentation on the analysis results, thereby improving temporal consistency, object continuity, and regional localization stability during the dynamic analysis of cardiac cavities.

[0018] Based on this, the present invention extracts the volume values ​​of each phase from a unified temporal cardiac chamber object and constructs a full-cycle volume curve. Then, by combining peak values, trough values, inflection points, and the relationship between adjacent phase changes, key phases are identified. This further completes the calculation of cardiac function parameters and the closed-loop quality control processing for abnormal phases, abnormal curves, and abnormal parameters. Because key phase identification and parameter calculation are based on a full-cycle continuous curve and a unified temporal cardiac chamber object, rather than on isolated phases or local extrema, it can suppress misjudgments caused by spurious extrema and abnormal fluctuations, improving the stability and reliability of cardiac function parameters such as stroke volume and ejection fraction.

[0019] The aforementioned technical effects are not obtained by the natural superposition of individual technical means, but rely on the synergistic effect of the integrated mechanism of cross-phase continuous correspondence and consistency correction; without this integrated mechanism, existing technologies are unable to obtain stable time-series consistency analysis results. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA proposed in this invention; Figure 2 This is a schematic diagram of the integrated mechanism of cross-phase continuous correspondence and consistency correction in the 4D-CT or 4D-CTA cardiac chamber volume curve construction and cardiac function assessment method proposed in this invention. Figure 3 This is a schematic diagram illustrating the construction of full-cycle volume curves and key phase identification based on a unified temporal cardiac cavity object for the 4D-CT or 4D-CTA cardiac chamber volume curve construction and cardiac function assessment method proposed in this invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0022] refer to Figures 1-3 The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA includes the following steps: Step 1: Acquire 4D-CT or 4D-CTA time-series images of the same subject, read at least one of the following: phase identification information, acquisition time information, ECG gating information, and reconstruction time information, organize the phases according to the acquisition time, and construct a standardized cardiac cycle coordinate system; Step 2: Perform cardiac chamber segmentation on each phase image in the standardized cardiac cycle coordinate system to obtain the initial segmentation results of each phase cardiac chamber; Step 3: Select a reference phase from the initial segmentation results of the cardiac chambers in each phase. Using the spatial transformation relationship or temporal correlation information of adjacent phases, propagate the initial segmentation results of the reference phase cardiac chambers to the other phases. Then, fuse the propagation results with the initial segmentation results of the current phase cardiac chambers. Under the conditions of forward and reverse consistency verification, long-range temporal consistency constraints and physiological consistency constraints, the establishment of cross-phase continuous correspondence and consistency correction are completed simultaneously to form a unified temporal cardiac chamber object. Step 4: Extract the volume values ​​of each phase from the unified temporal cavity object, and construct the full-cycle volume curve according to the time order in the standardized cardiac cycle coordinate system; Step 5: Search for peaks, valleys, and inflection points along the full-cycle volume curve, and identify key phases by combining the relationship between adjacent phase changes, local neighborhood stability, peak-valley stability, curve continuity, or inter-chamber consistency. Step Six: Calculate cardiac function parameters based on the corresponding volume values ​​of key phases in the full-cycle volume curve. During the construction of the unified temporal chamber object, generation of the full-cycle volume curve, identification of key phases, and calculation of cardiac function parameters, perform quality control and reliability assessment on abnormal phases, abnormal curves, and abnormal parameters. Based on the assessment results, perform closed-loop processing to generate at least one of the following: reliability score, abnormal cause code, and review prompt. Output the cardiac function assessment result.

[0023] In this embodiment, step one specifically includes: When acquiring 4D-CT or 4D-CTA time-series images of the same subject, each phase image is extracted one by one from the 4D-CT or 4D-CTA time-series images. Then, the acquisition time information is read from the data record corresponding to each phase image. Each acquisition time information is associated with a phase image in a one-to-one correspondence manner, thereby forming a correspondence between the acquisition time information and each phase image. When sorting the phase images according to the acquisition time information, phase identification information, or ECG gating information, the phase images are arranged in the order of acquisition time information, so that the phase images acquired earlier are placed in the first position and the phase images acquired later are placed in the last position. Then, the phase images are normalized according to the phase identification, so that the phase identification under different identification methods is converted into a unified identification form, thereby forming an initial phase sequence. When performing an integrity check on the initial phase sequence, the temporal distribution of each phase image is checked along the arrangement order of the initial phase sequence to identify phase missing, phase duplication, uneven phase density, and cross-cardiac cycle mixed cases. Based on the identification results, one of the following processes is performed: deduplication, filtering, interpolation completion, and resampling. Deduplication is used to delete duplicate phase images, filtering is used to retain phase images within the same cardiac cycle, interpolation completion is used to add phase images corresponding to the missing time positions, and resampling is used to adjust the distribution of each phase image on the time axis, thereby forming a corrected phase sequence. Based on the time position corresponding to each phase image in the corrected phase sequence, when mapping each phase image to the unified cardiac cycle time axis, the time position corresponding to each phase image in the corrected phase sequence is read first, and then the position is allocated according to the arrangement order of the time position in the unified cardiac cycle time axis, so that each phase image in the corrected phase sequence forms an orderly distribution on the unified cardiac cycle time axis, thereby forming a standardized cardiac cycle coordinate system.

[0024] In this embodiment, step two specifically involves: The phase images in the standardized cardiac cycle coordinate system are arranged in chronological order to form a phase image sequence. First, the phase images corresponding to each time position are extracted from the standardized cardiac cycle coordinate system. Then, they are arranged in chronological order from front to back, so that the phase images corresponding to each time position are arranged continuously in the same sequence, thus forming a phase image sequence consistent with the standardized cardiac cycle coordinate system. Heart cavity segmentation is performed phase by phase along the phase image sequence. In the phase image corresponding to each time position, the heart cavity region is determined by one of the following methods: boundary contour determination, region segmentation, model segmentation, and deep learning segmentation. In the phase image corresponding to each time position, the image range where the target heart cavity is located is first located. Then, a segmentation method is selected to perform region recognition on the image range where the target heart cavity is located, so that the target heart cavity and the non-target region form a clear boundary in the image space. Thus, a heart cavity region is determined in the phase image corresponding to each time position. The target heart cavity includes at least one or a combination of the left ventricle, right ventricle, left atrium, or right atrium. The cardiac chamber region is converted into a cardiac chamber segmentation labeling result corresponding to the time position, so that each time position corresponds to a cardiac chamber segmentation labeling result. In the phase image corresponding to each time position, the image range covered by the cardiac chamber region is assigned a cardiac chamber segmentation label, and the image range outside the cardiac chamber region is assigned a non-cardiac chamber segmentation label. A one-to-one correspondence is established between the cardiac chamber segmentation labeling result and the corresponding time position, so that each time position forms an independent cardiac chamber segmentation labeling result. The heart chamber segmentation and labeling results are matched with the phase image sequence according to the time position to form the initial segmentation results of the heart chambers in each phase. First, the heart chamber segmentation and labeling results corresponding to each time position are extracted according to the arrangement order in the phase image sequence. Then, the heart chamber segmentation and labeling results corresponding to each time position are paired and collected with the phase image corresponding to the same time position. The results are written into a unified record set in the order of time position, so that the correspondence between the phase image and the heart chamber segmentation and labeling results is preserved at each time position, thereby forming the initial segmentation results of the heart chambers in each phase.

[0025] In this embodiment, step three specifically includes: The initial segmentation results of the heart chambers of each phase are selected according to their time positions. The temporal continuity between the reference phase and the adjacent phase is determined based on the standardized cardiac cycle coordinate system. First, the initial segmentation results of the heart chambers of each phase are arranged from front to back according to their time positions. Then, the initial segmentation results of the heart chambers located at the reference time position of the standardized cardiac cycle coordinate system are determined as the initial segmentation results of the reference phase. The continuity order between the reference phase and the adjacent phase is determined according to the preceding and following time positions of the reference phase in the standardized cardiac cycle coordinate system, so that the reference phase and the adjacent phase form a temporal adjacency relationship that unfolds continuously along the time position. Based on the spatial transformation relationship or temporal correlation information of adjacent phases, the initial segmentation result of the reference phase is propagated phase by phase along the time position to each adjacent phase, forming the forward propagation result corresponding to each time position. Then, it is propagated phase by phase along the opposite time direction, forming the reverse propagation result corresponding to each time position. According to the temporal continuity relationship between the reference phase and the adjacent phases, the initial segmentation result of the reference phase is mapped to the position of the heart cavity region in the adjacent time position. Then, the mapped position of the heart cavity region is passed to the next time position, so that the initial segmentation result of the reference phase is continuously extended along the time sequence and forms the forward propagation result corresponding to each time position. At the same time, the same mapping and transmission are performed in the opposite time direction, so that the initial segmentation result of the reference phase is continuously extended along the reverse time sequence and forms the reverse propagation result corresponding to each time position. The forward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase heart cavity at the same time position. The backward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase heart cavity at the same time position, forming fusion results corresponding to each time position. Within the same time position, the heart cavity region positions in the forward propagation results and the heart cavity region positions in the initial segmentation results of the current phase heart cavity are compared for overlap, boundary fit, and region alignment. Regions that meet the joining conditions are then merged to form a forward fusion result. Within the same time position, the heart cavity region positions in the backward propagation results and the heart cavity region positions in the initial segmentation results of the current phase heart cavity are compared and merged in the same time position to form a reverse fusion result. The forward and reverse fusion results are then organized according to time position correspondence to form fusion results corresponding to each time position. The forward and reverse consistency checks are performed on the fusion results corresponding to each time position to determine the continuous mapping relationship between the initial segmentation results of the reference phase heart cavity and the corresponding heart cavity regions at each time position, forming a cross-phase continuous correspondence relationship. In the cross-phase continuous correspondence, the continuous mapping relationship between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position is extracted according to the time position order to form a time-series mapping sequence. First, the cross-phase continuous correspondence is unfolded sequentially along the time position order in the standardized cardiac cycle coordinate system. Then, the mapping record between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position is extracted item by item and written into the same arrangement sequence from front to back according to the time position, so that the mapping order from the initial segmentation result of the reference phase heart cavity to the corresponding heart cavity region at each time position is continuously maintained, forming a time-series mapping sequence. Along the temporal mapping sequence, a long-range mapping consistency comparison is performed between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity regions at non-adjacent time positions. Boundary deviation values, regional deviation values, and mapping continuity deviation values ​​are calculated to form long-range temporal consistency constraints. Between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity regions at each time position at least one time position apart, the differences in boundary contour positions are compared and boundary deviation values ​​are calculated. The differences in regional coverage are compared and regional deviation values ​​are calculated. The mapping connection order and mapping connection state are compared and mapping continuity deviation values ​​are calculated. Then, the boundary deviation values, regional deviation values, and mapping continuity deviation values ​​are organized according to the time position to form long-range temporal consistency constraints. Along the temporal mapping sequence, a physiological consistency comparison is performed between corresponding cardiac chamber regions at adjacent time positions. The direction of volume change, the magnitude of volume change, and the trend of boundary change are calculated to form physiological consistency constraints. Between the cardiac chamber regions corresponding to the previous time position and the cardiac chamber regions corresponding to the next time position, the increase or decrease of volume value is compared to determine the direction of volume change, the difference in volume value is compared to determine the magnitude of volume change, and the expansion and contraction of the boundary contour are compared to determine the trend of boundary change. Then, the direction of volume change, the magnitude of volume change, and the trend of boundary change are written in the order of adjacent time positions to form physiological consistency constraints. The physiological consistency constraints include at least one of the following: logical constraints on volume change between chambers, approximate conservation constraints on left and right ventricular output, and reasonable constraints on volume change before and after key phases.

[0026] Based on long-term temporal consistency constraints and physiological consistency constraints, boundary correction and region correction are performed on the cardiac chamber regions corresponding to each time position. The corrected cardiac chamber regions corresponding to each time position are then connected in chronological order to form a unified temporal cardiac chamber object. First, boundary position adjustment and region range adjustment are performed on time positions with large boundary deviation values ​​based on long-term temporal consistency constraints. Then, boundary correction and region correction are performed on time positions with abnormal volume change direction, abnormal volume change amplitude, and abnormal boundary change trend based on physiological consistency constraints. Finally, the cardiac chamber regions corresponding to each time position after correction are connected continuously from front to back according to time position, so that the cardiac chamber regions corresponding to each time position maintain a coherent mapping relationship in chronological order, forming a unified temporal cardiac chamber object.

[0027] This invention performs forward and backward propagation of the initial segmentation result of the reference phase heart cavity along the time position, and fuses the propagation result with the initial segmentation result of the current phase heart cavity at the same time position. Then, it establishes a cross-phase continuous correspondence by combining forward and backward consistency verification. At the same time, it introduces long-range temporal consistency constraints and physiological consistency constraints into the cross-phase continuous correspondence and performs boundary correction and region correction on the heart cavity region corresponding to each time position. It can suppress single-frame segmentation error, local boundary drift, phase misalignment, long-distance mapping deviation and unreasonable morphological fluctuations between adjacent time positions within the entire cardiac cycle. The resulting unified temporal heart cavity object has both temporal continuity and physiological rationality, thereby improving the accuracy, stability and reliability of establishing cross-phase continuous correspondence, constructing full-cycle volume curves, identifying key phases and calculating cardiac function parameters.

[0028] In this embodiment, step four specifically includes: The unified temporal cavity objects are unfolded sequentially along the time positions in the standardized cardiac cycle coordinate system. The cavity regions corresponding to each time position are extracted sequentially. The cavity region corresponding to the starting time position is written to the beginning of the sequence. Then, the cavity regions corresponding to subsequent time positions are written to the subsequent positions in a progressive time order, so that the cavity regions corresponding to all time positions are arranged continuously in the same time order, forming a cavity region sequence. Volume statistics are performed on each cardiac cavity region in the cardiac cavity region sequence. Voxel units with cardiac cavity segmentation marks are detected one by one in the cardiac cavity region corresponding to each time position. The voxel units with cardiac cavity segmentation marks are accumulated and counted to obtain the number of voxel units corresponding to the current time position. Then, the number of voxel units is multiplied by the spatial volume corresponding to a single voxel unit to obtain the volume value corresponding to the current time position, so that each cardiac cavity region corresponding to each time position is associated with a certain volume value. The volume value corresponding to each time position is paired and organized with the corresponding cardiac chamber region at the same time position according to the time position. An independent record is created at each time position. The time position is written into the record start field, the corresponding cardiac chamber region at the same time position is written into the region field, and the corresponding volume value at the same time position is written into the value field. This allows a record to simultaneously represent the cardiac chamber region and volume value at a time position. All records are then arranged from front to back according to the time position to form a volume record sequence. The volume record sequence is unfolded sequentially along the time position. The volume value corresponding to the previous time position and the volume value corresponding to the next time position are extracted between two adjacent records. The volume value corresponding to the previous time position is determined as the connection start point and the volume value corresponding to the next time position is determined as the connection end point. Then, the start point and the end point are sequentially connected in the direction of time progression. The same process is repeated for all adjacent time positions so that the volume values ​​form a continuous connection relationship on the time axis, forming a volume change trajectory. The volume change trajectory is continuously unfolded at all time positions in the standardized cardiac cycle coordinate system. The volume connection relationships corresponding to the starting time position, the intermediate time position, and the ending time position are all incorporated into the same continuous trajectory, and the connection order is kept consistent with the time order in the standardized cardiac cycle coordinate system. This allows the volume change process of the target cardiac chamber within the complete cardiac cycle to be expressed in the form of a continuous curve, forming a full-cycle volume curve. The volume values ​​corresponding to each time position in the full-cycle volume curve are written into the curve record set in chronological order. A record position is assigned to each time position in the curve record set, and the volume values ​​corresponding to each time position are written into the corresponding record position in chronological order. The time continuity between adjacent record positions is maintained, so that the curve record set continuously saves the volume change results of the target heart chamber in the complete cardiac cycle, forming the full-cycle volume curve result of the target heart chamber.

[0029] In this embodiment, step five specifically includes: The volume values ​​corresponding to each time position are extracted along the full-cycle volume curve in chronological order. The volume change is calculated based on the difference between the volume values ​​corresponding to the previous time position and the volume values ​​corresponding to the next time position, forming a volume change sequence. First, the volume values ​​corresponding to each time position in the full-cycle volume curve are expanded sequentially from front to back according to the time position. Then, a subtraction operation is performed between two adjacent time positions. The volume value corresponding to the previous time position is subtracted from the volume value corresponding to the next time position to obtain the volume change between the current adjacent time positions. Finally, all volume changes are written into the same sequence in chronological order to form a volume change sequence. Based on the direction of increase or decrease of volume change at each time position in the volume change sequence, the volume increase interval and the volume decrease interval are divided, and the time position where the direction of volume change changes is determined. The consecutive time positions where the volume change is greater than zero are classified into the volume increase interval, and the consecutive time positions where the volume change is less than zero are classified into the volume decrease interval. Then, the sign changes of the volume change corresponding to adjacent time positions are compared, and the time positions where the direction of volume change changes are recorded when the change is from positive to negative, from negative to positive, from continuous increase to continuous decrease, or from continuous decrease to continuous increase. The peak position is determined at the time position where the volume increase interval and the volume decrease interval intersect. The trough position is determined at the time position where the volume decrease interval and the volume increase interval intersect. The turning point is determined at the time position where the volume change changes from increasing to decreasing and from decreasing to increasing. The time position that is at the end of the volume increase interval and connects with the beginning of the volume decrease interval is written into the peak position set. The time position that is at the end of the volume decrease interval and connects with the beginning of the volume increase interval is written into the trough position set. The time position where the rate of change of volume change changes in the opposite direction is written into the turning point set, thus forming the peak position, trough position and turning point. The peak, trough, and turning points are checked for consistency with adjacent time positions, local neighborhood stability, cross-phase consistency, or inter-chamber physiological consistency. Peak, trough, and turning points that meet the condition of continuous change are retained, and the retained results are determined as key phases. The volume values ​​and volume changes of adjacent time positions are extracted on both sides of each peak, trough, and turning point. It is checked whether the adjacent time positions maintain a continuous rising, continuous falling, or reverse direction relationship. If the condition of continuous change is met, the corresponding time position is retained; if the condition of continuous change is not met, the corresponding time position is removed. The retained peak, trough, and turning points are then grouped according to time position to determine key phases.

[0030] In this embodiment, step six specifically includes: Extract the corresponding volume values ​​of each key phase in the full-cycle volume curve from the key phases, and record them according to the time position of the key phase in the standardized cardiac cycle coordinate system to form a key phase volume record. Sequentially search the curve records in the full-cycle volume curve that correspond one-to-one with each time position in the key phase, and write the retrieved volume values ​​into the recording units corresponding to each time position. Then, collect the key phase identifier, time position, and corresponding volume value into the same record set according to the pairing relationship, so that each key phase forms an independent key phase volume record. The maximum and minimum volume values ​​are determined in the key phase volume records. The stroke volume is calculated based on the difference between the maximum and minimum volume values. The ejection fraction is calculated based on the ratio of stroke volume to the maximum volume value, thus forming a cardiac function parameter record. First, the corresponding volume values ​​of each record are compared item by item in the key phase volume records. The record with the largest value is determined as the maximum volume value record, and the record with the smallest value is determined as the minimum volume value record. Then, the stroke volume is obtained by subtracting the minimum volume value from the maximum volume value. The ejection fraction is obtained by dividing the stroke volume by the maximum volume value. The maximum volume value, minimum volume value, stroke volume, and ejection fraction are written into the same parameter record set according to their corresponding relationship to form a cardiac function parameter record. Quality control and reliability assessment are performed based on key phase volume records, full-cycle volume curves, and cardiac function parameter records. This involves checking the key phase time positions, continuity of volume changes, and range of cardiac function parameter values ​​to identify abnormal phases, curves, and parameters, thus generating quality control results. First, the distribution of key phase time positions in the standardized cardiac cycle coordinate system is checked to determine if they correspond to the peak, trough, and inflection points in the full-cycle volume curve. Then, the direction, amplitude, and continuity of volume changes between adjacent time positions in the full-cycle volume curve are checked to determine if there are any abrupt changes, breaks, or abnormal fluctuations. Next, the maximum, minimum, stroke volume, and ejection fraction values ​​in the cardiac function parameter records are checked to determine if they fall within preset ranges. Finally, the results of each check are written into the quality control record set according to pass / fail status to form the quality control results. Based on the quality control results, closed-loop processing is performed on abnormal phases, abnormal curves, and abnormal parameters. This closed-loop processing includes one of the following: removing abnormal phases and performing interpolation compensation based on neighboring phases; reducing the weight of abnormal results; performing restrictive corrections on local curves; retaining results and outputting warning messages; marking abnormal regions; and prompting for manual review. After the closed-loop processing is completed, the key phase volume records, cardiac function parameter records, and quality control results are aggregated according to their time position to form at least one of the following: a confidence score, an anomaly cause code, and a suggested review information, thus forming a cardiac function assessment result. In the quality control results, the records corresponding to abnormal phases, abnormal curves, and abnormal parameters are first located. Then, the closed-loop processing method is selected according to the type of abnormal record. When performing abnormal phase removal and interpolation compensation based on neighboring phases, the records corresponding to the abnormal phases are removed from the key phase volume. The system records the volume values ​​before and after the abnormal phase, then extracts the corresponding volume values ​​and performs interpolation to fill in the abnormal phase volume position. When performing weight reduction processing on abnormal results, the weight of abnormal phase, abnormal curve, and abnormal parameters in the evaluation results is reduced. When performing restrictive correction on local curves, the volume change trajectory of the time segment where the abnormal curve is located is locally limited and connected. When performing result retention and warning information output, the abnormal record is retained in the result set and a warning label is written synchronously. When performing abnormal region marking, the heart chamber region corresponding to the abnormal phase is written into the abnormal region label. When performing manual review prompt, the abnormal record is written into the review prompt field. Finally, the key phase volume record, cardiac function parameter record, and quality control result that have completed closed-loop processing are written into the same result set from front to back according to time position to form the cardiac function evaluation result.

[0031] In this embodiment, the method can be implemented by a software system, a medical image post-processing workstation, a server, a terminal device, or a computer-readable storage medium storing program instructions, and the aforementioned carrier is used to execute the various steps of this method.

[0032] In this embodiment, the physiological consistency constraint includes at least one of the following: logical constraint on volume change between chambers, constraint on approximate conservation of left and right ventricular output, and constraint on the rationality of volume change before and after key phases.

[0033] In this embodiment, when there are multiple competing extreme points, local pseudo-extreme points, or phase conflicts, abnormal extreme points are excluded based on local neighborhood stability, cross-phase consistency, or inter-compartment physiological consistency, and unstable extreme points are downweighted or marked as pending review.

[0034] In this embodiment, the cardiac function assessment results include at least one of the following: cardiac function parameter values, full-cycle volume curve, key phase number or time position, quality control conclusion, confidence score, abnormality cause code, and recommended phase number or heart chamber name for review.

[0035] In this embodiment, the 4D-CT or 4D-CTA cardiac chamber volume curve construction and cardiac function assessment system includes a data acquisition and temporal standardization module, an initial segmentation module, a cross-phase continuous correspondence and consistency correction module, a volume calculation module, a full-cycle volume curve construction module, a key phase identification module, a cardiac function parameter calculation module, a quality control and reliability assessment module, and a result output module. The data acquisition and temporal standardization module is used to acquire 4D-CT or 4D-CTA temporal images and construct a standardized cardiac cycle coordinate system. The initial segmentation module is used to form the initial segmentation results of cardiac chambers in each phase. The cross-phase continuous correspondence and consistency correction module is used to establish cross-phase continuous correspondence relationships and form a unified temporal cardiac chamber object. The volume calculation module and the full-cycle volume curve construction module are used to extract the volume values ​​of each phase and form a full-cycle volume curve. The key phase identification module is used to identify key phases. The cardiac function parameter calculation module is used to calculate cardiac function parameters. The quality control and reliability assessment module is used to identify abnormal phases, abnormal curves, and abnormal parameters and perform closed-loop processing. The result output module is used to output the cardiac function assessment results.

[0036] In this embodiment, the electronic device includes a processor, a memory, and a computer program stored in the memory. When the processor executes the computer program, it performs the following steps: acquiring 4D-CT or 4D-CTA time-series images of the same subject, constructing a standardized cardiac cycle coordinate system, forming initial segmentation results of each phase of the cardiac chamber, establishing cross-phase continuous correspondence, forming a unified time-series cardiac chamber object, constructing a full-cycle volume curve, identifying key phases, calculating cardiac function parameters, and performing quality control and reliability assessment and completing closed-loop processing. The electronic device can be a server, an image post-processing workstation, a medical imaging terminal, a cloud computing node, or a dedicated computing device with medical image analysis capabilities.

[0037] In this embodiment, a computer-readable storage medium stores a computer program. When the computer program is loaded and executed by the processor, it is used to implement various processes such as sorting phases according to the acquisition time, constructing a standardized cardiac cycle coordinate system, performing cardiac chamber segmentation, establishing cross-phase continuous correspondence, performing consistency correction to form a unified temporal cardiac chamber object, extracting the volume values ​​of each phase to construct a full-cycle volume curve, identifying key phases, calculating cardiac function parameters, and performing quality control and reliability assessment on abnormal phases, abnormal curves and abnormal parameters, and completing closed-loop processing based on the assessment results. The computer-readable storage medium can be a hard disk, solid-state drive, mobile storage device, optical disk, flash drive or storage medium capable of carrying program instructions.

[0038] Example 1: To verify the feasibility of this invention in practice, it was applied to a scenario of quantitative assessment of cardiac function based on 4D-CT or 4D-CTA temporal images. In this scenario, after a cardiac 4D-CT or 4D-CTA scan, continuous phase images covering the entire cardiac cycle can be obtained. Although the images are sequential in time, they are still easily affected by factors such as heart rate fluctuations, differences in reconstruction beats, uneven local contrast, motion artifacts, and blurred right ventricular boundaries during actual processing. This often results in problems such as uneven phase distribution, local boundary drift, single-frame segmentation jumps, and local bulges in the volume curve. Traditional methods typically segment each phase image independently, then calculate the volume separately and directly extract extreme points as key phases. This method is prone to misinterpreting local jumps caused by missegmentation as true peaks or troughs when image quality is unstable, leading to increased deviations in key phase localization. Consequently, parameters such as stroke volume and ejection fraction fluctuate significantly, and the workload of manual review also increases.

[0039] In this embodiment, the acquired 4D-CT or 4D-CTA temporal images are first organized into phases according to the acquisition time, and a standardized cardiac cycle coordinate system is established so that all phase images are unfolded within a unified time frame. Then, cardiac chamber segmentation is performed on each phase image to obtain the initial segmentation results of each phase cardiac chamber. Subsequently, a reference phase is selected from the initial segmentation results of each phase cardiac chamber, and the initial segmentation results of the reference phase cardiac chamber are propagated to the forward and backward time positions using the adjacent phase space transformation relationship. The initial segmentation results of the reference phase cardiac chamber are then fused with the initial segmentation results of the cardiac chamber at the corresponding time position, and a cross-phase continuous correspondence is established through forward and reverse consistency verification. After the continuous correspondence is formed, long-range temporal consistency constraints and physiological consistency constraints are added to jointly correct the boundaries and regions, forming a unified temporal cardiac chamber object. After establishing a unified temporal cardiac chamber object, volume values ​​are extracted from various time positions to construct a full-cycle volume curve. Peaks, troughs, and inflection points are then searched along the curve, and key phases are identified by combining the relationship between adjacent phase changes. Finally, cardiac function parameters are calculated based on the volume values ​​corresponding to the key phases. Simultaneously, quality control and reliability assessments are performed on abnormal phases, abnormal curves, and abnormal parameters. Closed-loop processing is completed based on the assessment results, and the assessment results are output. Through this processing flow, the originally scattered and fluctuating single-frame analysis results are unified into a continuous temporal object, reducing the interference of local missegmentation, boundary drift, and phase mismatch on the overall judgment, improving the stability of the full-cycle volume curve construction, the accuracy of key phase identification, and the reliability of cardiac function parameter results.

[0040] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for constructing cardiac chamber volume curves and assessing cardiac function using 1.4D-CT or 4D-CTA, characterized in that: Includes the following steps: Step 1: Acquire 4D-CT or 4D-CTA time-series images of the same subject, read at least one of phase identification information, acquisition time information, ECG gating information, and reconstruction time information, organize the phases according to at least one of the phase identification information, acquisition time information, ECG gating information, and reconstruction time information, and construct a standardized cardiac cycle coordinate system. Step 2: Perform cardiac chamber segmentation on each phase image in the standardized cardiac cycle coordinate system to obtain the initial segmentation results of each phase cardiac chamber; Step 3: Select a reference phase from the initial segmentation results of the cardiac chambers in each phase. Using the spatial transformation relationship or temporal correlation information of adjacent phases, propagate the initial segmentation results of the reference phase cardiac chambers to the other phases. Then, fuse the propagation results with the initial segmentation results of the current phase cardiac chambers. Under the conditions of forward and reverse consistency verification, long-range temporal consistency constraints and physiological consistency constraints, the establishment of cross-phase continuous correspondence and consistency correction are completed simultaneously to form a unified temporal cardiac chamber object. Step 4: Extract the volume values ​​of each phase from the unified temporal cavity object, and construct the full-cycle volume curve according to the time order in the standardized cardiac cycle coordinate system; Step 5: Search for peaks, valleys, and inflection points along the full-cycle volume curve, and identify key phases by combining the relationship between adjacent phase changes, local neighborhood stability, peak-valley stability, curve continuity, or inter-chamber consistency. Step Six: Calculate cardiac function parameters based on the corresponding volume values ​​of key phases in the full-cycle volume curve. During the construction of the unified temporal chamber object, generation of the full-cycle volume curve, identification of key phases, and calculation of cardiac function parameters, perform quality control and reliability assessment on abnormal phases, abnormal curves, and abnormal parameters. Based on the assessment results, perform closed-loop processing to generate at least one of the following: reliability score, abnormal cause code, and review prompt. Output the cardiac function assessment result.

2. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step one specifically involves: Acquire 4D-CT or 4D-CTA time-series images of the same subject, read at least one of the acquisition time information, phase identification information, ECG gating information, and reconstruction time information corresponding to each phase image, and establish the correspondence between the acquisition time information and each phase image. Based on the acquisition time information, phase identification information, or ECG gating information, the phase images are sorted by time, and the phase images are normalized according to the phase identification to form an initial phase sequence; An integrity check is performed on the initial phase sequence to identify phase missing, phase repetition, uneven phase density, and cross-cardiac cycle confounding conditions. Based on the identification results, one of the following processes is performed: deduplication, filtering, interpolation completion, or resampling to form a corrected phase sequence. Based on the time position of each phase image in the corrected phase sequence, each phase image is mapped to a unified cardiac cycle time axis to form a standardized cardiac cycle coordinate system.

3. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step two specifically involves: Arrange the phase images in the standardized cardiac cycle coordinate system according to their time position to form a phase image sequence; Heart chamber segmentation is performed phase by phase along the phase image sequence. In the phase image corresponding to each time position, the heart chamber region is determined by one of the following methods: boundary contour determination, region segmentation, model segmentation, and deep learning segmentation. The target heart chamber includes at least one or a combination of the left ventricle, right ventricle, left atrium, or right atrium. The cardiac chamber region is converted into a cardiac chamber segmentation labeling result corresponding to the time position, so that each time position corresponds to a cardiac chamber segmentation labeling result; The cardiac chamber segmentation and labeling results are matched with the phase image sequence according to the time position to form the initial segmentation results of the cardiac chamber for each phase.

4. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step three specifically involves: The initial segmentation result of the reference phase is selected from the initial segmentation results of the cardiac chambers in each phase according to the time position, and the temporal continuity relationship between the reference phase and the adjacent phase is determined according to the standardized cardiac cycle coordinate system. Based on the spatial transformation relationship or temporal correlation information of adjacent phases, the initial segmentation result of the reference phase is propagated phase by phase along the time position to each adjacent phase, forming the forward propagation result corresponding to each time position, and propagated phase by phase along the opposite time direction, forming the reverse propagation result corresponding to each time position. The forward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase of the heart cavity corresponding to the same time position, and the backward propagation results corresponding to each time position are fused with the initial segmentation results of the current phase of the heart cavity corresponding to the same time position, to form the fused results corresponding to each time position. Perform forward and reverse consistency checks on the fusion results corresponding to each time position to determine the continuous mapping relationship between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position, forming a cross-phase continuous correspondence relationship; In the cross-phase continuous correspondence, the continuous mapping relationship between the initial segmentation result of the reference phase heart cavity and the corresponding heart cavity region at each time position is extracted according to the time position order to form a time-series mapping sequence; Along the temporal mapping sequence, a long-range mapping consistency comparison is performed between the initial segmentation result of the cardiac cavity at the reference phase and the corresponding cardiac cavity region at non-adjacent time positions. Boundary deviation value, region deviation value, and mapping continuity deviation value are calculated to form a long-range temporal consistency constraint. Along the time-series mapping sequence, a physiological change consistency comparison is performed between the corresponding cardiac chamber regions at adjacent time positions. The direction of volume change, the magnitude of volume change, and the trend of boundary change are calculated to form physiological consistency constraints. The physiological consistency constraints include at least one of the following: logical constraints on volume change between chambers, approximately conservation constraints on left and right ventricular output, and reasonable constraints on volume change before and after key phases. Based on long-term temporal consistency constraints and physiological consistency constraints, boundary correction and region correction are performed on the corresponding cardiac cavity regions at each time position, and the corrected cardiac cavity regions at each time position are connected in chronological order to form a unified temporal cardiac cavity object.

5. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step four specifically involves: In a unified temporal cardiac cavity object, the cardiac cavity regions corresponding to each time position are extracted according to the time position order in the standardized cardiac cycle coordinate system to form a cardiac cavity region sequence. Volume statistics are performed on the cardiac cavity regions corresponding to each time position in the cardiac cavity region sequence. The number of voxels contained in the cardiac cavity region corresponding to each time position is calculated, and the volume value corresponding to each time position is determined based on the number of voxels and the spatial volume corresponding to each voxel. According to the time position in the standardized cardiac cycle coordinate system, the volume value corresponding to each time position and the corresponding cardiac chamber region are matched and organized to form a volume recording sequence; By sequentially connecting the volume values ​​corresponding to the previous time position with the volume values ​​corresponding to the next time position in the volume record sequence, a volume change trajectory is formed. The volume change trajectory is continuously unfolded at all time points in the standardized cardiac cycle coordinate system to form a full-cycle volume curve. The volume values ​​corresponding to each time position in the full-cycle volume curve are written into the curve record set in chronological order to form the full-cycle volume curve result of the target heart chamber within a complete cardiac cycle.

6. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step five specifically involves: The volume values ​​corresponding to each time position are extracted along the full-cycle volume curve in chronological order, and the volume change is calculated based on the difference between the volume value corresponding to the previous time position and the volume value corresponding to the next time position, thus forming a volume change sequence. Based on the direction of increase or decrease of volume change at each time position in the volume change sequence, divide the volume increase interval into volume increase interval and volume decrease interval, and determine the time position at which the direction of volume change changes. The peak position is determined at the time position where the volume increase interval intersects with the volume decrease interval; the trough position is determined at the time position where the volume decrease interval intersects with the volume increase interval; and the turning point is determined at the time position where the volume change changes from increasing to decreasing and from decreasing to increasing. The peak position, trough position, and turning point are checked for consistency with the changes in adjacent time positions, local neighborhood stability, cross-phase consistency, or inter-chamber physiological consistency. Peak positions, trough positions, and turning points that meet the condition of continuous change are retained, and the retained results are determined as key phases.

7. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, Step six specifically involves: Extract the corresponding volume value of each key phase in the full-cycle volume curve from the key phases, and record them according to the time position of the key phases in the standardized cardiac cycle coordinate system to form a key phase volume record. The maximum and minimum volume values ​​are determined in the key phase volume records. The stroke volume is calculated based on the difference between the maximum and minimum volume values. The ejection fraction is calculated based on the ratio of the stroke volume to the maximum volume value, thus forming a record of cardiac function parameters. Quality control and reliability assessment are performed based on key phase volume records, full-cycle volume curves, and cardiac function parameter records. The time position of key phases, continuity of volume changes, and range of cardiac function parameters are checked to identify abnormal phases, abnormal curves, and abnormal parameters, and to form quality control results. Based on the quality control results, closed-loop processing is performed on abnormal phases, abnormal curves, and abnormal parameters. The closed-loop processing includes one of the following: removing abnormal phases and interpolating compensation based on neighboring phases, reducing the weight of abnormal results, performing restrictive correction on local curves, retaining results and outputting warning information, marking abnormal areas, and prompting manual review. After the closed-loop processing is completed, the key phase volume records, cardiac function parameter records, and quality control results are collected according to time position to form at least one of the following: a confidence score, an abnormality cause code, and a suggested review information, and a cardiac function assessment result is formed.

8. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, The physiological consistency constraints include at least one of the following: logical constraints on inter-chamber volume changes, approximately conservation constraints on left and right ventricular output, and reasonable constraints on volume changes before and after key phases.

9. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, When multiple competing extreme points, local pseudo-extreme points, or phase conflicts exist, abnormal extreme points are excluded based on local neighborhood stability, cross-phase consistency, or inter-compartment physiological consistency, and unstable extreme points are downweighted or marked as pending review.

10. The method for constructing cardiac chamber volume curves and assessing cardiac function using 4D-CT or 4D-CTA according to claim 1, characterized in that, The cardiac function assessment results include at least one of the following: cardiac function parameter values, full-cycle volume curve, key phase number or time position, quality control conclusion, confidence score, abnormality cause code, and phase number or heart chamber name recommended for review.

Images