Method and system for correcting contrast-enhanced images

Abstract

A method and system for correcting for contrast agent density differences in a sequence of contrast-enhanced image frames. A reference image frame is defined in the sequence of contrast-enhanced image frames, and segmentation is performed on the reference image frame to determine the location of a region of interest within the reference image frame. The region of interest is a region of the reference image frame that contains contrast agent. Other image frames in the sequence of contrast-enhanced images are corrected based on differences in contrast agent density / image intensity in the region of interest relative to the region of interest in the reference image frame.

Description

Technical Field

[0001] This invention relates to the field of correcting contrast-enhanced images, and more particularly to the field of correcting differences in contrast agent density in a sequence of contrast-enhanced image frames. Background Technology

[0002] In diagnostic applications based on medical imaging, clinicians frequently compare images acquired at different time frames. In the case of contrast-enhanced imaging, the degradation and dilution of the contrast agent means that images of the same anatomical features corresponding to different time frames within the acquisition will often show varying levels of opacity. This makes image comparison difficult for clinicians.

[0003] Furthermore, when segmentation algorithms (such as model-based segmentation) are applied to sequences of contrast-enhanced image frames, the temporal variability of contrast agent density within the sequence can introduce errors. For example, a time-varying distribution of contrast agent can lead to the detection of boundaries between regions with different contrast agent densities that do not correspond to physiological structures. This means that automated determination of clinical information cannot be reliably obtained from sequences of contrast-enhanced image frames.

[0004] These are particular problems in contrast-enhanced imaging applications, where the temporal variability of the contrast agent is further complicated by physiological processes such as the cardiac cycle. For example, as new blood containing undamaged contrast agent flows into the left atrium and then into the left ventricle, the temporal variability of the contrast agent in the left ventricular opacity (LVO) is caused by both the partial destruction of the contrast agent over time due to its interaction with emitted ultrasound waves and the changes in contrast agent density during the cardiac cycle.

[0005] Therefore, there is a need for a method to reduce the effects of contrast agent degradation and / or dilution in a sequence of contrast-enhanced image frames. Summary of the Invention

[0006] This invention is defined by the claims.

[0007] According to an example of one aspect of the present invention, a computer-implemented method is provided for correcting differences in contrast agent density in a sequence of contrast-enhanced image frames.

[0008] The computer-implemented method includes: selecting a reference image frame from the sequence of contrast-enhanced image frames; performing segmentation on the reference image frame; identifying a region of interest in the reference image frame based on the segmentation, wherein the region of interest is a region of the reference image frame containing the contrast agent; and correcting for differences in contrast agent density in a set of one or more image frames in the sequence of image frames based on changes in image intensity within the region of interest between each of the one or more contrast-enhanced image frames and the reference image frame, wherein the set of one or more image frames includes at least one image frame different from the reference image frame.

[0009] Even in cases with complex time-dependent contrast agent density (such as contrast-enhanced cardiac ultrasound), this method allows for the correction of differences (i.e., variations) in contrast agent density between image frames in a sequence of contrast-enhanced image frames. In other words, the embodiments propose using a region of interest from a reference image to correct for inter-frame differences in contrast agent density. Correcting for differences in contrast agent density means that image frames can be compared more easily and the reliability of interpreting features in image frames is improved because it reduces the need for clinicians to distinguish between effects caused by the time dependence of the contrast agent and effects caused by physiologically relevant processes.

[0010] The sequence of contrast-enhanced image frames can be based on one-dimensional (1D), two-dimensional (2D), or three-dimensional (3D) images.

[0011] The reference image frame can be selected as one with a high average contrast agent density, making it more suitable for segmentation than one with a lower average contrast agent density. Once the correction has been applied, segmentation can be reliably performed on frames where the contrast agent density is too low for segmentation purposes. Using the segmentation from the reference image frame to identify the region of interest improves the reliability of correctly identifying the appropriate region for the region of interest.

[0012] A region of interest (ROI) in a reference image frame can be defined as a region suitable for tracking contrast agent density across image frames. The ROI may include one or more anatomical structures containing the majority of the contrast agent, and can be defined such that the region is expected to include the same one or more anatomical structures in each image frame of the sequence.

[0013] The inventors have recognized that basing the correction on changes in image intensity within a region of interest (ROI) across a sequence of images, rather than on intensity values ​​from a complete image frame, results in more accurate correction. When the ROI is selected as a region containing contrast agent, the image intensity within the ROI acts as a proxy for the contrast agent density within that ROI.

[0014] This correction can be applied to all image frames in a sequence of contrast-enhanced image frames, or to a subset of the sequence (i.e., not all). For example, the correction can be applied to a subset of the sequence that includes all image frames following the reference image frame.

[0015] In a particular example, the set of one or more image frames in the sequence of contrast-enhanced images includes multiple image frames. In some examples, the set of one or more image frames does not include the reference image (because this provides reference information for other image frames in the correction sequence).

[0016] The correction of the contrast agent density difference in a set of one or more image frames in the sequence of image frames is based on the change (i.e., difference) in image intensity within the region of interest between each of the one or more contrast-enhanced image frames and a reference image frame. In other words, the difference in image intensity between the first image frame and the reference image frame is used to correct the image intensity in the first image frame.

[0017] Optionally, the sequence of contrast-enhanced image frames is a sequence of ventricular opacity image frames, such as left ventricular opacity image frames or right ventricular opacity image frames.

[0018] This method is particularly suitable for correcting left ventricular opacity (LVO) images, such as LVO ultrasound images, because the density of contrast agent in the left ventricle varies significantly and in a complex manner over time. This temporal variability is due to blood circulation, and also due to the destruction of the contrast agent by ultrasound waves when imaging is performed using one or more ultrasound transducers.

[0019] The reference image frame can be selected by: performing an initial segmentation on a subset of the first M frames in the sequence including the contrast-enhanced image frames, where M is a predetermined number; determining, based on the initial segmentation, which image frame in the subset corresponds to the first end-diastolic frame; and selecting the determined image frame as the reference image frame.

[0020] The first end-diastolic frame in a sequence of LVO images typically has the maximum or near-maximum contrast agent density, thus forming a suitable reference image frame.

[0021] The region of interest may include the estimated location of the blood pool in the reference image frame. The blood pool contains most of the contrast agent in the left ventricular opacity and is typically the first anatomical structure in the left ventricle to respond to changes in contrast agent density, thus making it a suitable location for the region of interest. Furthermore, the simple shape of the blood pool allows for easier identification and tracking of image intensity within it compared to, for example, thin structures such as myocardium.

[0022] The step of identifying the region of interest may further include: identifying end-contraction frames in the sequence of contrast-enhanced image frames; estimating the location of the blood pool in the identified end-contraction frames; and defining the region of interest as a region that includes the estimated location of the blood pool in the reference image frame and the estimated location in the end-contraction frames.

[0023] In this way, the region of interest can be defined such that it can be expected to be located in the blood pool across all image frames in the sequence. This allows for tracking of contrast agent density in the blood pool across sequences.

[0024] The reference image frame can be selected by determining which image frame among the first N frames in the sequence of contrast-enhanced image frames has the maximum average intensity, where N is a predetermined number; and selecting the determined image frame as the reference image frame.

[0025] Contrast agent density generally decreases over time, therefore the earliest image frames in the expected sequence will include those with appropriately high contrast agent densities. The image frame with the highest average intensity among these frames will correspond to the image frame with the highest contrast agent density.

[0026] N can be a predetermined number, which is preferably less than 10, for example, preferably less than 5.

[0027] The reference image frame can be one of the following image frames in the sequence of contrast-enhanced image frames, in which the contrast agent density is highest. The image frame with the highest contrast agent density is the most suitable frame for performing segmentation.

[0028] The step of correcting the set of one or more image frames in the sequence of image frames may include: defining a nonlinear transfer function for intensity values ​​for each image frame in the set of one or more image frames, such that the average uncorrected image intensity of the region of interest in the image frame is mapped to the average image intensity of the region of interest in the reference image frame; and applying the nonlinear transfer function defined for the image frame to at least a portion of the image frame for each image frame in the set of one or more image frames.

[0029] The nonlinear transfer function is designed to modify images so that they resemble a hypothetical scenario in which the density of the contrast agent does not change over time, thereby producing a more uniform appearance across a sequence of contrast-enhanced image frames.

[0030] The continuous mapping of intensity in the nonlinear transfer function means avoiding the introduction of spurious intensity edges during the correction process.

[0031] The step of correcting the set of one or more image frames in the sequence of image frames may include: defining a function based on histogram matching for each image frame in the set of one or more image frames, such that the gray value statistics of at least a portion of the image frame are consistent with the gray value statistics of the region of interest in the reference image frame; and applying the function defined for the image frame to at least a portion of the image frame for each image frame in the set of one or more image frames.

[0032] This method aligns the grayscale value distribution between the image frame being corrected and the reference frame to a higher quantile.

[0033] For each image frame in a set of one or more image frames, a nonlinear transfer function can be applied to the entire image frame or (only) a portion of the image frame, preferably where the portion includes the region of interest. The temporal variability of the contrast agent primarily affects the image intensity in anatomical structures containing the contrast agent; these regions are typically found close to the region of interest. In the context of this application, "portion" does not include the entire image.

[0034] The segmentation performed on the reference image frame can be model-based segmentation. Alternatively, any other suitable segmentation method can be used, such as a deep learning-based voxel segmentation method.

[0035] A computer program product comprising code units is also proposed, which, when executed on a computing device having a processing system, cause the processing system to perform all steps of any of the methods described herein.

[0036] A processing system for correcting differences in contrast agent density in a sequence of contrast-enhanced image frames is also proposed. The processing system is configured to: select a reference image frame from the sequence of contrast-enhanced image frames; perform segmentation on the reference image frame; identify a region of interest (ROI) in the reference image frame based on the segmentation, wherein the ROI is a region of the reference image frame containing the contrast agent; and correct differences in contrast agent density in a set of one or more image frames in the sequence of image frames based on changes in image intensity within the ROI between each of the one or more contrast-enhanced image frames and the reference image frame, wherein the set of one or more image frames includes at least one image frame different from the reference image frame.

[0037] The reference image frame may be one of the following image frames in the sequence of contrast-enhanced image frames, in which the contrast agent density is highest.

[0038] A system is also proposed, comprising: an imaging device for acquiring medical images of a subject; and a previously described processing system further configured to receive a sequence of contrast-enhanced image frames from the imaging device. The imaging device may be, for example, an ultrasound imaging device, a magnetic resonance (MR) imaging device, a computed tomography (CT) imaging device, an X-ray imaging device, or any imaging device suitable for contrast-enhanced imaging.

[0039] These and other aspects of the invention will be apparent and illustrated by referring to one or more embodiments described below. Attached Figure Description

[0040] Examples of the invention will now be described in detail with reference to the accompanying drawings, in which:

[0041] Figure 1 The illustration depicts a method for correcting differences in contrast agent density in a sequence of contrast-enhanced image frames according to an embodiment of the present invention;

[0042] Figure 2 The illustration shows examples of different levels of opacity between the first end-diastolic frame and the first end-systolic frame in a sequence of left ventricular opacity image frames;

[0043] Figure 3 The illustration shows the segmentation of a reference image frame and the identification of a region of interest in a sequence of left ventricular opacity images according to an embodiment of the present invention.

[0044] Figure 4 The figure shows a graph of contrast agent density versus time for an example sequence of contrast-enhanced image frames;

[0045] Figure 5 The figure shows a graph of the average image intensity in the example region of interest versus the image frame index of a sequence of left ventricular opacity image frames;

[0046] Figure 6 An example nonlinear transfer function for correcting differences in contrast agent density in contrast-enhanced image frames according to an embodiment of the present invention is illustrated.

[0047] Figure 7 The illustration shows example results of a sequence of image frames corrected using a nonlinear transfer function for left ventricular opacity; and

[0048] Figure 8 The illustration shows a system for acquiring and correcting a sequence of contrast-enhanced image frames according to an embodiment of the present invention, the system comprising an imaging apparatus and a processing system. Detailed Implementation

[0049] The invention will be described with reference to the accompanying drawings.

[0050] It should be understood that the detailed descriptions and specific examples, while indicating exemplary embodiments of the apparatus, systems, and methods, are intended for illustrative purposes only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems, and methods of the invention will be better understood from the following description, the appended claims, and the accompanying drawings. It should be understood that the drawings are merely schematic and not drawn to scale. It should also be understood that the same reference numerals are used throughout the drawings to denote the same or similar parts.

[0051] According to the concept of the present invention, a method and system for correcting contrast agent density differences in a sequence of contrast-enhanced image frames are proposed. A reference image frame is defined in the sequence of contrast-enhanced image frames, and segmentation is performed on the reference image frame to determine the location of a region of interest (ROI) within the reference image frame. The ROI is the region of the reference image frame containing the contrast agent. Other image frames in the sequence of contrast-enhanced images are corrected based on the difference in contrast agent density / image intensity in the ROI relative to the ROI in the reference image frame.

[0052] The embodiments are based, at least in part, on the understanding that correcting for differences in contrast agent density based on variations or differences in intensity values ​​in appropriate regions of an image frame provides a more accurate correction than correction based on intensity values ​​of the entire image frame, and that the appropriate region on which the correction is based can be identified by first identifying a reference image frame suitable for segmentation and performing segmentation on the reference image frame to identify anatomical structures containing the contrast agent.

[0053] Illustrative embodiments may be employed, for example, in contrast-enhanced medical imaging systems (such as contrast-enhanced cardiac ultrasound systems).

[0054] Figure 1 The illustration depicts a method 100 for correcting differences in contrast agent density within a sequence of contrast-enhanced image frames. The sequence of contrast-enhanced image frames can be any sequence of contrast-enhanced image frames obtained during medical imaging acquisition. For example, the sequence can be a sequence of contrast-enhanced ultrasound image frames, such as a sequence of ventricular opacity image frames, e.g., left ventricular opacity image frames or right ventricular opacity image frames.

[0055] The method begins with step 110, in which a reference image frame is selected from a sequence of contrast-enhanced image frames. The reference image frame is an image frame with a high contrast agent density; for example, the reference image frame may be one of the following image frames in the sequence of contrast-enhanced image frames, in which the contrast agent density is highest.

[0056] Various methods for selecting a suitable reference frame in a sequence of contrast-enhanced images are envisioned. In some examples, the reference frame can be determined directly from the image data corresponding to the sequence of contrast-enhanced image frames by identifying the image frame with the highest average intensity in the sequence. A large average intensity indicates a high contrast agent density.

[0057] It is expected that the image frame with the highest contrast agent density will be found early in the sequence of contrast-enhanced image frames, as contrast agent density typically decreases over time. Therefore, in some examples, only the first few image frames of the sequence can be used to select a reference image frame. For example, it is possible to determine which of the first N image frames in the sequence has the maximum average intensity, where N is a predetermined number, and that frame can be selected as the reference image frame. N can be, for example, a number less than 10, such as less than 5. In the example, N = 3.

[0058] In other examples, knowledge of how physiological processes affect contrast agent density can be used when selecting an appropriate reference image frame. For instance, in cases of left / right ventricular opacity, contrast agent density increases during cardiac diastole as new blood containing contrast agent flows into the left / right ventricle, and then decreases again as the ultrasound used to obtain the contrast-enhanced image frame interacts with the contrast agent.

[0059] The proposed method is particularly useful for left ventricular opacity due to the influence of the lungs on contrast agent density, as the contrast agent (typically introduced via intravenous injection) tends to be diluted as blood flows through / through the lungs. However, it should be understood that the advantages of the invention are achieved when used in both left and right ventricular opacity techniques.

[0060] Figure 2 The illustration shows examples of different levels of opacity between the first end-diastolic (ED) frame 200 and the first end-systolic (ES) frame 250 in a sequence of left ventricular opacity image frames. Those skilled in the art will understand that a sequence of right ventricular opacity image frames can be processed in a functionally equivalent manner.

[0061] like Figure 2 As shown, the contrast agent density is significantly higher in the first ED frame than in the first ES frame. Due to the degradation and dilution of the contrast agent over time, a higher contrast agent density is expected in the first ED frame than in subsequent ED frames. In a sequence of left ventricular opacity image frames, the first ED frame typically has the highest or near-highest contrast agent density.

[0062] Therefore, the first ED frame in the sequence of left ventricular opacity image frames is a suitable choice for the reference image frame. Thus, a reference image frame can be selected from the sequence of left ventricular opacity image frames by identifying the first ED frame and selecting it as the reference image frame. For example, initial segmentation (such as model-based segmentation) can be performed to determine which image frame in the sequence corresponds to the first ED frame.

[0063] Since the first ED frame will be found to be oriented towards the beginning of the sequence, initial segmentation can be performed only on the first few image frames. For example, an initial model-based segmentation can be performed on a subset of image frames, including the first M frames in a sequence containing left ventricular opacity image frames, where M is a predetermined number. Then, based on the initial model-based segmentation, it can be determined which image frame in the subset corresponds to the first ED frame, and the image frame determined to correspond to the first ED frame can be selected as a reference image frame.

[0064] Return to Figure 1 In step 120, segmentation is performed on the selected reference image frame. Segmentation can be, for example, model-based segmentation, but alternative segmentation methods are also envisioned, such as AI-based segmentation methods (i.e., machine learning methods, such as neural networks).

[0065] Because the reference image frame is selected to have a high contrast agent density, it will generally be well-suited for segmentation without the need for image correction.

[0066] At step 130, a region of interest (ROI) is identified in the reference image frame based on segmentation. An ROI is a region, portion, or fraction of the reference image frame that contains contrast agent. For example, an ROI may correspond to one or more anatomical structures that typically contain a large portion of contrast agent, identified in segmentation based on known parameters or literature. The identity of an ROI may be predetermined, for example, based on a specific use case scenario.

[0067] The Region of Interest (ROI) preferably forms only a portion of the entire reference image frame, i.e., not the entire reference image frame. It will be apparent that the ROI (when identified within the reference image frame) can be mapped to the same location / region in every other image frame of the sequence of contrast-enhanced image frames. Therefore, the ROI identified in the reference image frame corresponds to a ROI of the same size and location in every other image frame of the sequence.

[0068] In the case of left ventricular opacity, for example, the Region of Interest (ROI) can include the estimated location of the blood pool in a reference image frame. The estimation of the blood pool's location is based on the segmentation results. Therefore, the ROI can (in this scene) correspond to the portion of the image representing the blood pool. The location of this portion of the image can be identified using the segmentation results.

[0069] As another example, the ROI can be the location of a lesion or tumor. For instance, a tumor or lesion in the brain may cause leakage or rupture of the blood-brain barrier, meaning that contrast agents present in the blood will leak into the brain regions surrounding the tumor / lesion. This causes the appearance of the tumor / lesion to be highlighted in the image frame.

[0070] Other suitable examples of regions of interest containing most of the contrast agent will be apparent to those skilled in the art, and will depend on the use case scenario (i.e., implementation details).

[0071] In a sequence of contrast-enhanced images, the size and position of anatomical structures can vary across the frame sequence. For example, in a sequence of images showing the opacity of the left ventricle, the anatomy of the heart changes in size and position throughout the cardiac cycle. As another example, in a series of images of the brain, changes in blood flow or patient position may indicate movement within the brain.

[0072] In some examples, this variation is taken into account when determining the size and location of the ROI. For instance, for a sequence of left ventricular opacity image frames, when defining the ROI, the estimated size and location of the blood pool at end-systole can be used in addition to the estimated location of the blood pool in the reference image frame. In this way, the ROI can be defined such that it lies (at least approximately) within the blood pool at all image frames in the sequence, assuming that the image frames in the sequence are approximately aligned.

[0073] Various methods are envisioned for identifying suitable regions of interest (ROIs) based on the estimated location of the blood pool in a reference image frame, as well as the estimated end-systolic (ES) size and location of the blood pool. In one example, ES frames are identified in a sequence of left ventricular opacity images based on, for example, an initial segmentation used to identify the first ED frame. The location and size of the blood pool in the identified ES frames can then be estimated using a difference pattern between the ED and ES frames, averaged across several datasets. The ROI can be defined as having the estimated location and size of the blood pool in the ES frame. Alternatively, the size and location of the ROI in the reference image can be obtained by scaling the estimated ES blood pool relative to its centroid using a scaling factor f, where f < 1 (e.g., f = 0.8). Scale-downscaling the estimated ES blood pool to define the ROI increases the likelihood that the ROI of interest will remain entirely within the blood pool across all frames, as there will be some uncertainty in the estimation of the blood pool's size and location due to uncertainties in both the segmentation of the first ED frame and the mapping from the difference pattern to the ES frame. In another example, the ROI can be defined using a suitable morphological operation (e.g., erosion).

[0074] Therefore, the region of interest can be defined by identifying the region containing the contrast agent of the reference frame, and each subsequent frame in the sequence of predicted frames for that region has a corresponding region of interest containing the contrast agent (at the same location and size). The description provided above provides an example of this method.

[0075] Figure 3 The illustration shows the segmentation of reference image frames and the identification of Regions of Interest (ROIs) in a sequence of left ventricular opacity images as described above. Image 300 shows the segmentation of the reference image frame used to estimate the location of the blood pool 305. Image 350 shows the identification of ROI 355. ROI 355 is smaller than the blood pool 305 in the reference image frame because it has been defined as being within the blood pool in all frames of the left ventricular image sequence as described above.

[0076] Return to Figure 1 At step 140, differences in contrast agent density are corrected within a set of one or more image frames in the sequence of contrast-enhanced images. The set of one or more image frames includes at least one image frame that is different from a reference image. In a particular example, the set of one or more image frames in the sequence of contrast-enhanced images includes multiple image frames. In some examples, the set of one or more image frames does not include a reference image (because this provides reference information for correcting other image frames in the sequence).

[0077] The purpose of correction is to modify one or more images to resemble or simulate (e.g., as closely as possible) how they would look if the contrast agent density were the same as in a reference image frame. Therefore, inter-frame differences in contrast agent density can be corrected. Figure 4 The image illustrates this idea.

[0078] Figure 4 A schematic graph 400 shows the contrast agent density versus time for an example sequence of contrast-enhanced image frames. The density of the uncorrected image frame 410 varies over time. The reference image frame has been selected as the frame with the highest contrast agent density. The corrected image frame 420 has the same contrast agent density as the reference image frame.

[0079] The correction applied to each image frame is based on the change in image intensity within the Region of Interest (ROI) between the image frame and the reference image frame. Using only the image intensity of the ROI, rather than the intensity of the entire image, improves the accuracy of the correction.

[0080] Therefore, the difference between the image intensity in the ROI of the reference image frame and the ROI of the image frame to be changed (at the same location and size) is used to correct the image frame to be changed.

[0081] Figure 5This illustrates how average image intensity within a ROI can be used as a representation of contrast agent density. Figure 5 The figure 500 illustrates a graph of the average image intensity in an example ROI versus the image frame index of a sequence of left ventricular opacity image frames. As shown in Figure 500, the average image intensity in the ROI illustrates the expected variability of contrast agent density during the cardiac cycle.

[0082] In some examples, the difference in contrast agent density is corrected by defining a nonlinear transfer function for the intensity values ​​of each of one or more image frames. This nonlinear transfer function is designed such that the average image intensity of the ROI in the uncorrected image frame is mapped to the average image intensity of the ROI in the reference image frame.

[0083] Figure 6 The diagram illustrates an example nonlinear transfer function that can be used to correct differences in contrast agent density between contrast-enhanced image frames. The intensity, equal to the average ROI intensity in the image frame being corrected, is adjusted to a corrected intensity equal to the average ROI intensity in the reference image frame. This continuous mapping of intensity in the function allows for the correction of image frames without introducing spurious intensity edges. Furthermore, note that both the minimum and maximum intensity values ​​(0 and 255 here, respectively) remain unchanged through the transfer function; this ensures that no overall offset of intensity values ​​is applied to the image.

[0084] In other examples, histogram matching methods can be used to correct one or more image frames, wherein the grayscale statistics of the image frame being corrected (or a region thereof, as described below) are consistent with the grayscale statistics of the ROI in the reference frame. In other words, the histogram matching function can be defined such that the grayscale statistics of at least a portion of the image frame being corrected are consistent with the grayscale statistics of the region of interest in the reference image frame. Compared to the methods described above that are based solely on average intensity values, this method also aligns to higher quantiles of the grayscale distribution between the image being corrected and one of the reference frames. Further methods for correcting differences in contrast agent density based on variations in image intensity within the ROI between the image frame and the reference image frame will be apparent to those skilled in the art.

[0085] The corrected image frames may include all image frames in the sequence except for the reference image frame. Alternatively, correction may be applied only to a subset of the sequence of contrast-enhanced images. For example, the corrected image frames may include all image frames in the sequence following the reference image frame, or the corrected image frames may include only the image frames selected by the clinician for comparison.

[0086] Correction for each of one or more image frames can be applied to the entire image frame, or the application of correction can be limited to a portion of the image frame. For example, correction can be applied only to regions / portions of the image that are close to and / or include the Region of Interest (ROI), since regions close to the ROI are generally more likely to contain contrast agent than regions far from the ROI, and are therefore generally more susceptible to the temporal variability of the contrast agent. For example, correction can be applied entirely within the ROI and smoothly transition to the identity function far outside the ROI (i.e., no correction), where the transition parameter for each voxel is the nearest distance to the ROI.

[0087] Figure 7 The diagram illustrates the usage reference. Figure 6 Example result 750 of a sequence 700 of image frames with described nonlinear transfer function correction for left ventricular opacity. In this example, the correction for each image has been applied to the entire image.

[0088] like Figure 7 As shown, in the uncorrected sequence 700, reference frame 710 has the highest contrast agent density, and the contrast agent density decreases over time (as the sequence progresses from left to right). In the corrected sequence 750, the contrast agent density is approximately constant across the sequence.

[0089] Figure 8 The illustration shows a system 800 for acquiring and correcting a sequence of contrast-enhanced image frames according to an embodiment of the present invention. The system 800 includes an imaging device 810 and a processing system 820. The processing system 820 itself is an embodiment of the present invention.

[0090] Imaging device 810 is a medical imaging device configured to acquire a sequence 815 of contrast-enhanced image frames of object 830. Figure 8 In this context, the imaging device is an ultrasound probe used to obtain images of the opacity of the left ventricle, but any imaging device suitable for obtaining contrast-enhanced image frames can be used, such as an MRI system, a CT system, etc.

[0091] The processing system 820 includes one or more processors configured to receive a sequence 815 of contrast-enhanced image frames from the imaging device 810 and to correct differences in contrast agent density in the sequence.

[0092] Processing system 820 corrects the difference in contrast agent density by selecting a reference image frame from sequence 815, performing segmentation (e.g., model-based segmentation) on the reference image frame, identifying regions of interest (ROIs) in the reference image frame based on the segmentation, and correcting one or more image frames in the sequence based on changes in image density within the ROIs between each of the one or more image frames and the reference image frame, as described above. Figure 1 As described.

[0093] In some embodiments, the processing system 820 may select an image frame with the highest contrast agent density in sequence 815 as a reference image frame. The processing system may select a reference image frame by determining which image frame among the first few frames in the sequence has the highest average intensity. In the case of a sequence of left ventricular opacity image frames, the processing system may select a reference image frame by identifying the first end-diastolic frame (e.g., by performing initial segmentation on the first few frames in the sequence).

[0094] In some embodiments, when sequence 815 is a sequence of left ventricular opacity images, processing system 820 can identify blood pools as regions of interest. The processing system can estimate the location of the blood pools in a reference image frame based on the segmentation results. The processing system can also define the region of interest by identifying end-systolic frames in sequence 815, estimating the location of the blood pools in the end-systolic frames, and defining the region of interest as including the estimated locations of the blood pools in both the reference image frame and the end-systolic frames.

[0095] In some embodiments, the processing system 820 can correct one or more images in sequence 815 by defining a nonlinear transfer function for intensity values ​​for each image frame and applying the nonlinear transfer function defined for each image frame to at least a portion of the image frames. The nonlinear transfer function is defined for each image frame such that the average uncorrected image intensity of the region of interest in the image frame is mapped to the average image intensity of the region of interest in a reference image frame. In some examples, the processing system may apply the nonlinear transfer function to the entire image frame. In other examples, the processing system may apply the nonlinear transfer function only to the portion of the image that includes the image frame.

[0096] For the purpose of improving understanding of the background, the embodiments described in this detailed description have focused on sequences of left ventricular opacity images. However, those skilled in the art will understand that the proposed mechanism can be configured to be used with any suitable sequence of image frames (e.g., a sequence of right ventricular opacity image frames).

[0097] It should be understood that the disclosed methods are computer-implemented methods. Therefore, the concept of a computer program is also proposed, which includes code units for implementing any of the described methods when the program is run on a processing system.

[0098] Any advantages of the computer-implemented methods according to the present invention are similarly and analogously applicable to the systems and computer programs disclosed herein.

[0099] Those skilled in the art will understand that two or more of the above-described options, embodiments, and / or aspects of the present invention can be combined in any manner deemed useful.

[0100] Various aspects of the present invention can be implemented in a computer program product, which may be a collection of computer program instructions stored on a computer-readable storage device and executable by a computer. The instructions of the present invention can reside in any interpreted or executable code mechanism, including but not limited to scripts, interpreted programs, dynamic link libraries (DLLs), or Java classes. These instructions can be provided as a complete executable program, a partial executable program, as a modification (e.g., an update) to an existing program, or as an extension (e.g., a plugin) to an existing program. Furthermore, the various parts of the processing of the present invention can be distributed across multiple computers or processors.

[0101] As described above, the processing unit (e.g., a controller) implements the control method. The controller can be implemented in various ways using software and / or hardware to perform a variety of desired functions. A processor is an example of a controller employing one or more microprocessors, which can be programmed using software (e.g., microcode) to perform desired functions. However, a controller can be implemented with or without a processor, and can also be implemented as a combination of dedicated hardware performing some functions and processors (e.g., one or more programmed microprocessors and associated circuitry) performing other functions.

[0102] As described above, the system utilizes a processor to perform data processing. Processors can be implemented in various ways, using software and / or hardware, to perform a variety of required functions. Processors typically employ one or more microprocessors that can be programmed using software (e.g., microcode) to perform desired functions. A processor can be implemented as a combination of dedicated hardware performing some functions and one or more programmed microprocessors and associated circuitry performing other functions.

[0103] Examples of circuits that may be employed in various embodiments of this disclosure include, but are not limited to, conventional microprocessors, application-specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0104] In various implementations, the processor may be associated with one or more storage media, such as volatile and non-volatile computer memories, such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when run on one or more processors and / or controllers, perform the desired functions. The various storage media may be fixed within the processor or controller, or may be transportable, allowing one or more programs stored thereon to be loaded into the processor or controller.

[0105] Those skilled in the art, through studying the accompanying drawings, the disclosure, and the claims, will be able to understand and implement other variations of the disclosed embodiments when practicing the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality. Although specific measures are recited in dissimilar dependent claims, this does not imply that combinations of these measures cannot be advantageously used. No reference numerals in the claims should be construed as limiting the scope.

Claims

1. A computer-implemented method (100) for correcting differences in contrast agent density in a sequence (700, 815) of contrast-enhanced image frames, the computer-implemented method comprising: Select a reference image frame from the sequence of the contrast-enhanced image frames (710). Segmentation is performed on the reference image frame; Based on the segmentation, a region of interest (355) is identified in the reference image frame, wherein the region of interest is the region of the reference image frame containing the contrast agent; and The difference in contrast agent density in the set of one or more image frames in the sequence of image frames is corrected based on the change in image intensity in the region of interest between each of the one or more contrast-enhanced image frames and the reference image frame, wherein the correction modifies each image frame in the set of one or more image frames to simulate how the image frame would look if the contrast agent density were the same as in the reference image frame, wherein the set of one or more image frames includes at least one image frame that is different from the reference image frame.

2. The computer-implemented method (100) according to claim 1, wherein, The sequence of angiography-enhanced image frames (700, 815) is a sequence of ventricular opacity image frames.

3. The computer-implemented method (100) according to claim 2, wherein, The reference image frame (710) is selected in the following manner: An initial segmentation is performed on a subset of the first M frames in the sequence of image frames including the contrast-enhanced image frames, where M is a predetermined number; Based on the initial segmentation, determine which image frame in the subset of image frames corresponds to the first end-diastolic frame; and The determined image frame is selected as the reference image frame.

4. The computer-implemented method (100) according to claim 2 or 3, wherein, The region of interest (355) includes the estimated location of the blood pool (305) in the reference image frame (710).

5. The computer-implemented method (100) according to claim 4, wherein, The step of identifying the region of interest (355) further includes: Identify the terminal contraction frames in the sequence (700, 815) of the contrast-enhanced image frames; Estimate the location of the blood pool in the identified end-contraction frames; and The region of interest is defined as the region that includes the estimated location of the blood pool in the reference image frame and the estimated location in the end-contraction frame.

6. The computer-implemented method (100) according to claim 1 or 2, wherein, The reference image frame (710) is selected in the following manner: Determine which of the first N frames in the sequence of contrast-enhanced image frames has the maximum average intensity, where N is a predetermined number; and The determined image frame is selected as the reference image frame.

7. The computer-implemented method (100) according to any one of claims 1 to 6, wherein, The reference image frame (710) is the following image frame in the sequence of contrast-enhanced image frames (700, 815), in which the contrast agent density is highest.

8. The computer-implemented method (100) according to any one of claims 1 to 7, wherein, The step of correcting the set of one or more image frames in the sequence of image frames (700, 815) includes: For each image frame in the set of one or more image frames, a non-linear transfer function for intensity values ​​is defined such that the average uncorrected image intensity of the region of interest in the image frame is mapped to the average image intensity of the region of interest in the reference image frame. For each of the set of one or more image frames, the nonlinear transfer function defined for the image frame is applied to at least a portion of the image frame.

9. The computer-implemented method (100) according to any one of claims 1 to 7, wherein, The step of correcting the set of one or more image frames in the sequence of image frames (700, 815) includes: For each image frame in the set of one or more image frames, a function is defined based on histogram matching such that the grayscale statistics of at least a portion of the image frames are consistent with the grayscale statistics of the region of interest in the reference image frame; and For each image frame in the set of one or more image frames, the function defined for that image frame is applied to at least a portion of that image frame.

10. The computer-implemented method (100) according to any one of claims 1 to 9, wherein, The difference in contrast agent density is corrected throughout the entire image frame of each image frame in the set of one or more image frames.

11. The computer-implemented method (100) according to any one of claims 1 to 9, wherein, The difference in contrast agent density is corrected in a portion of each image frame in the set of one or more image frames, wherein the portion includes the region of interest (355).

12. A computer program product comprising code units, which, when executed on a computing device having a processing system, cause the processing system to perform all the steps of the method according to any one of claims 1 to 11.

13. A processing system (820) for correcting differences in contrast agent density in a sequence (700, 815) of contrast-enhanced image frames, said processing system being configured to: Select a reference image frame from the sequence of the contrast-enhanced image frames (710). Segmentation is performed on the reference image frame; Based on the segmentation, the region of interest (355) in the reference image frame is identified, wherein... The region of interest is the region of the reference image frame that contains the contrast agent; and The difference in contrast agent density in the set of one or more image frames in the sequence of image frames is corrected based on the change in image intensity in the region of interest between each of the one or more contrast-enhanced image frames and the reference image frame, wherein the correction modifies each image frame in the set of one or more image frames to simulate how the image frame would look if the contrast agent density were the same as in the reference image frame, wherein the set of one or more image frames includes at least one image frame that is different from the reference image frame.

14. The processing system (820) according to claim 13, wherein, The reference image frame (710) is the following image frame in the sequence of contrast-enhanced image frames (700, 815), in which the contrast agent density is highest.

15. A system (800) comprising: An imaging device (810) for acquiring medical images of an object (830); as well as The processing system (820) according to claim 13 or 14 is further configured to receive a sequence of the contrast-enhanced image frames (700, 815) from the imaging device.

Images