Method for dynamically characterizing an object by analyzing ultrasonic signals

Abstract

Method for characterizing an object using an ultrasonic transducer, the method comprising the steps of: Acquiring (201) at a first instant, using the ultrasonic transducer, a first ultrasonic signal resulting from the reflection of an ultrasonic field emitted by the transducer on a region of interest including the object, Identifying (202) a point of interest in the first signal, Extracting (203) from the first signal a so-called reference portion including the point of interest, Acquiring (204) at a second instant, a second ultrasonic signal, Calculating (205) a similarity criterion between the reference portion and the second signal, Determining (206) the position of the point of interest in the second signal from the position in the second signal corresponding to the maximum of the similarity criterion and the relative position of the point of interest in the reference portion. Figure 2

Description

Title of the invention: Method for the dynamic characterization of an object by analysis of ultrasonic signals

[0001] The invention relates to the field of ultrasound imaging, in particular medical imaging and more specifically to a method for characterizing and tracking over time the position of points of interest of objects, in particular anatomical objects such as blood vessels.

[0002] In the field of medical imaging, there is a need to accurately characterize anatomical objects, in particular their position and size, and to track the evolution of these objects over time.

[0003] For example, there is a need for precise characterization of blood vessel dimensions, particularly arteries. Indeed, arteries are a true indicator of health, as they provide essential information about the state of the human body. Real-time arterial monitoring allows for the tracking of biological markers such as blood pressure to better understand the functioning of the human body and anticipate the development of diseases. The development of ultrasound-based technologies makes it possible to design compact imaging systems that offer a non-invasive solution. Real-time monitoring requires that the entire measurement chain, from acquisition and artery localization to the extraction of relevant information, be fully automated.

[0004] For example, arteries can be characterized by the detection of points of interest such as their center or points on the distal and proximal walls which then allow the measurement of their diameter.

[0005] Furthermore, the diameter of the arteries and the position of their walls can vary over time due to physiological parameters related to the individual, for example the individual's breathing cycle.

[0006] There is therefore a general need to accurately characterize points of interest of anatomical objects using ultrasound imaging techniques, but also to ensure follow-up over time of these points of interest.

[0007] Ultrasound imaging methods that offer solutions for characterizing blood vessels are most often focused on the longitudinal dimension of the vessel, for example the carotid artery.

[0008] These methods are diverse but can be grouped into three non-exhaustive categories which include manual, semi-automatic and automatic methods.

[0009] Manual methods rely on human expertise to manually determine the position of arteries, for example using graphic tools applied to ultrasound images.

[0010] In the case of semi-automatic methods, assistance algorithms operate in collaboration with a human. For example, the algorithm can continue to automatically locate the artery, relying on the human as a guide, who validates, corrects, and indicates reference information to help the algorithm find the correct location.

[0011] In the category of fully automatic methods, the algorithm automatically determines artery localization. Most of these automatic methods specifically address the problem of carotid artery localization and do not function optimally with other arteries. Indeed, the carotid artery is a less demanding case than other arteries such as the radial artery for several reasons. The carotid artery is larger with thicker walls and exhibits a different intensity profile than the radial artery, with less noise and more contrast. Since the diameter of the carotid artery is greater than that of the radial artery, ultrasound images exhibit better contrast. It is thus easier to locate the center of the carotid artery simply by searching for the area with the lowest intensity profile.Compared to the radial artery, ultrasound images of the carotid artery naturally show fewer anatomical features that could be mistaken for the artery itself. The larger size of the carotid artery also contributes to this, as it limits the fields of view of the ultrasound images, resulting in images with fewer noisy objects. Most methods developed for the carotid artery focus on the longitudinal dimension, further simplifying localization by simply identifying the horizontal trace with the lowest intensity profile. Other anatomical features have a less noisy footprint in this dimension. Most algorithms vertically locate the area with the lowest intensity by scanning the entire image; subsequently, the artery is located based on a consensus of the identified points after removing outliers.These methods, based on the longitudinal dimension, are sensitive to probe orientation and all require precise probe positioning on the skin during data acquisition. Uncertainty in orientation will result in inaccurate measurements of diameter and other properties. This problem can be circumvented by using multidimensional probes, but these are more expensive.

[0012] Although the transverse dimension is more difficult to treat, because the shape of anatomical objects in this dimension appears frontal and disrupts localization, particularly in the case of the radial artery, the advantage of the transverse dimension is that it does not require precise positioning of the probe on the skin, and the diameter measurements remain robust and reliable against probe orientation errors.

[0013] Existing solutions for automatic localization along the transverse dimension of arteries, or more generally blood vessels, are scarce and all suffer from one or more limitations. Methods based on deep learning are effective but consume a lot of energy, are not suitable for small portable systems, are complex to implement, and require enormous resources. Achieving real-time performance is difficult, especially on a portable computer. These methods are found integrated into assistive interfaces, for which time and resource requirements are not limited. Similarly, neural networks, characterized by their black-box nature, suffer from the problem of unpredictability. It has been shown that a single pixel in an image can lead a neural network to provide aberrant data.This problem is, in fact, the main reason slowing down the adoption of artificial intelligence in the medical field, and in areas with serious consequences.

[0014] Doppler-based solutions allow localization through frequency changes related to blood flow. These methods are limited to vertical localization, where the depth of the artery is determined. Horizontal localization is performed manually. Another drawback of these methods is that Doppler localization requires monitoring ultrasound frequencies over time; this implies that all these methods require the temporal dimension to localize the artery.

[0015] Other limitations are evident in several aspects, such as, for example, execution speed and compliance with real-time standards. Iterative search algorithms, for instance, are slow and require considerable resources to achieve real-time operation.

[0016] Algorithms requiring a reference model and prior information are limiting. Another aspect relates to algorithms that are sensitive to noise and require preprocessing to condition the received data before executing the localization method. Some existing solutions rely on landmarks such as contours, which makes them vulnerable to data imperfections. These methods often use preprocessing, consensus, and outlier removal algorithms to ensure robustness, which adds complexity and computational effort.

[0017] Other methods are applied to other technological solutions that fall outside the scope of ultrasound, such as intravascular ultrasound (IVUS) and coherence tomography (CTT) methods. Optical coherence tomography (OCT) is used in both methods. Although these methods process images of arteries, they do not address the same problem because the profile of the processed data differs in terms of intensity, noise, and complexity. Furthermore, these methods inherit the drawbacks of these technologies, such as invasiveness in the case of IVUS methods and depth limitations in the case of non-invasive OCT methods.

[0018] Patent application WO2016057233 describes a method for estimating a non-invasive continuous blood pressure waveform using ultrasound and an automated cuff. It belongs to the category of non-invasive continuous blood pressure measurements. The proposed system measures physical characteristics such as the geometry, elasticity, and deformation of a blood vessel, as well as other external physical parameters. Computer modeling and processing of the measured signals are used during the inflation and / or deflation of the cuff to iteratively estimate the subject's blood pressure. The ultrasound component allows for the estimation of the artery's diameter and the calculation of its elasticity.

[0019] The localization of the artery in this solution is limited by several factors: first, the transducer arrays are fixed by a qualified operator at an approximate location, and the artery is localized in three dimensions using an iterative search method based on assumed geometric characteristics and ultrasound response. The search is not only iterative, which is computationally intensive, but it is also performed in a 3D search space, thus involving a large amount of data. This limits the execution speed, particularly for devices with limited resources and power consumption. Furthermore, this 3D search is performed by comparing the results at each iteration with a reference artery model constructed using personal information about the subject, such as their age, height, etc.Besides requiring a reference model and prior information, this solution also necessitates a two-dimensional probe to enable acquisition in the 3D space on which the iterative search algorithm runs. With slow localization, refreshing the artery's position imposes a compromise on speed, accuracy, and the person's movement.

[0020] Document [1] presents a method for characterizing the carotid artery that can be applied to its longitudinal and transverse dimensions. It consists of a contour detection algorithm based on the "Snake" method and equipped with a mathematical artery reference model in the form of B-splines that deform in scale, translation, and rotation. The method converges to the carotid artery contours through an optimization algorithm. The convergence criterion is ensured by a sum of the absolute differences between the model and the image at each point in the image. The model parameters are optimized The model is iteratively analyzed using an energy minimization algorithm, and then the sum of absolute differences is calculated at each point. The circular shape of the reference model is used to simplify the calculation. A gradient-based optimization algorithm is applied to refine the precise segmentation of the detected artery contours.

[0021] In this document, the carotid artery is depicted as quite large and circular, and its wall is fairly thick, making it easy to distinguish the contours of the layers on which the algorithm is based. This solution is not applicable to other types of arteries, such as the radial artery, for several reasons. First, the radial artery often appears to be neither perfectly circular nor perfectly rectangular because of its small size. The wall of the radial artery is thin, making it impossible to distinguish its layers with standard probes at low center frequencies; this requires high-quality probes at higher ultrasound frequencies.For this reason, and given that the reference method [1] relies on differences in local contrast between inner and outer contours, it will be difficult for the algorithm to accurately locate the region of interest in cases where the artery or surrounding tissue lacks texture or contrast, as is the case with the radial artery.

[0022] Also, anatomical objects in transverse images of the carotid artery, such as the jugular vein, have shapes that are not very circular and are easily distinguishable from the carotid artery. For this reason, the algorithm is unlikely to converge on other objects. In the case of the radial artery, several anatomical objects have a shape and positioning very similar to the radial artery, which means that the method is likely to converge on other objects instead of the radial artery, because the criterion of the sum of absolute differences will also be satisfied, for example, by the veins neighboring the artery.

[0023] The method requires image enhancement preprocessing based on adaptive histograms to reduce noise. This is because noise can significantly increase the convergence time of the optimization algorithms on which the method is based. The convergence time is variable and depends on the intensity profiles of the acquired image, its complexity, the position of the artery, and its similarity to the model. It also depends on the initialization parameters integrated into the model. If the image contains complex anatomical structures or if the artery itself has irregular shapes, the comparison approach with the reference model may not function correctly within the required timeframe.

[0024] Also, the model may not be flexible enough to handle these variations, resulting in misalignment. For this reason, this method is more suitable for applications that do not require strict real-time constraints, and a Regular time-of-flight measurements are taken at fixed intervals. The method compares image pixels with a model at each image point, calculating the sum of absolute differences criterion for each pixel at each time. This operation is computationally intensive, necessitating the use of optimization algorithms to reduce computation time. The drawback of this technique is that it makes the solution entirely dependent on the convergence time of the optimization algorithm, which is an iterative algorithm dependent on image complexity and vulnerable to convergence problems such as local minima. Furthermore, the optimization may converge towards veins rather than the artery in transverse scans of the radial artery.

[0025] The method presented in document [2] relates to the localization of the carotid artery in transverse images. The method is based on the circular Hough transform, a method used to detect lines in images that has been adapted for detecting circles. The method processes sequences of 5 to 15 consecutive images. Each image is preprocessed to optimize brightness and contrast. Then, a powerful Gaussian filter reduces noise. Although several fine morphological structures are destroyed in this process, the carotid artery appears clearer. Next, the Hough transform detects all dark circles by considering all rays that fall within a reasonable range. The coordinates of the detected centers (xi, yi) and the rays (ri) are collected. For each of the detected dark circles, the brightness values ​​of their pixels are evaluated.The circle with the darkest values, i.e., the minimum brightness, is elected as the "candidate circle," and its center coordinates (xci, yci) along with a radius (rci) are stored in a matrix of potential candidates. Once all the B-mode images have been processed, the matrix contains the results obtained from each image. The most frequent triplet (xc, yc, rc) appearing in the matrix is ​​selected as the final choice.

[0026] This third method is also unsuitable for the radial artery for the following reasons. The carotid artery has a fairly circular shape; it stands out as a rather dark area and is easily distinguished from other anatomical objects. The shape of the radial artery on ultrasound images does not appear perfectly circular. Moreover, there are several anatomical objects similar to the artery, for example, veins, which are nearby, sometimes darker than the radial artery, and also have a circular shape. The Hough transform can be adapted to other, more complex shapes for application to the radial artery; however, this considerably increases the computational intensity, because the circular shape as presented in document [2] already performs a search in 3D space, which requires much more computation than with the Hough transform. standard. Indeed, for each circle diameter, the method calculates a new accumulator across the entire image. The radial artery can easily deform, for example, into an elliptical shape, which implies an additional parameter to be considered by the Hough transform. The shape of the radial artery is much more distorted than that of the carotid artery because its size is smaller, resulting in smaller contours that are therefore more distorted by the noise observed in ultrasound images. The Hough transform provides multiple detections, and choosing based on the darkest area will not work in the case of the radial artery because neighboring veins may be darker. These multiple detections cause a large number of false detections, which necessitates either the use of more complex criteria or consideration of the temporal dimension to improve localization robustness.For this reason, the method requires 5 successive images. The circular Hough transform requires an edge detection step and threshold adjustments. Edge detection is vulnerable to noise in the image, which implies the need for preprocessing enhancement and noise filtering. For this reason, the method described in the document specifies that strong Gaussian filtering is necessary.

[0027] In general, the aforementioned prior art methods also do not address the problem of temporal tracking of points of interest of the artery detected at a given time.

[0028] There is therefore a need for a new method of detection and temporal tracking of points of interest of anatomical objects by ultrasound imaging which is low complexity and which makes it possible to overcome small variations of ultrasound signals over time which are linked to physiological variations of said objects.

[0029] The proposed method is based on defining a reference segment around a point of interest in an acquired ultrasonic signal and then calculating the similarity between this reference segment and a new ultrasonic acquisition at a later time.

[0030] The invention makes it possible to take into account the overall response of the point of interest or more generally of the area of ​​interest in the ultrasonic signal and to follow the evolution of this response over time.

[0031] The invention has the advantage of enabling dynamic in-vivo monitoring of a point of interest on an anatomical object. For example, the invention makes it possible to obtain the variation in the diameter of an artery and the movement of its walls over time.

[0032] Unlike existing methods, the proposed method allows for simple and highly efficient real-time management of signal loss problems, interference between characteristic signal peaks, and interference related to measurements of person movements, uncertainties and anatomical variabilities, with accuracy and a low sampling frequency.

[0033] The principle of the method consists of finding a similarity between local information measured on the ultrasound signal, for example, peaks, points / areas of interest, and global information, resulting in high accuracy. For example, if the anatomical object is an artery, one or more points of interest are identified on the signal, such as the positions of peaks related to the interface between the blood and the arterial wall, allowing the internal diameter of an artery to be determined by the difference in the time of flight of the ultrasound waves. The method consists of extracting segments around these points of interest to use as a reference. The coordinates of the points of interest in the reference frame of each corresponding segment are stored.Subsequently, for each new signal acquired, a similarity calculation is applied between each reference segment and the new signal, allowing the new position of the segment to be located by similarity measurement. The position of the point of interest, for example the proximal or distal peak of the internal wall, in the frame of reference of the wall segment is used to update the new positions dynamically over time.

[0034] The proposed method is of low complexity and can be implemented in real time on resource-constrained embedded devices.

[0035] The invention relates to a method for characterizing an object using an ultrasonic transducer, the method comprising the steps of: - To acquire, at a first instant, using the ultrasonic transducer, a first ultrasonic signal resulting from the reflection of an ultrasonic field emitted by the transducer on an area of ​​interest including the object, - Identify a point of interest in the first signal, - Extract from the first signal a so-called reference portion including the point of interest, - Acquire, at a second instant, a second ultrasonic signal, - Calculate a similarity criterion between the reference portion and the second signal, - Determine the position of the point of interest in the second signal from the position in the second signal corresponding to the maximum of the similarity criterion and the relative position of the point of interest in the reference portion.

[0036] According to a particular aspect of the invention, the similarity criterion is calculated for a portion of the second signal including the reference portion and the position in the second signal corresponding to the maximum of the similarity criterion is determined relative to the beginning of said portion.

[0037] In one embodiment, the method further comprises the steps of: i. Evaluate whether the reference portion is obsolete according to a predefined obsolescence criterion, ii. If the reference portion is obsolete, replace it with a portion of the second signal including the point of interest.

[0038] According to a particular aspect of the invention, the obsolescence criterion consists of comparing a parameter measured on the similarity criterion to a predefined threshold value.

[0039] According to a particular aspect of the invention, the similarity criterion is an intercorrelation.

[0040] According to a particular aspect of the invention, the parameter measured on the similarity criterion is a function of the amplitude of the maximum of the cross-correlation.

[0041] According to a particular aspect of the invention, the object is an anatomical object, for example a blood vessel.

[0042] According to a particular aspect of the invention, the method is applied to two points of interest corresponding to a first point on the distal wall of a blood vessel and a second point on the proximal wall of the blood vessel, diametrically opposite to the first point, and the method further includes an estimation of the diameter of the blood vessel from the extrapolated times of flight of the two points.

[0043] According to a particular aspect of the invention, the ultrasonic signals are one-dimensional or two-dimensional, the position of the point of interest is defined by the coordinates of the point in the signal.

[0044] The invention also relates to an ultrasound imaging device comprising an ultrasound transducer and a processing unit configured to perform the steps of the method according to the invention.

[0045] Other features and advantages of the present invention will become more apparent from the following description in relation to the following accompanying drawings.

[0046] [Fig. 1] represents a diagram of a multi-element ultrasonic probe capable of performing an ultrasonic signal acquisition sequence to image an anatomical object,

[0047] [Fig.2] represents a flowchart describing the steps for implementing a method for characterizing an anatomical object according to an embodiment of the invention,

[0048] [Fig.3] represents a first example of an ultrasound signal measured for an artery radial at a given instant t,

[0049] [Fig.4a] illustrates the definition of reference segments on the signal of [Fig.3],

[0050] [Fig.4b] represents a diagram of a first corresponding reference segment to the characterization of a proximal artery wall,

[0051] [Fig.4c] represents a diagram of a second reference segment corresponding to the characterization of a distal artery wall,

[0052] [Fig.5a] represents a second example of an ultrasound signal measured for a radial artery at time t+1,

[0053] [Fig.5b] represents the superposition of the first ultrasonic signal and the second ultrasonic signal on the same time diagram,

[0054] [Fig.6a] represents an example of a first search space corresponding to a proximal wall of an artery in the second signal,

[0055] [Fig.6b] represents an example of a second search space corresponding to a distal wall of an artery in the second signal,

[0056] [Fig.7a] represents a result of an intercorrelation between the first reference segment and the first search space,

[0057] [Fig.7b] represents a result of an intercorrelation between the second reference segment and the second search space,

[0058] [Fig.8a] represents the superposition of the first search space with the first reference segment at the position of the maximum cross-correlation,

[0059] [Fig.8b] represents the superposition of the second search space with the second reference segment at the position of the maximum intercorrelation,

[0060] Figure 2 represents, on a flowchart, the main steps of implementation of a method for characterization and temporal monitoring of an anatomical object, according to an embodiment of the invention.

[0061] The invention is described for a non-limiting example where the anatomical object is a radial artery, but it extends to any type of anatomical object, in particular any type of blood vessel.

[0062] The method begins in step 201 with the acquisition of at least one ultrasonic signal.

[0063] This acquisition is carried out using a probe shown in [Fig. 1].

[0064] The acquisition probe is advantageously a linear multi-element probe 101, that is to say, one which comprises several ultrasonic elements 102, for example piezoelectric elements, aligned in a row. Alternatively, a two-dimensional probe, that is to say, comprising an array of ultrasonic elements, can also be used.

[0065] In the example of [Fig.1], the probe 101 is positioned in contact with the skin 103 so that the row of elements 102 is substantially in a transverse plane of an artery 104 to be imaged.

[0066] Advantageously, the acquisition carried out in step 101 is done in such a way that the active ultrasonic elements of the probe 101 are centered on the artery 104 so as to measure a signal corresponding to a path crossing the artery 104 in its center.

[0067] More generally, the acquisition is carried out in such a way that the ultrasound signal intercepts the anatomical object at one or more points of interest that we seek to characterize. In the case of artery 104, these points of interest are, for example, diametrically opposed points on the inner wall of the artery.

[0068] In one embodiment of the invention, the acquisition in step 101 is not limited to a single time-domain signal but aims to determine an ultrasound image of the artery. In this case, the acquisition 101 is performed by means of a successive scan during which, at each step, an ultrasound beam is emitted by a group of elements 102 comprising at least one element, in a direction perpendicular to the axis of the alignment of the elements. The same elements are active in reception to generate an ultrasound signal corresponding to a predefined acquisition time and to an axis substantially perpendicular to the group of elements 102.

[0069] This step allows the acquisition of an ultrasound signal. It is then iterated by shifting the active ultrasound elements by one element according to a sliding window, and then performing a new acquisition. Thus, by performing several successive acquisitions using a fixed-size group of elements that scans all the elements of the probe, several ultrasound signals are obtained which together form an ultrasound image of a transverse plane of artery 104.

[0070] This acquisition makes it possible to obtain in two dimensions a transverse ultrasonic imprint of the targeted radial artery as well as any other anatomical objects present around it such as veins.

[0071] Figure 3 shows an example of a 300 ultrasound signal measured for a radial artery, the signal corresponding to a path passing through the artery at its center. The time axis gives the position of the echo relative to the source, in other words its depth, via a simple time-distance conversion based on knowledge of the signal propagation speed.

[0072] In step 202 of the method, one or more points of interest are identified on the ultrasonic signal which correspond to corresponding points of interest of the anatomical object to be characterized.

[0073] For example, in the diagram in [Fig. 3], two peaks Pp and Pd of the signal have been identified, corresponding respectively to the reflections of the signal from the proximal and distal walls of the artery. From these two peaks, the diameter of the artery and the position C of its center can be deduced.

[0074] Step 202 can be performed by any suitable signal processing method for identifying a point of interest, in particular by thresholding. For example, it can be implemented using the methods described in the Applicant's French patent applications filed under numbers FR2314398 and FR2314348.

[0075] In step 203, for each identified point of interest, a reference segment is determined around the point of interest.

[0076] Figure 4a represents a first proximal reference segment SegP around the first point of interest Pp, this segment having PsegP as its starting point.

[0077] Similarly, a second distal reference segment SegD around the second point of interest Pd is represented, this segment having PsegD as its starting point.

[0078] Each segment corresponds to an overall signal response over a region of interest around the point of interest. In the example in [Fig. 4a], the regions of interest correspond to the entire proximal (respectively distal) wall of the artery, which includes various internal layers. A margin of tolerance can be provided at the beginning and end of the segment. The beginning of the reference segment PsegP, PsegD is chosen as the reference point for positioning the point of interest Pp, Pd. The coordinate of the point of interest in the reference segment is given by the relation: Pp = PsegP + DP, where Dp is the distance between the point of interest and the beginning of the segment. This distance serves as a reference for updating the position measurements of the reference point over time, as will be explained later.

[0079] Figures 4b and 4c respectively represent the two reference segments SegP and SegD extracted from the signal. The diagrams in Figures 4a, 4b, and 4c show the amplitude of the ultrasonic signal as a function of depth, expressed in samples.

[0080] The choice of the length of the reference segment depends on the characteristics of the area of ​​interest whose response in the ultrasound signal is to be captured. In the example of [Fig. 4a], the lengths of the reference segments are chosen so as to encompass the entire response of the whole of an artery wall.

[0081] In step 204, a new acquisition of an ultrasound signal is then performed at a later time t+1. [Fig. 5a] shows an example 500 of such a signal. [Fig. 5b] shows the superposition of signals 300 and 500. As can be seen in [Fig. 5b], slight changes appear, which are due in particular to movements of the artery between the two acquisition times.

[0082] An objective of the invention is to determine the position of the points of interest in the new signal 500 without again applying a complete detection method such as in step 202.

[0083] For this purpose, in step 205, for each point of interest, a similarity score is calculated between each reference segment and the new ultrasonic signal 500. In other words, the position in the new signal 500 that best corresponds to the shape of the signal in the reference segment is sought.

[0084] For example, the calculation of the similarity score is a calculation of cross-correlation between the ultrasonic signal and the reference segment.

[0085] This calculation can be performed on the entire new signal 500 or on a reduced search area encompassing a region of interest within the signal. Using a reduced search area has the advantage of limiting the computation time required to calculate the similarity score.

[0086] Fig. 6a shows a portion 501 of the new signal 500 corresponding to a search area of ​​the proximal wall of the artery.

[0087] Fig. 6b shows a portion 502 of the new signal 500 corresponding to a search area of ​​the distal wall of the artery.

[0088] Fig. 7a shows the result of an intercorrelation calculation between the first reference segment SegP and portion 501 of the signal corresponding to the search area of ​​the proximal wall of the artery.

[0089] Fig. 7b shows the result of an intercorrelation calculation between the second reference segment SegD and portion 502 of the signal corresponding to the search area of ​​the distal wall of the artery.

[0090] On each of the figures 7a, 7b the abscissa of the maximum of the cross-correlation function gives the position of the reference segment in the new signal 500.

[0091] Figures 8a and 8b show respectively the superposition of the reference segments and the portions 501, 502 of the signal, the reference segments being positioned at the position given by the maximum of the cross-correlation calculation.

[0092] If we denote CCd(t+l) the position of the maximum of intercorrelation for the distal search area, the position of the point of interest corresponding to the distal wall Pd in ​​the 502 portion of the signal is given by the relation:

[0093] Pd(t+1) = Rd(t+1) + CCd(t+1) + Dd

[0094] Rd(t+1) is the beginning of portion 502 of the distal search area in the signal 500,

[0095] Dd is the distance between the point of interest Pd and the beginning of the reference segment.

[0096] In the same way, the point of interest corresponding to the wall is determined proximal via the same relationship:

[0097] Pp(t+1) = Rp (t+1) + CCp(t+1) + Dp

[0098] Thus, in step 206, the position of the point of interest in the new ultrasonic signal acquired at time t+1 is determined via the previous relations.

[0099] Cross-correlation is calculated by varying the offset of a reference segment relative to the signal. Maximum cross-correlation is obtained for an offset that aligns the position of the reference segment with the position of the portion of the new signal that most closely resembles that segment. Cross-correlation is sufficiently robust without normalization. Because similarity is measured on two signals that are quite close in intrinsic properties, it is also possible to normalize the cross-correlation. The cross-correlation measure r between two signals Si and S2 is defined by the following formula for a given lag of 1.

[0100] 101011

[0102] d = argm®r(|yrils2[z]|

[0103] Where d is the translational position that maximizes the cross-correlation.

[0104] In an alternative embodiment shown in dotted lines in [Fig.2], the reference segments are updated when they are obsolete according to the following update mechanism.

[0105] In step 207, an obsolescence test is performed to determine if the reference segment is up to date.

[0106] In a simplest embodiment, this test simply consists of activating the update of the reference segment periodically according to a predefined time period.

[0107] In another embodiment, an obsolescence criterion is defined. This criterion is, for example, based on the analysis of the cross-correlation function calculated in step 205.

[0108] For example, the criterion consists of comparing the amplitude of the maximum of the cross-correlation function to a predefined threshold. When the amplitude is below the threshold, this indicates that the reference segment is less correlated with the current signal and should be updated. Alternatively, other parameters measured on the cross-correlation function may be taken into account, such as the variance, standard deviation, general shape of the function, or any mathematical criterion that allows the shape of the cross-correlation function to be characterized and compared to an expected theoretical shape that should correspond to a Dirac delta function at the position that maximizes the correlation between the two signals.

[0109] When the reference sequence is deemed obsolete, it is then replaced (step 208) by returning to step 203 to extract a new reference sequence on the current signal around the position of the point of interest detected in step 206.

[0110] In one embodiment of the invention, the similarity calculation performed in step 205 is carried out using criteria other than cross-correlation. For example, the cross-correlation calculation is replaced by a calculation of the Euclidean distance between the reference segment and the new signal, or by a calculation of mutual information, an entropy measurement, a statistical comparison criterion, or a mixture. of Gaussians or a criterion determined via an artificial intelligence model trained to predict a level of similarity between two signals.

[0111] Similarity calculation can be applied to descriptors of signals instead of the signals themselves.

[0112] Although the invention has been presented above for the case of one-dimensional ultrasonic signals, it also applies to multidimensional ultrasonic signals, for example, two-dimensional images. The image can be composed of several one-dimensional ultrasonic signals measured at several positions of the ultrasonic probe so as to image an area of ​​interest. The reference segments are replaced by 2D reference areas. The similarity calculation, for example, cross-correlation, is performed by comparing two 2D signals.

[0113] In another embodiment, when updating the reference segment, the new reference segment is calculated as a combination of several reference segments, for example, a weighted sum of these segments. References

[0114] [1] JH Gagan et al., “Automated Segmentation of Common Carotid Artery in Ultrasound Images," in IEEE Access, vol. 10, pp. 58419-58430, 2022

[0115] [2] « System A feasibility study of a PMUT-based wearable sensor for the automatic monitoring of carotid artery parameters », 2021 IEEE International Ultrasonics Symposium (IUS)

Claims

Demands

1. Method for characterizing an object using an ultrasonic transducer, the method comprising the steps of: - Acquiring (201) at a first instant, using the ultrasonic transducer, a first ultrasonic signal from the reflection of an ultrasonic field emitted by the transducer on an area of ​​interest including the object, - Identifying (202) a point of interest of the first signal, - Extracting (203) from the first signal a so-called reference portion including the point of interest, - Acquiring (204), at a second instant, a second ultrasonic signal, - Calculating (205) a similarity criterion between the reference portion and the second signal, - Determining (206) the position of the point of interest in the second signal from the position in the second signal corresponding to the maximum of the similarity criterion and the relative position of the point of interest in the reference portion.

2. Method of characterizing an object according to claim 1 wherein the similarity criterion is calculated (205) for a portion of the second signal including the reference portion and the position in the second signal corresponding to the maximum of the similarity criterion is determined relative to the beginning of said portion.

3. Method of characterizing an object according to any one of the preceding claims further comprising the steps of: - Evaluating (207) whether the reference portion is obsolete according to a predefined obsolescence criterion, - If the reference portion is obsolete, replacing it (208) with a portion of the second signal including the point of interest.

4. Method of characterizing an object according to claim 3 in which the obsolescence criterion (207) consists of comparing a parameter measured on the similarity criterion to a predefined threshold value.

5. Method of characterizing an object according to any one of the preceding claims wherein the similarity criterion (205) is an intercorrelation.

6. Method of characterizing an object according to claim 5 in combination with claim 4 wherein the parameter measured on the similarity criterion is a function of the amplitude of the maximum of the cross-correlation.

7. Method of characterizing an object according to any one of the preceding claims wherein the object is an anatomical object, for example a blood vessel.

8. Method of characterizing an object according to claim 7 wherein the method is applied to two points of interest corresponding to a first point on the distal wall of a blood vessel and a second point on the proximal wall of the blood vessel, diametrically opposite to the first point and the method further comprises an estimation of the diameter of the blood vessel from the extrapolated times of flight of the two points.

9. Method of characterizing an object according to any one of the preceding claims wherein the ultrasonic signals are one-dimensional or two-dimensional, the position of the point of interest is defined by the coordinates of the point in the signal.

10. Ultrasonic imaging device comprising an ultrasonic transducer and a processing unit configured to perform the steps of the method according to any one of the preceding claims.

Images