Infusion port catheter implantation length estimation method and system based on medical image recognition

Abstract

The invention discloses an infusion port catheter implantation length estimation method and system based on medical image recognition, particularly relates to the field of medical information processing, and is used for solving the problem of tip periodic drifting when an existing infusion port catheter is implanted. A continuous edge attenuation area is extracted, and a frame-level fuzzy boundary contour curve set is constructed; and performing phase calibration on the image frame in combination with a heart cycle signal, generating a phase stable contour cluster through phase clustering, and performing interpolation fitting on a contour point trajectory to generate a fuzzy domain deformation envelope diagram reflecting the maximum deformation range of the boundary. And further extracting a drift trajectory of the tip of the catheter in the envelope diagram, constructing a tip drift path cluster, identifying a high-frequency aggregation region through spatial density estimation, determining a stable interval band, calculating an extension length range of each path in the interval, and outputting an effective implantation length interval of the catheter.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical information processing, more particularly, the present application relates to a method and system for estimating the length of a catheter implanted in a port based on medical image recognition. BACKGROUND

[0002] In the process of implanting a central venous port catheter, the accuracy of the position of the catheter tip directly relates to the safety of drug delivery and the treatment effect. At present, the clinical community generally relies on enhanced CT or X-ray images to make a static position judgment of the catheter tip, and usually takes whether the catheter tip reaches the junction area of the superior vena cava and the right atrium as the qualified standard. However, this method ignores the periodic drift phenomenon of the catheter tip caused by physiological activities such as breathing and heartbeat, especially in the superior vena cava entrance area, the drift is particularly significant. In addition, due to the dense structure of this area, the static image often has the problems of boundary blur and signal attenuation, which further aggravates the catheter positioning error.

[0003] In some patients, even if the intraoperative image shows that the catheter position is compliant, postoperative stimulation arrhythmia or catheter dysfunction may still occur due to the displacement of the tip to the atrium, reflecting the problem that the existing estimation means lacks quantitative support for the "catheter drift tolerance range". Especially for the image signal noise fluctuation in the fuzzy area of the superior vena cava, there is still a lack of systematic modeling and utilization. At present, there is no scheme to establish a catheter drift tolerance model based on the characteristics of the imaging fuzzy area and to reversely deduce the optimal implantation length, and there is still a significant gap in the technical path. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a method for estimating the length of a catheter implanted in a port based on medical image recognition to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: The method for estimating the length of a catheter implanted in a port based on medical image recognition comprises the following steps: S1: Obtain an enhanced CT image sequence of the superior vena cava entrance segment, and extract a continuous edge attenuation area in the image sequence; S2: Construct a time-series fuzzy boundary profile change curve set based on the continuous edge attenuation area; S3: Collect a heart beat signal and calibrate the heart beat phase to which the image sequence belongs, and classify and integrate the fuzzy boundary profile change curve set through phase clustering; S4: Interpolate and fit the fuzzy boundary profile in the phase clustering to generate a fuzzy domain deformation envelope graph in a phase stable form; S5: Extract the drift trajectory of the catheter tip in the fuzzy domain deformation envelope graph, and construct a catheter tip drift path cluster; S6: constructing a stable interval band of the tip position according to the cluster of the drift path, and calculating the corresponding effective insertion length range of the catheter in the stable interval band.

[0006] In a preferred embodiment, in S1, the enhanced CT image sequence of the superior vena cava inlet segment is acquired, and the continuous edge attenuation region in the image sequence is extracted, which specifically includes: Acquiring an enhanced CT image sequence covering the spatial range from the superior vena cava to the posterior wall of the atrium of the patient; According to the anatomical position of the superior vena cava inlet, locating the cross-sectional frame set containing the superior vena cava inlet segment in the image sequence; Performing a multi-scale directional convolution operation on the cross-sectional frame set of the inlet segment to extract the edge response graph between the superior vena cava lumen and the adjacent tissue; Identifying the region with continuously decreasing edge intensity in the edge response graph, and constructing a candidate mask set of the continuous edge attenuation band; Performing spatial connectivity screening on the candidate mask set to exclude non-enclosed structure regions and retain closed regions with continuous structures as edge attenuation regions.

[0007] In a preferred embodiment, in S2, the time-series fuzzy boundary contour change curve set is constructed based on the continuous edge attenuation region, which specifically includes: Projecting the edge attenuation region boundary into a two-dimensional plane coordinate system to generate a frame-level boundary point set, and corresponding pairing the boundary points at the same spatial position in the order of frame index; Tracking the moving path of the boundary point evolution over time at each corresponding spatial position, and recording the boundary drift trajectory in the form of coordinate change sequence; Performing a spline interpolation operation on all boundary drift trajectories to generate a continuous time-series fuzzy boundary contour change curve set.

[0008] In a preferred embodiment, in S3, the heart beat signal is collected and the heart beat phase to which the image sequence belongs is calibrated, and the fuzzy boundary contour change curve set is classified and integrated by phase clustering, which specifically includes: Discrete time mapping processing is performed on the heart beat cycle signal collected based on the continuous electrical signal to generate a frame-level heart beat phase label sequence; Wherein, the heart beat phase refers to a specific time segment obtained by time axis segmentation of a complete heart beat cycle according to a preset division rule; Constructing a mapping structure between the fuzzy boundary contour change curve set and the frame-level heart beat phase label sequence, rearranging the fuzzy boundary contour change curve frames, and grouping all the contour change curve frames into phase label sets; The fuzzy boundary profile change curve frame sequence under the same phase label is discriminated based on the spatial distribution form similarity of the profile boundary points, and a phase stable profile curve cluster is generated by unsupervised clustering as the phase clustering fuzzy boundary profile.

[0009] In a preferred embodiment, the fuzzy boundary profile change curve frame sequence under the same phase label is discriminated based on the spatial distribution form similarity of the profile boundary points, and a phase stable profile curve cluster is generated by unsupervised clustering, which specifically includes: Each fuzzy boundary profile curve under the same phase label is uniformly sampled for profile boundary points to construct a point sequence set of uniform dimension; The Euclidean distance change sequence and the angle change sequence between adjacent points are calculated for each profile boundary point sequence, and the geometric feature trajectory of the spatial boundary change is extracted; Based on the geometric feature trajectory, a dynamic time warping operation is performed on all profiles to obtain a form deformation distance matrix between any two profile curves; Hierarchical clustering analysis is performed on the distance matrix, and the profile curves are grouped according to the preset deformation tolerance threshold to generate a similarity form clustering cluster as a phase stable profile curve cluster.

[0010] In a preferred embodiment, in S4, the fuzzy boundary profile in the phase clustering is interpolated and fitted to generate a fuzzy domain deformation envelope map under the phase stable form, which specifically includes: The boundary point set in each phase stable profile curve cluster is extracted, and the starting points of each profile curve are repositioned according to a uniform coordinate direction; The boundary points in the profile curve cluster are arranged in order according to their corresponding positions in the frame sequence to construct a multi-frame boundary point trajectory set; Each trajectory in the boundary point trajectory set is subjected to spatial curve interpolation fitting to generate a continuous deformation path between multiple frames, and the boundary maximum envelope is reconstructed based on the continuous deformation path; The boundary curve structure of the maximum envelope framework is closed to a closed three-dimensional curved surface graph, and the graph is defined as a fuzzy domain deformation envelope map under the corresponding heart phase.

[0011] In a preferred embodiment, in S5, the drift trajectory of the catheter tip in the fuzzy domain deformation envelope map is extracted to construct a catheter tip drift path set, which specifically includes: The catheter structure edge line bundle is identified in the enhanced CT image sequence, and the end point coordinates in each frame image are subjected to image frame extraction to generate a catheter tip point sequence; Each end point in the catheter tip point sequence is subjected to three-dimensional space remapping, and the tip point set located inside the phase deformation envelope graph is screened out; Reconstruct the spatial path lines arranged in ascending order of frame index based on the screened tip point set, and connect them on the time axis to form the catheter tip drift trajectory line; Apply an angle transformation rate sequencing process on the drift trajectory line, segment the trajectory sections with continuous directional trends in the path curvature sequence, and generate a set of trajectory structure fragments; Merge the trajectory structure fragments according to the point set similarity in spatial distribution, and classify the trajectory fragments that meet the continuity and distribution consistency standards into a unified path cluster, and output the catheter tip drift path cluster.

[0012] In a preferred embodiment, in S6, a stable interval band of tip position is constructed according to the drift path cluster, and the effective insertion length range of the catheter in the stable interval band is calculated, which specifically includes: Perform spatial density estimation on the tip endpoints of all paths in the drift path cluster, and construct a probability distribution map of the tip aggregation area; Extract the spatial closed region with continuous distribution density higher than the set threshold in the probability distribution map, and define it as the candidate region of the tip stable interval band; Calculate the projection distance of the intersection of each path in the drift path cluster and the boundary of the stable interval band as the line segment extension length of the corresponding path in the stable interval band; Merge the upper and lower boundaries of all line segment extension lengths to generate a set of insertion depth length intervals of the catheter tip in the stable interval band.

[0013] On the other hand, the application provides an infusion port catheter insertion length estimation system based on medical image recognition, which includes an image extraction module, a contour change construction module, a contour change clustering module, a deformation envelope map generation module, and a stable insertion estimation module: The image extraction module: collects enhanced CT image sequences covering the superior vena cava to the atrial posterior wall, locates the cross-sectional frames containing the superior vena cava inflow segment, performs multi-scale convolution to extract edge response maps, identifies gray continuous attenuation regions, and constructs edge attenuation regions through spatial connectivity screening; The contour change construction module: projects the edge attenuation region boundary to a two-dimensional coordinate system to generate frame-level boundary point sets, traces the path of point changes over time and interpolates, and constructs a set of fuzzy contour curves reflecting the dynamic changes of the boundary; The contour change clustering module: maps the contour change curve to the heart beat phase label, classifies and aggregates the contour sequence according to the label, and performs unsupervised clustering based on spatial morphological similarity to generate phase-stable contour curve clusters; The deformation envelope map generation module: extracts the boundary point trajectory of the stable contour cluster, performs interpolation fitting to reconstruct the deformation path, constructs the maximum envelope structure of the boundary, and generates a three-dimensional closed graph as the fuzzy domain deformation envelope map; Stable implantation estimation module: extract the catheter tip point sequence and screen the trajectory falling within the envelope diagram, identify the tip drift path segment and cluster, combine the spatial density distribution to construct the stable interval band, and output the effective implantation length range.

[0014] The medical image recognition-based infusion port catheter implantation length estimation method and system of the present application has the following technical effects and advantages: By extracting the continuous edge attenuation region of the superior vena cava inlet segment, combining the construction of the time sequence fuzzy boundary profile change curve and the dynamic calibration of the heart beat phase, the spatial drift behavior of the catheter tip under the physiological fluctuation background is modeled; by interpolating and fitting the fuzzy boundary profile in the phase clustering to generate a fuzzy domain deformation three-dimensional envelope diagram, the possible activity boundary of the catheter is effectively described; further, by combining the structural clustering of the catheter tip drift trajectory, the path cluster is constructed and the stable interval band is identified, thereby deducing the individualized effective catheter implantation length range, which has the technical advantages of strong dynamic adaptability, low anatomical dependence and high intraoperative guidance value. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The flowchart of the medical image recognition-based infusion port catheter implantation length estimation method of the present application is shown in the figure. Figure 2 The structural diagram of the medical image recognition-based infusion port catheter implantation length estimation system of the present application is shown in the figure. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0017] Embodiment 1 Figure 1 The medical image recognition-based infusion port catheter implantation length estimation method of the present application is given, which comprises the following steps: S1: Obtain the enhanced CT image sequence of the superior vena cava inlet segment, and extract the continuous edge attenuation region in the image sequence; S2: Construct a time sequence fuzzy boundary profile change curve set based on the continuous edge attenuation region; S3: Collect the heart beat signal and calibrate the heart beat phase to which the image sequence belongs, and classify and integrate the fuzzy boundary profile change curve set by phase clustering; S4: Interpolate and fit the fuzzy boundary profile in the phase clustering to generate a fuzzy domain deformation envelope diagram under the stable phase form; S5: Extract the drift trajectory of the catheter tip within the fuzzy domain deformation envelope, and construct a cluster of catheter tip drift paths; S6: According to the drift path cluster, construct a stable interval band of the tip position, and calculate the corresponding effective insertion length range of the catheter within the stable interval band.

[0018] In S1, enhanced CT image sequences of the superior vena cava inlet segment are obtained, and continuous edge attenuation regions in the image sequences are extracted.

[0019] Enhanced CT data acquisition is performed, and the acquisition range is set from the middle segment of the patient's superior vena cava to the posterior wall region of the right atrium to fully cover the expected activity segment of the catheter tip. The image acquisition uses a 64-slice spiral CT device, and a non-ionic contrast agent (iohexol) is injected. The dose and flow rate are automatically adjusted according to the patient's weight, and the acquisition time is the resting period of chest scanning. The image reconstruction resolution is set to 0.5mm x 0.5mm in the transverse axis, the layer thickness is 1mm, and the total number of image frames is controlled between 80-150 frames.

[0020] After the acquisition is completed, a frame-level image sequence matrix is constructed, and preprocessing is performed on all images, including window width and window level normalization, noise filtering (using median filtering), and body position correction (based on skeleton registration). Then, structure positioning is performed, combining with reference anatomical features such as the ascending aorta, right atrial wall, and intercostal space distribution, to calculate the index interval of the superior vena cava terminal segment and the right atrium junction region in the image sequence. The center point trajectory backtracking technique is used to track the vessel path from the middle segment of the SVC downward, and finally the transverse cross-section frame set containing the superior vena cava into the atrium segment is located, usually 20-30 frames, referred to as the analysis frame set.

[0021] For this frame set, multi-scale directional edge convolution operations are performed on each image. First, a Gaussian function is used to construct derivative convolution kernels in different directions, including 0°, 45°, 90°, and 135°, and the kernel size is adjusted to 3x3, 5x5, and 7x7 in three levels according to the scale space strategy. The maximum response direction of each pixel point in the convolution result is selected to form a direction-enhanced edge response map. Then, in the edge response map, a gray scale scanning line is generated in the direction of the vessel radius, the gray scale change curve is extracted, and the first derivative is calculated. When the gray scale intensity of a certain segment decreases continuously for multiple pixels, and the derivative value is stable within the set slope threshold interval, it is considered as an edge intensity attenuation region.

[0022] All pixel sets that meet the attenuation conditions are defined as candidate attenuation masks, forming the initial set of edge attenuation bands for this frame. A spatial connectivity analysis method is introduced to perform 8-neighborhood connectivity judgment on the candidate mask set, identify closed structures, and remove boundary breaks, insufficient areas, or disconnected regions, finally retaining edge attenuation closed regions with complete structure and surrounding catheter space. The closed region output of each frame is a binary mask map, and the center trajectory and boundary coordinate point set are stored.

[0023] In S2, a set of temporal fuzzy boundary profile variation curves is constructed based on the continuity edge attenuation region.

[0024] The closed boundary of each frame edge attenuation region is output in the form of a set of pixel coordinate points. To unify the reference system position of the boundary points in space, all boundary point sets are projected to a two-dimensional plane coordinate system (X-Y plane under the Axial view) registered with the CT image cross section. The boundary point set of each frame image is labeled as a frame-level boundary point set and stored in an ordered structure with the frame index as the label.

[0025] Based on the point position correspondence in the time dimension, the boundary points at the "same spatial position" are found in all frames. Using a polar angle uniform sampling method, the boundary curve is resampled into N uniformly distributed sampling points (e.g. N=360, one point per degree) with the centroid coordinates of each frame edge region as the polar coordinate center. Each frame's boundary profile is represented as a sequence of points with uniform point position numbers, with consistent numbering representing consistent polar angles, i.e. consistent approximate spatial positions.

[0026] In all frames, the same numbered points are paired to construct their trajectories with frame sequence (or time) variation. For example, the boundary point at a certain position with a frame number corresponds to a specific coordinate position in each frame of the image sequence. After arranging these coordinates in the time order of the frame index, a sequence of the boundary point's movement trajectory with time variation is formed, representing its position evolution process in the entire image sequence. This sequence reflects the position variation of the boundary position in the entire image sequence due to heartbeats or other physiological factors, i.e. the boundary drift trajectory.

[0027] For each boundary point drift trajectory, further smoothing processing is performed. In this embodiment, cubic spline interpolation is used to interpolate and reconstruct the coordinate changes in the x-axis and y-axis directions. Normalization is performed on the boundary point time index during interpolation to ensure that the interpolation process is performed on a unified time scale. After interpolation, each trajectory becomes a continuous spatial curve, reflecting the smooth drift behavior of the point in the entire frame sequence.

[0028] After completing the trajectory interpolation of all numbered points, the interpolation results of each frame are recombined into a new profile curve. Specifically, the corresponding point set at each time point on the interpolated trajectory curve is extracted, and these point sets are connected in the order of numbering to form a complete frame-level profile line. In this way, a continuous profile curve set sequence is formed on the time axis, with each curve composed of interpolated points, which has higher temporal consistency and smoothness compared to the jumping behavior of the original image profile points.

[0029] In S3, the heart beat signal is collected and the heart beat phase to which the image sequence belongs is calibrated, and the fuzzy boundary profile change curve set is classified and integrated through phase clustering.

[0030] While the patient is undergoing enhanced CT scanning, the patient's electrocardiogram signal is synchronously collected. The collection device is an electrocardiogram monitor with a synchronous recording function, and the sampling frequency is not less than 500 Hz. During the collection process, the peak point of the electrical activity signal of each heart beat start is automatically detected, and the time period between each adjacent peak point is defined as a complete heart beat cycle. The entire image collection time usually spans multiple heart beat cycles, and the system records the division points of these cycles on the time axis. The image frames are matched with the heart beat rhythm, and the CT device provides accurate collection time information for each image frame. The system matches the collection time point of each image frame to the heart beat cycle in which it is located, and further calculates the position of the image in the corresponding heart beat cycle. This embodiment divides each heart beat cycle into several equal-width time segments, for example, based on QRS onset, T wave peak, diastolic initial stage, etc. as the time segment division basis, and then according to the relative position of each frame collection time in the heart beat cycle, the frame is given a fixed phase label. After completing the heart beat phase labeling of all image frames, the frame sequence information with the labeled phase label is matched with the fuzzy boundary profile change curve set constructed earlier. Each image frame corresponds to a boundary profile curve, which is classified into the curve set under the corresponding phase label according to the time index. After the construction is completed, each heart beat phase segment contains a group of boundary profile curves extracted under similar rhythm background.

[0031] The fuzzy boundary profile change curve frame sequence under the same phase label is discriminated based on the similarity of the spatial distribution form of the profile boundary points, and a phase stable profile curve cluster is generated through unsupervised clustering, which is the specific fuzzy boundary profile in the phase clustering. The point set of each fuzzy boundary profile curve is standardized. Since the number of points of different profile curves in the original image may not be consistent, and the distance between discrete points is uneven, it will interfere with subsequent form analysis, so the profile point sequence resampling operation needs to be performed. This embodiment adopts the polar angle uniform sampling strategy: taking the geometric centroid of the profile as the reference center, the profile is projected into the polar coordinate system, and the boundary points are extracted at equal angle intervals, so that each profile is expressed as an ordered point sequence set with a uniform number of points (such as 128 points), and the point sequence number corresponds to a unified spatial orientation.

[0032] After obtaining the boundary point sequences in uniform dimension, the geometric variation features of each contour are extracted. Specifically, the Euclidean distance sequence between each point and its adjacent point on the contour is calculated to depict the local boundary line segment scale feature, and the angle variation sequence of the boundary direction is calculated as the quantitative expression of the curve bending degree. The two types of sequences together constitute the geometric feature trajectory describing the spatial structure variation of the contour, which is used to measure the deformation relationship between curves.

[0033] To compare the morphological differences between any two contour curves, dynamic time warping method is used to perform paired analysis on the geometric feature trajectories. The DTW algorithm allows elastic alignment of two feature sequences under the premise of non-linear point correspondence, thereby obtaining a relatively stable morphological deformation distance. After paired analysis of all curve pairs, a complete morphological distance matrix between contours is constructed, and the matrix elements represent the morphological variation degree between two curves.

[0034] After obtaining the distance matrix, hierarchical clustering method is used for contour clustering analysis. In the clustering process, the curve subsets are gradually merged according to the distance matrix value, and the termination condition in the clustering process is judged according to the system set deformation tolerance threshold (based on the contour construction accuracy, the default setting is the average distance). When the maximum morphological distance between contours does not exceed the tolerance range, it is considered that they have consistency in spatial morphology. Finally, multiple contour clustering clusters are formed, each cluster containing several contour curves with similar structure.

[0035] In S4, the ambiguous boundary contours in the phase cluster are interpolated and fitted to generate the ambiguous domain deformation envelope diagram under the phase-stable morphology.

[0036] Boundary point data is extracted from the obtained multiple contour curve clustering clusters, and each clustering cluster corresponds to a group of contour curves considered to be relatively stable in spatial structure. To ensure the formation of a unified direction boundary structure in space, the starting points of all contour curves need to be relocated. This embodiment uses contour reparameterization strategy: taking the contour centroid as the reference, the direction angle of the contour starting point and the centroid is calculated, and rotation and translation are performed on all curves to make their starting points in the same direction in the coordinate system. After such processing, the boundary point positions of different contours in time sequence can be guaranteed to have the same starting point alignment characteristics.

[0037] Subsequently, a "boundary point trajectory set" is constructed. Specifically, each contour curve is arranged in sequence according to the frame sequence, and its boundary points are paired in sequence to form a group of spatial point position correspondence between multiple frames. Taking the same numbered point group as a unit, the spatial positions of the points in different frames along the time axis are organized to generate a point position trajectory. Each trajectory represents the deformation path experienced by a fixed boundary position in the time sequence (i.e., in the frame sequence change).

[0038] A spline curve interpolation method is used to construct a continuous deformation path of each trajectory point sequence in the spatial dimension. The interpolated trajectory is smoothly transitioned on the frame axis and can reflect the boundary evolution trend of the structure induced by the heartbeat. After processing, the interpolation results of all trajectories are aggregated into a three-dimensional coordinate grid to form a spatial trajectory group describing the change of the contour outer edge under time continuity. Based on the outermost distribution of each point in the trajectory group in space, a convex hull fitting method and a direction scanning method are used to extract the maximum outer package structure of the point group in the three-dimensional space. The generated boundary shell is encapsulated into a closed three-dimensional curved surface graph by a cross-section closing algorithm, and the structure is defined as the "fuzzy domain deformation envelope" under the condition of the heartbeat phase. The internal structure of the graph contains all the point trajectories of the stable contour curve, and reflects the maximum deformation range. The closed structure is stored using point set grid encoding.

[0039] In S5, the drift trajectory of the catheter tip in the fuzzy domain deformation envelope is extracted, and a catheter tip drift path cluster is constructed.

[0040] An image enhancement processing method based on gray difference and structure refinement is used, and a multi-scale Frangi filter is used to perform line bundle enhancement on the image sequence to highlight the slender highlight area of the catheter in the blood vessel background. The processed image is subjected to region growing and morphological thinning algorithm to extract the catheter structure edge line bundle, and the farthest end point consistent with the image boundary direction in each frame is marked as the catheter tip position of the frame. The catheter tip coordinates identified in each frame are extracted in the order of image frame index to construct a catheter tip point sequence. To ensure that the extracted catheter trajectory is truly within the permissible area of the biological structure, each tip point is projected into the fuzzy domain deformation envelope generated in the previous step and a three-dimensional space remapping operation is performed. Specifically, the image coordinates (based on pixels) of the tip point are converted into actual space coordinates (based on CT image resolution and scanning parameters), and it is checked whether the point is inside the three-dimensional envelope graph closed surface body under the current phase condition. This embodiment uses a three-dimensional point-grid containment test algorithm, such as the ray casting method and the barycentric triangle method, to judge the envelope relationship of the point. Only the catheter tip points falling within the envelope graph are retained as the basic point set for constructing the effective drift trajectory.

[0041] After obtaining the continuous and effective tip point set, the points are connected in a three-dimensional space in the order of increasing image frame index to form a catheter tip drift trajectory line. The trajectory line represents the actual motion path of the catheter tip in the space in the continuous frames, and has time continuity. Considering that the heartbeat, breathing and slight body position changes of the patient may cause non-single direction stable changes of the trajectory, the trajectory is further analyzed for its morphological changes.

[0042] Specifically, a trajectory curvature analysis mechanism is introduced to calculate the angle change rate of each point on the trajectory in space, thereby forming a curvature change sequence. When the trajectory shows a trend of stable direction and small angle fluctuation amplitude in a certain segment of continuous point positions, it is considered that this segment can be classified as a same directional drift segment. The entire trajectory curve is dissected in this way to form a plurality of trajectory structure segment sets, and each segment represents the same directional stable movement behavior of the catheter tip within a continuous time period.

[0043] The trajectory structure segments are merged. Specifically, the spatial statistical features of each trajectory segment, such as the point set distribution center, direction vector, and distribution density, are extracted, and based on the Euclidean distribution overlap and direction consistency between point sets, the similarity index between segments is calculated. Hierarchical clustering is used to perform spatial clustering analysis on all segments, and based on the preset spatial distribution tolerance and direction variation threshold, the trajectory segments that meet the continuity and distribution consistency conditions are classified into a unified path cluster. The finally output catheter tip drift path cluster contains a plurality of trajectory sub-clusters, and the tip trajectories in each sub-cluster are continuous in time, have similar trends in space, and have segmented features that can be used for drift modeling and stable interval identification.

[0044] In S6, a stable interval band of the tip position is constructed based on the drift path cluster, and the effective insertion length range of the catheter in the stable interval band is calculated.

[0045] In the catheter tip drift path cluster that has been constructed, the endpoint coordinates of each path are extracted. Each path endpoint represents the final stable position of the catheter tip within a certain time period, and therefore all the endpoints in the entire path cluster constitute the spatial distribution samples of the catheter tip. After collecting all the endpoint coordinates, a three-dimensional point cloud set is constructed, and a spatial kernel density estimation operation is performed based on the point set. In the kernel density estimation process, a three-dimensional Gaussian kernel function is used to smooth the catheter tip endpoint point cloud to form a spatial continuous density field, and the density value represents the probability intensity of the tip point aggregation within a unit volume. The kernel bandwidth parameter is adaptively adjusted according to the spatial distribution range of the point cloud to ensure that the estimated probability distribution map can not only preserve the structural features of the concentrated area, but also avoid the loss of details caused by excessive smoothing. The processing generates a tip distribution probability value at each position in the three-dimensional space, i.e., a probability density map of the tip aggregation area.

[0046] In the density map, the system extracts the region with density value higher than the set threshold in the continuous region as the candidate region. The threshold can be set to more than 80% quantile of the density value in the overall distribution according to experience, or automatically adjusted according to the aggregation. All spatial position points that meet the density condition form a spatial closed region, and the geometric boundary of the region is extracted by connected component analysis and three-dimensional boundary generation technology, which is defined as the "catheter tip stable interval band candidate region". This region represents the spatial range where the catheter tip is most likely to stay in the dynamic state. After completing the extraction of the stable interval band, the crossing of each catheter path in the interval needs to be evaluated. For this purpose, the intersection of each path in the drift path cluster and the boundary of the stable interval band is identified, and the length of the spatial path segment from the entry point of the path into the interval band to the exit point of the interval boundary is calculated, which is defined as the "extension length" of the path in the interval band. If a path only exists in the interval in multiple non-continuous segments, the longest segment can be retained or the cumulative length of all segments can be performed to enhance the adaptability of the estimation.

[0047] The extension lengths of all paths in the stable interval band are analyzed, and the maximum extension length and the minimum extension length are extracted as the upper and lower boundaries of the current sample catheter placement length, that is, the "catheter effective placement length interval set" is constructed. The set can be expressed as a set of length interval values, or further statistical modeling can be performed to form a confidence interval expression as a reference for the depth of the catheter during the operation. Avoiding the over-reliance on static anatomical points in traditional placement depth judgment, and instead constructing a dynamic and reliable spatial tolerance band based on the actual drift behavior of the catheter tip, and combining the trajectory analysis results to generate individualized depth suggestions, has significant clinical adaptability, dynamic variability and modeling explanatory power, providing direct data support for navigation in infusion port implantation.

[0048] Embodiment 2 The difference between the embodiment 2 and the embodiment 1 of the present application is that the embodiment 2 introduces a catheter placement length estimation system based on medical image recognition for infusion port.

[0049] Figure 2 The structural schematic diagram of the catheter placement length estimation system based on medical image recognition for infusion port is given, which includes an image extraction module, a contour change construction module, a contour change clustering module, a deformation envelope generation module, and a stable placement estimation module: The image extraction module: collect enhanced CT image sequences covering the superior vena cava to the atrial posterior wall, locate the cross-sectional frame containing the superior vena cava inflow segment, perform multi-scale convolution to extract edge response maps, identify gray continuous attenuation regions, and construct edge attenuation regions through spatial connectivity screening; Contour variation construction module: projecting the edge attenuation region boundary to a two-dimensional coordinate system to generate a frame-level boundary point set, tracking the path of the point position over time and interpolating, constructing a fuzzy contour curve set reflecting the dynamic changes of the boundary; Contour variation clustering module: mapping the contour variation curve with the heart beat phase label, classifying and gathering the contour sequence according to the label, and performing unsupervised clustering based on spatial morphological similarity to generate phase-stable contour curve clusters; Deformation envelope map generation module: extracting the boundary point trajectory of the stable contour cluster, performing interpolation fitting to reconstruct the deformation path, constructing the maximum envelope structure of the boundary, and generating a three-dimensional closed graph as the fuzzy domain deformation envelope map; Stable placement estimation module: extracting the catheter tip point sequence and screening the trajectories falling within the envelope map, identifying the tip drift path segments and clustering, combining the spatial density distribution to construct a stable interval band, and outputting the effective placement length range.

[0050] The above formulas are all dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula closest to the actual situation. The preset parameters and threshold values in the formula are set by a person skilled in the art according to the actual situation.

[0051] The above embodiments can be realized wholly or partially by software, hardware, firmware or any other combination. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer readable storage medium or transferred from one computer readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center and the like containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0052] Those of skill in the art would understand that the modules and algorithms described in connection with the examples described herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. The

[0053] Those of skill in the art would understand that, for the purposes of description and enabling clarity, the specific process of the system, device and module described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0054] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, for example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed modules can be indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0055] The modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical modules, which can be located in one place or distributed on multiple network modules. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0056] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically, or two or more modules can be integrated in one module.

[0057] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0058] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0059] Finally: the above is only a preferred embodiment of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for estimating the length of a catheter placed in a port based on medical image recognition, characterized in that, The method comprises the following steps: S1: obtaining an enhanced CT image sequence of the superior vena cava inlet segment, and extracting a continuous edge attenuation region in the image sequence; S2: constructing a time-series fuzzy boundary profile change curve set based on the continuous edge attenuation region; S3: collecting a heart beat signal and calibrating a heart beat phase to which the image sequence belongs, and classifying and integrating the fuzzy boundary profile change curve set through phase clustering; S4: interpolating and fitting the fuzzy boundary profile in the phase cluster to generate a fuzzy domain deformation envelope graph in a phase-stable shape; S5: extracting a drift trajectory of the catheter tip in the fuzzy domain deformation envelope graph to construct a catheter tip drift path cluster; S6: constructing a stable interval band of the tip position according to the drift path cluster, and calculating a corresponding effective insertion length range of the catheter in the stable interval band.

2. The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S1, obtaining an enhanced CT image sequence of the superior vena cava inlet segment, and extracting a continuous edge attenuation region in the image sequence specifically comprises: Collecting an enhanced CT image sequence covering the space range from the superior vena cava to the posterior wall of the atrium; According to the anatomical position of the superior vena cava inlet, locating a cross-section frame set containing the superior vena cava inlet segment in the image sequence; Performing a multi-scale directional convolution operation on the cross-section frame set of the inlet segment to extract the edge response graph between the superior vena cava lumen and the adjacent tissue; Identifying the region with continuously decreasing edge intensity in the edge response graph to construct a candidate mask set of the continuous edge attenuation band; Performing spatial connectivity screening on the candidate mask set to exclude non-closed structure regions and retain closed regions with continuous structures as edge attenuation regions. 3.The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S2, constructing a time-series fuzzy boundary profile change curve set based on the continuous edge attenuation region specifically comprises: Projecting the edge attenuation region boundary onto a two-dimensional plane coordinate system to generate a frame-level boundary point set, and corresponding pairing the boundary points at the same spatial position in the order of frame index; Tracking the moving path of the boundary points at each corresponding spatial position over time and recording the boundary drift trajectory as a coordinate change sequence; Performing spline interpolation operation on all boundary drift trajectories to generate a continuous time-series fuzzy boundary profile change curve set. 4.The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S3, collecting a heart beat signal and calibrating a heart beat phase to which the image sequence belongs, and classifying and integrating the fuzzy boundary profile change curve set through phase clustering specifically comprises: Performing discrete time mapping processing on the heart beat cycle signal collected based on continuous electrical signal to generate a frame-level heart beat phase label sequence; Wherein, the heart beat phase refers to a specific time segment obtained by time axis segmentation of a complete heart beat cycle according to a preset division rule; Constructing a mapping structure between the fuzzy boundary profile change curve set and the frame-level heart beat phase label sequence, rearranging the fuzzy boundary profile change curve frame sequence, and grouping all profile change curve frames into phase label sets; Performing similarity judgment on the spatial distribution form of the fuzzy boundary profile change curve frame sequence under the same phase label based on the profile boundary point, and generating a phase-stable profile curve cluster through unsupervised clustering as the fuzzy boundary profile in the phase cluster.

5. The medical image recognition-based implantable port catheter length estimation method of claim 4, wherein, The fuzzy boundary profile variation curve frame sequence under the same phase label is discriminated based on the similarity of the spatial distribution form of the profile boundary points, and a phase-stable profile curve cluster is generated by unsupervised clustering, specifically comprising: Uniform sampling of the profile boundary points of each fuzzy boundary profile curve under the same phase label is performed to construct a point sequence set of uniform dimension; The Euclidean distance variation sequence and the angle variation sequence between adjacent points are calculated for each profile boundary point sequence, and the geometric feature trajectory of the spatial boundary variation is extracted; Based on the geometric feature trajectory, a dynamic time warping operation is performed on all profiles to obtain a morphological deformation distance matrix between any two profile curves; Hierarchical clustering analysis is performed on the distance matrix, and the profile curves are grouped according to the preset deformation tolerance threshold to generate a similarity morphological cluster as a phase-stable profile curve cluster.

6. The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S4, the fuzzy boundary profile in the phase cluster is interpolated and fitted to generate a fuzzy domain deformation envelope map under the phase-stable form, specifically comprising: Extract the boundary point set in each phase-stable profile curve cluster, and reposition the starting point of each profile curve in the uniform coordinate direction; Arrange the boundary points in the profile curve cluster in the order of their corresponding positions in the frame sequence to construct a multi-frame boundary point trajectory set; Perform spatial curve interpolation fitting on each trajectory in the boundary point trajectory set to generate a continuous deformation path between multiple frames, and reconstruct the maximum envelope based on the continuous deformation path; Close the boundary curve structure of the maximum envelope framework to a closed three-dimensional surface graph, and define this graph as the fuzzy domain deformation envelope map under the corresponding heart phase.

7. The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S5, a catheter tip drift trajectory set is constructed by extracting the drift trajectory of the catheter tip in the fuzzy domain deformation envelope map, specifically comprising: Identify the catheter structure edge line bundle in the enhanced CT image sequence, and perform image frame extraction on the end point coordinates in each frame to generate a catheter tip point sequence; Apply three-dimensional space remapping to each endpoint in the catheter tip point sequence to screen out the tip point set located inside the phase deformation envelope map; Reconstruct the spatial path line arranged in ascending order of frame index based on the screened tip point set, and connect it on the time axis to form a catheter tip drift trajectory line; Apply angle transformation rate sequencing processing to the drift trajectory line to segment the trajectory segments with continuous direction trend in the path curvature sequence, and generate a trajectory structure fragment set; According to the point set similarity of each trajectory structure fragment in the spatial distribution, perform merging operation, and classify the trajectory fragments meeting the continuity and distribution consistency standards into a unified path cluster to output the catheter tip drift path set. 8.The medical image recognition-based implantable port catheter length estimation method of claim 1, wherein, In S6, a stable interval band of the tip position is constructed based on the drift path set, and the effective insertion length range of the catheter in the stable interval band is calculated, specifically comprising: Perform spatial density estimation on the tip endpoints of all paths in the drift path set to construct a probability distribution map of the tip aggregation area; Extract the spatial closed area with continuous distribution density higher than the set threshold in the probability distribution map, and define it as a tip stable interval band candidate area; The intersection of each path in the drift path cluster and the boundary of the stable interval band is projected to calculate the distance, which is the length of the line segment extension of the corresponding path in the stable interval band; The upper and lower boundary merging of all line segment extensions generates the set of insertion depth length intervals of the catheter tip within the stable interval band range.

9. A medical image recognition based on estimation system of length of catheter placement for port, for realizing the medical image recognition based on estimation method of length of catheter placement of any one of claims 1-8, characterized in that, It includes an image extraction module, a contour change construction module, a contour change clustering module, a deformation envelope map generation module, and a stable insertion estimation module: The image extraction module: collects enhanced CT image sequences covering the superior vena cava to the atrial posterior wall, locates the cross-sectional frame containing the superior vena cava inflow segment, performs multi-scale convolution on the frame set to extract edge response maps, identifies gray continuous attenuation regions, and constructs edge attenuation regions through spatial connectivity screening; The contour change construction module: projects the edge attenuation region boundary to a two-dimensional coordinate system to generate a frame-level boundary point set, tracks the change path of the point position over time and interpolates, and constructs a fuzzy contour curve set reflecting the dynamic change of the boundary; The contour change clustering module: maps the contour change curve to the heart beat phase label, classifies and aggregates the contour sequence according to the label, and performs unsupervised clustering based on spatial morphological similarity to generate phase-stable contour curve clusters; The deformation envelope map generation module: extracts the boundary point trajectory of the stable contour cluster, performs interpolation fitting to reconstruct the deformation path, constructs the maximum envelope structure of the boundary, and generates a three-dimensional closed graph as the fuzzy domain deformation envelope map; The stable insertion estimation module: extracts the catheter tip point sequence and filters the trajectories falling within the envelope map, identifies the tip drift path segments and clusters them, constructs a stable interval band based on the spatial density distribution, and outputs the effective insertion length range.

Images