Processing laser speckle images

Abstract

A mechanism for processing one or more laser speckle images. The spatial frequency transformation is used to transform the one or more laser speckle images into spatial frequency data. The spatial frequency data is used to define a plurality of blood flow velocity indicators, each indicating an amount of blood moving at a particular velocity within the illuminated region of interest.

Description

Technical Field

[0001] The present disclosure relates generally to the field of laser speckle images, and in particular to the processing and analysis of laser speckle images. Background Art

[0002] The amount of microvascular dysfunction in subjects with peripheral arterial disease (PAD) is increasingly recognized as an important factor affecting their clinical outcome. Microvascular dysfunction affects the ability of the vasculature to respond to the changing oxygen demands in the parenchyma. The pathophysiological roots of vascular dysfunction are often difficult to determine and are thought to be caused by the interaction of different conditions including capillary ischemia, structural microangiopathy, or impaired hyperemic response.

[0003] A useful indicator of microvascular dysfunction is blood flow velocity. Laser speckle contrast imaging (LSCI), sometimes simply referred to as laser speckle imaging (LSI), has historically been used to construct a perfusion index, K, that varies more or less linearly with mean blood flow velocity.

[0004] The measurement principle is based on observing the speckle pattern of light reflected from the (light) perfused surface of the object, which is represented by the laser speckle image. The speckle pattern in the laser speckle image originates from the interfering light backscattered from the distribution of scattering particles. This includes not only the tissue matrix, but also any red blood cells residing in the tissue. The movement of the latter leads to the decorrelation of the reflected electric field, where the decorrelation time constant is proportional to the velocity of the red blood cells. Therefore, the speckle pattern in the laser speckle image will change in response to the velocity of the red blood cells (i.e., the blood flow velocity).

[0005] This results in an exposure time-dependent spatial blurring of the speckle pattern in the laser speckle image. This means that the standard deviation of the pixel values ​​σ(I) can be calculated from the average intensity of the pixel values ​​over the region of interest. Normalization is used to calculate the perfusion index K(I) of the laser speckle image of the region of interest as follows:

[0006]

[0007] There is a continuing desire to provide additional and / or more accurate clinical information. In particular, it would be advantageous to increase knowledge about blood flow velocity in a region of interest of a subject to aid in the diagnosis, assessment and treatment of microvascular dysfunction and / or PAD. Summary of the invention

[0008] The invention is defined by the claims.

[0009] A computer-implemented method for processing one or more laser speckle images is proposed, each laser speckle image being an image containing a speckle pattern produced by reflection of light illuminating a region of interest of an object, the computer-implemented method comprising: processing the one or more laser speckle images using a spatial frequency transform, thereby generating spatial frequency data; and defining a plurality of blood flow velocity indicators using the spatial frequency data, each blood flow velocity indicator indicating or representing an amount of blood moving at a corresponding different blood flow velocity within the region of interest.

[0010] The present disclosure recognizes that it would be particularly advantageous to determine an indicator that represents the amount or intensity of blood moving at different speeds. In particular, this can be used to determine or represent a blood flow velocity distribution. It should be recognized that the blood flow velocity distribution is particularly indicative of a subject's hyperemic response, such that changes in the blood flow velocity distribution can indicate an altered or impaired hyperemic response.

[0011] Therefore, it would be advantageous to generate data that changes in response to changes in the blood flow velocity distribution. This would allow identification of changes in the hyperemic response. Such data may also be processed to determine one or more characteristics of the subject, as there is a causal relationship between such data and the blood flow velocity distribution.

[0012] It should also be recognized that blood flowing in vessels of different sizes (i.e., vessels of different calibers or inner diameters) will produce an expected velocity distribution. Changes in the expected velocity distribution can reflect pathological patterns or problems, such as the presence of shunts or static microvascular dysfunction (e.g., caused by microangiopathy).

[0013] These methods are based on the recognition that the spatial frequency information of a (single) laser speckle image contains information about the blood velocity distribution. The present invention proposes not to calculate a single quantity K(I) from a laser speckle image or a spatial speckle pattern, but to calculate a frequency spectrum across all spatial frequencies and use this information to directly define a blood velocity index representing the amount or amplitude of blood moving at different blood flow velocities, thereby establishing more detailed perfusion state information.

[0014] Considering the frequency spectrum, the spatial standard deviation historically used to calculate the perfusion index K(I) is the average of the entire power spectrum. In contrast, rather than averaging the power spectrum across spatial frequencies (as is implicit in calculating K(I)), the present proposal evaluates the amplitude at different spatial frequencies. This provides additional and more clinically useful information, for example for pinpointing or identifying relevant blood velocity information.

[0015] It will be appreciated that multiple blood flow velocity indices will together provide information about the velocity distribution within a region of interest. The blood flow velocity distribution provides insights into the direction of blood flow, as velocity decreases as the size of the vessel caliber decreases. This information is useful for assessing or understanding the condition of a subject / patient. For example, if a disease causes increased shunting from the capillary bed, the proportion of slower moving red blood cells will decrease as blood bypasses the larger caliber AV confluence. This can be identified by changes in the blood flow velocity indices. Similarly, if the dynamic regulation of blood flow by vasomotor mechanisms (vasoconstriction / dilation) or sphincter regulation (shunting) is impaired, this can be observed not only in the behavior of the mean velocity, but also in the velocity distribution.

[0016] In some examples, the step of processing the one or more laser speckle images to generate spatial frequency data includes converting the one or more laser speckle images into a spatial power spectrum, wherein the spatial power spectrum identifies the power of the spatial frequency in the one or more laser speckle images for each of a plurality of spatial frequencies. Each power in the spatial power spectrum can serve as a corresponding blood flow velocity indicator.

[0017] Specifically, the magnitude of the spatial frequencies (ie, the power of the spatial power spectrum) can together serve as a multivariate marker or vector of the blood velocity distribution and perfusion of the tissue.

[0018] In some examples, the step of processing the laser speckle images to generate spatial frequency data includes: converting each laser speckle image into an initial spatial power spectrum, the initial spatial power spectrum identifying the power of a spatial frequency in the laser speckle image for each of a plurality of spatial frequencies; defining a plurality of spatial frequency ranges; and generating the spatial power spectrum by combining, for each of the plurality of spatial frequency ranges, the power of spatial frequencies in each initial spatial power spectrum that fall within the spatial frequency range to generate a power for the spatial power spectrum. For example, the combining may include summing, adding, or averaging the powers of spatial frequencies (in the same spatial frequency range).

[0019] This method effectively converts the initial spatial power spectrum into a spatial power spectrum by grouping the spatial frequency intervals into a fixed number of bins. This method allows for a higher signal-to-noise ratio while also compensating for different sampling resolutions in the spatial frequency domain by normalizing it to a fixed grid or a fixed set of spatial frequency ranges.

[0020] Thus, a defined plurality of spatial frequency ranges may be predetermined.

[0021] The computer-implemented method may also include processing one or more of the blood flow velocity indicators using a computer-implemented method to determine one or more values ​​of one or more characteristics of the object, wherein the one or more characteristics include at least one of: blood flow velocity distribution and / or clinical parameters such as pathological prognosis.

[0022] The method recognizes that the magnitude of spatial frequencies can also be converted to improved or other metrics (e.g., more clinically interpretable metrics) using one or more inference models that are trained to perform the conversion to provide better interpretable data to a clinician or further processing system. This is because spatial frequencies are responsive or responsive to changes in blood velocity, blood velocity distribution, and resulting clinical parameters (e.g., pathological prognosis or subject health). If determined, the pathological prognosis can be a prognosis of a predetermined or known pathology (e.g., a disease or other condition) of the subject.

[0023] Computer-implemented models may include machine learning models.

[0024] The computer-implemented method may include: obtaining a time-varying signal representing the temporal variation of blood flow in the region of interest; transforming the time-varying signal into the frequency domain; defining a set of frequency amplitudes, the set of frequency amplitudes containing amplitudes of different frequency components or frequency ranges of the time-varying signal; and determining one or more parameters of the irradiated region of interest using the blood flow velocity indicator and the defined set of frequency amplitudes.

[0025] Example parameters include blood flow velocity distribution and / or values ​​of the previously described clinical parameter(s).

[0026] The computer-implemented method may further include a laser speckle image definition process, the process including, for each laser speckle image: obtaining an initial laser speckle image, the initial laser speckle image being an image containing a speckle pattern generated by reflection of light irradiating a region of interest of an object, wherein the initial laser speckle image contains the laser speckle image and has a higher resolution than the laser speckle image; processing the initial laser speckle image to identify one or more portions of the initial laser speckle image representing the region of interest; and extracting the identified one or more portions of the initial laser speckle image to define the laser speckle image.

[0027] In some examples, the step of processing the initial laser speckle image includes: identifying any portion of the initial laser speckle image representing a heterogeneous region of the object as a heterogeneous portion, the heterogeneous region including one or more capillaries and one or more of arteries and veins; and excluding the identified heterogeneous portion of the initial laser speckle image from one or more portions of the identified initial laser speckle image representing the region of interest.

[0028] If included in the laser speckle image, heterogeneous regions may lead to misinterpretation, since spectral analysis of small patches always assumes that the observed pattern is a periodic continuation. Therefore, to improve accuracy, severe heterogeneity needs to be detected and rejected or masked.

[0029] If the laser speckle image includes spatial heterogeneity (e.g., the imaged area contains large vessels surrounded by tissue), the spatial frequency transform will assume that signal jumps due to the vessels will reappear regularly outside the field of view and can include frequencies in the spectrum that are more related to the location of the vessels in the current image rather than the speckle pattern.

[0030] In some examples, processing the initial laser speckle image includes identifying one or more portions of the initial laser speckle image containing one or more desired text features of the initial laser speckle image as the portion of the initial laser speckle image representing the region of interest.

[0031] The one or more laser speckle images may include multiple laser speckle images with different exposure times. Images recorded with different exposure times will contain blood flow velocity information with different dynamic ranges. This improves the robustness and accuracy of the blood flow velocity index generated using multiple laser speckle images. In particular, using different exposure times, the decorrelation time of the backscattered electromagnetic waves can be determined or otherwise considered.

[0032] A computer program product comprising code which, when executed by a processor circuit, causes the processor circuit to perform the steps of any method described herein is also proposed. A non-transitory computer readable medium or data carrier comprising or carrying the computer program product is also proposed.

[0033] A processing system for processing one or more laser speckle images is also proposed, each laser speckle image being an image containing a speckle pattern produced by reflections of light illuminating a region of interest of an object.

[0034] The processing system is configured to: process the one or more laser speckle images using a spatial frequency transform to generate spatial frequency data; and define a plurality of blood flow velocity indicators using the spatial frequency data, each blood flow velocity indicator indicating or representing an amount of blood moving at a corresponding different blood flow velocity within the region of interest.

[0035] Those skilled in the art will be able to readily adapt the processing system to perform any of the methods described herein.

[0036] A laser speckle imaging system is also proposed, comprising: the processing system; and a laser speckle device configured to capture the one or more laser speckle images.

[0037] Any advantages of the computer-implemented method according to the present invention similarly and analogously apply to the system and computer program disclosed herein.

[0038] It will be appreciated by those skilled in the art that two or more of the above-mentioned options, embodiments, and / or aspects of the invention may be combined in any way deemed useful. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] For a better understanding of the invention, and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0040] Figure 1 A laser speckle imaging system including a processing system according to an embodiment is shown;

[0041] Figure 2 A method for processing one or more laser speckle images is shown;

[0042] Figure 3 A method of generating processing data using one or more laser speckle images is shown;

[0043] Figure 4 shows the effect of increasing the flow rate on the spatial power spectrum of one or more laser speckle images;

[0044] Figure 5 A method of generating output data using one or more laser speckle images and a time-varying signal is shown;

[0045] Figure 6 A method for obtaining and processing one or more laser speckle images is shown;

[0046] Figure 7 A method for obtaining one or more laser speckle images is shown; and

[0047] Figure 8 A processing system according to an embodiment is shown. DETAILED DESCRIPTION

[0048] The present invention will be described with reference to the accompanying drawings.

[0049] It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the apparatus, system, and method, are 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, system, and method of the invention will be better understood through the following description, claims, and 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 indicate the same or similar components.

[0050] The present invention provides a mechanism for processing one or more laser speckle images. A spatial frequency transform is used to transform the one or more laser speckle images into spatial frequency data. The spatial frequency data is used to define a plurality of blood flow velocity indicators, each of which indicates the amount of blood moving at a specific speed within an illuminated region of interest.

[0051] In the context of this disclosure, "object" may alternatively be labeled "current object" or "studied object". This is in contrast to "historical object". When used alone, the term "frequency" generally refers to time-derived frequency, rather than "spatial frequency".

[0052] Embodiments may be used in a variety of clinical scenarios where information about blood flow velocity in a subject's region of interest would be useful. Example scenarios include diagnosis, evaluation, and treatment (including initial and follow-up appointments) of microvascular dysfunction and / or PAD.

[0053] Another use case scenario for the proposed methods is their use during perfusion challenges such as hyperemia experiments triggered by occlusion or thermal stimulation. Here, changes in blood velocity metrics (e.g., spectral distribution of speckle function) serve as indicators of changes in blood velocity distribution. The latter probes the ability of the vasculature to respond and adapt to external stimuli and changing demands.

[0054] Figure 1 A laser speckle imaging system 100 according to an embodiment is conceptually shown. The laser speckle imaging system 100 includes a laser speckle device 110 and a processing system 150, which is itself an embodiment of the present invention.

[0055] The laser speckle device 110 is configured to generate one or more laser speckle images 115 of an object 140, for example, multiple laser speckle images 115 of the object 140. The laser speckle image 115 is an image containing a speckle pattern generated by reflection of light illuminating a region of interest of the object. Methods for generating laser speckle images of an object using a laser speckle device are well-established in the art.

[0056] As an example, the laser speckle device 110 can be configured to irradiate coherent light (e.g., red light or near infrared light) onto tissue and observe the backscattered speckle pattern (as a laser speckle image 115). This can be achieved using a light emitting system and a camera system (or other optical sensor) or an attachable electrode. The choice of a camera system or an attachable electrode may depend on the clinical scenario.

[0057] For example, the laser speckle device 110 may include and use an optical sensor or camera system (due to its low cost implementation) during diagnosis or follow-up of a PAD patient. Such an optical sensor or camera system may include a CCD camera and / or a CMOS sensor. As another example, a contact-based system (e.g., with attachable electrodes) may be used in an intraoperative setting to confirm treatment efficacy.

[0058] Each laser speckle image will be associated with a specific exposure time T. The exposure time T indicates the length of time over which the laser speckle image is captured, ie the length of time over which reflected light is collected, as is well known in the field of image acquisition.

[0059] The processing system 150 is configured to process the laser speckle image(s) 115 to generate output data.

[0060] In particular, the processing system 150 is configured to process the laser speckle image(s) using a spatial frequency transform to generate spatial frequency data. The processing system then uses the spatial frequency data to define a plurality of blood flow velocity indicators.

[0061] Multiple blood velocity indices extracted from the spatial frequency data will carry information about the blood velocity distribution within the imaging region. This provides clinicians with useful information to improve their understanding of regions and subjects of interest.

[0062] The information carried by the multiple blood flow velocity indicators can be considered to be in a coded format, e.g., not providing direct information about the velocity distribution or a 1:1 correspondence with the velocity distribution. Nevertheless, the multiple blood flow velocity indicators will still change in response to changes in the velocity distribution, thereby allowing inferences to be made from the generated data and / or the changes identified.

[0063] The processing system 150 may be configured to perform one or more pre-processing steps on the laser speckle image(s) prior to using the spatial frequency transform. Similarly, the processing system 150 may be configured to perform one or more post-processing steps on the plurality of blood flow velocity indicators, for example, mapping or converting the blood flow velocity indicators into a target quantity or prediction (e.g., actual velocity distribution, perfusion state, or outcome measure).

[0064] More detailed examples and explanations of possible pre-processing and post-processing steps or processes are provided later in this disclosure.

[0065] In some examples, the processing system 150 may be configured to control the operation of the laser speckle device, for example, to control how and / or when the laser speckle device captures the laser speckle image. For example, the processing system 150 may be able to control one or more capture properties of the laser speckle device, such as exposure time, time to capture the laser speckle image, illumination intensity, illumination position, etc.

[0066] More detailed examples and explanations of such procedures are provided later in this disclosure.

[0067] The processing system 150 generates output data, which may include a plurality of blood flow velocity indicators themselves and / or processed data. Processed data is data generated by performing one or more post-processing steps on (at least) a blood flow velocity indicator.

[0068] The laser speckle imaging system 100 may include a user interface 160, such as a display or screen. The processing system 150 may be configured to control the user interface 160 to provide a visual representation of the output data (e.g., multiple blood flow velocity indicators and / or processed data). Methods for controlling the user interface 160 to provide a visual representation of the data are mature in the art and are not explained for the sake of brevity.

[0069] Figure 2 2 is a flow chart showing a computer-implemented method 200 according to an embodiment. The computer-implemented method may be performed by Figure 1 The processing system 150 is shown executed or implemented.

[0070] The computer-implemented method includes a process 210 of processing one or more laser speckle images using a spatial frequency transform to generate spatial frequency data. The process 210 here performs the step of converting the one or more laser speckle images into a spatial power spectrum. The spatial power spectrum identifies the power of the spatial frequency in (one or more) laser speckle images for each of a plurality of spatial frequencies.

[0071] Spatial frequency transforms are well known in the art. The spatial power spectrum produced by a spatial frequency transform effectively identifies or defines how the pixel values ​​of an image vary in space. In practice, a spatial frequency transform is a multi-dimensional Fourier transform that decomposes the variation of pixel value or intensity over distance (e.g., in contrast to a conventional frequency transform that decomposes the variation of pixel values ​​over time).

[0072] One example of a space-frequency transform is a Fast Fourier Transform (FFT) or a Short-Time Fast Fourier Transform (STFFT).However, other suitable space-frequency transforms may be used, for example any suitable Fourier-based transform.

[0073] In some methods, performing a spatial frequency transform includes performing a multidimensional spatial frequency transform. This produces a multidimensional spatial power spectrum. Then, process 210 can convert the multidimensional spatial spectrum into a 1D (one-dimensional) spatial power spectrum by treating the data along each of the multiple angles within the multidimensional spatial spectrum as an initial 1D spatial spectrum. It can be assumed that the measured spectral features should not be (significantly) different in the orientation of the multidimensional spatial power spectrum. The initial 1D spatial spectra can be combined (e.g., by summing or averaging) to produce a combined 1D spatial spectrum, which serves as the output of the spatial frequency transform, that is, as the spatial power spectrum.

[0074] The process for generating a 1D spatial power spectrum from a multi-dimensional image may be labeled a multi-dimensional to single-dimensional spatial frequency transform.Other processes for generating a 1D spatial power spectrum from a multi-dimensional image will be apparent to those skilled in the art.

[0075] One technique for generating a spatial power spectrum is to simply perform a spatial frequency transform on each laser speckle image to generate a corresponding spatial power spectrum.

[0076] If there are multiple laser speckle images (thus generating multiple spatial power spectra), the generated spatial power spectra can be combined (e.g., summed or averaged) to generate a spatial power spectrum to be output by process 210. This method is particularly advantageous if the laser speckle images are obtained at the same point in the cardiac cycle (and / or other rhythm cycles that affect the imaging region, such as the respiratory cycle). Alternatively, if there are multiple laser speckle images, the generated spatial power spectra can be cascaded together. This method is particularly advantageous if the laser speckle images are obtained at different points in the cardiac cycle or at different exposure times. The cascaded information still provides useful data for the clinician or further processing system to evaluate the blood flow velocity distribution within the imaging region.

[0077] A more advanced technique for generating a spatial power spectrum is to perform frequency binning on the initial spatial power spectrum (or spectra) to generate a spatial power spectrum to be output by process 210. This technique allows for a higher signal-to-noise ratio and also compensates for different sampling resolutions in the spatial frequency domain by normalizing it to a fixed grid.

[0078] In this technique, each laser speckle image may be converted into an initial spatial power spectrum in step 211. This may include, for example, performing a multi-dimensional to single-dimensional spatial frequency transform. The initial spatial power spectrum identifies the power of a spatial frequency in the (corresponding) laser speckle image for each of a plurality of spatial frequencies.

[0079] Then, in step 212, a plurality of different spatial frequency ranges are defined.

[0080] In step 213, for each spatial frequency range, the powers of the spatial frequencies in the (one or more) initial power spectra that fall within that spatial frequency range are combined (e.g., summed or averaged) to produce the powers of the spatial power spectrum to be output by process 210. Thus, there are as many powers or values ​​in the spatial power spectrum (output by process 210) as there are spatial frequency ranges.

[0081] If there are multiple laser speckle images, and therefore multiple initial spatial power spectra, then there are various mechanisms for performing step 213 .

[0082] In one approach, step 213 includes combining (e.g., summing or averaging) all initial spatial power spectra to produce a combined initial spatial power spectrum. Then, for each spatial frequency range, step 213 generates the power of the spatial power spectrum by combining (e.g., summing or averaging) the powers of all spatial frequencies in the combined initial spatial power spectrum that fall within the spatial frequency range. This generates a spatial power spectrum.

[0083] In another method, step 213 includes, for each spatial frequency range, generating a group of intermediate powers. Each intermediate power in the same group is a combination (e.g., sum or mean) of the powers of all spatial frequencies that fall within the corresponding spatial frequency range in different initial spatial power spectra. Therefore, the number of intermediate powers in each group is the same as the number of initial spatial power spectra. Each group of intermediate powers is then processed to generate a single power of a spatial power spectrum for each group. In particular, the intermediate powers in the same group are combined (e.g., summed or averaged) to generate a single power of the group.

[0084] Each spatial frequency range may partially overlap with one or more other spatial frequency ranges. This provides a windowing effect to improve the signal-to-noise ratio.

[0085] Alternatively, if there are multiple laser speckle images, the iteration of step 213 may be performed on each initial spatial power spectrum, that is, frequency merging may be performed on each initial spatial power spectrum to generate multiple intermediate spatial power spectra. The intermediate spatial power spectra may then be concatenated together to generate a spatial power spectrum.

[0086] It is not necessary that process 210 or step 213 output only a single spatial power spectrum. Instead, multiple spatial power spectra may be output (eg, one for each laser speckle image).

[0087] The method 200 also includes a step 220 of defining a plurality of blood flow velocity indicators using the spatial frequency data, each blood flow velocity indicator indicating or representing the amount of blood moving at a corresponding different blood flow velocity within the region of interest. Preferably, step 220 includes defining no less than 5 blood flow velocity indicators, for example no less than 10 blood flow velocity indicators.

[0088] The relationship between each blood flow velocity index and the blood volume may not be a direct correlation. Instead, there may be a nonlinear relationship between the blood flow velocity index and the amount of blood at a specific blood flow velocity.

[0089] In a simple example, each power in the spatial power spectrum can serve as a corresponding blood flow velocity indicator. This provides a numerical representation of the blood flow velocity. In this way, the spatial power spectrum itself includes or defines the blood flow velocity indicator generated in step 220.

[0090] In another example, each power in the spatial power spectrum can be classified or categorized to generate a classified blood flow velocity indicator. For example, each power can be compared to one or more predetermined non-overlapping ranges. Each range can define a different category or classification (e.g., "low", "medium" or "high"). A classified blood flow velocity indicator can be generated by identifying which predetermined range the power falls into and generating a blood flow velocity indicator in response to that value.

[0091] In a more advanced example, step 210 includes processing a plurality of laser speckle images to generate a spatial power spectrum for each laser speckle image. As previously described, this may include using a reference Figure 2 The described method performs frequency binning (which produces a single value for the spatial frequency range within the spatial power spectrum).

[0092] It should be understood that each laser speckle image is captured at a different point in time or over a different period of time. This produces a series or a series of spatial power spectra showing how the spatial power spectrum changes over time. Therefore, a time-varying signal can be generated at each of a plurality of spatial frequencies or spatial frequency ranges. Each time-varying signal represents the change in amplitude over time at a specific spatial frequency or spatial frequency range. Therefore, each time-varying signal is actually a spatial frequency filtered version of a plurality of spatial power spectra. Each time-varying signal can then undergo a Fourier-based transform to produce a frequency power spectrum for each of a plurality of spatial frequencies or spatial frequency ranges. The amplitude or power at different frequencies of each frequency power spectrum can form or provide a corresponding blood flow velocity indicator.

[0093] In this way, multiple groups of blood flow velocity indicators can be generated. Each group of blood flow velocity indicators contains multiple blood flow velocity indicators, each blood flow velocity indicator indicates the amplitude of a different frequency or frequency range of the same spatial frequency or spatial frequency range (associated with the group). Different groups of blood flow velocity indicators are associated with different spatial frequencies or spatial frequency ranges.

[0094] This is conceptually illustrated by Table 1, which shows how different frequency components at different spatial frequencies can be generated. As previously described, the rows of Table 1 represent different groups of blood flow velocity indicators. The columns of Table 1 represent blood flow velocity indicators generated from the corresponding frequency power spectra.

[0095]

[0096] Table 1

[0097] Therefore, if step 210 includes processing a plurality of laser speckle images to generate a spatial power spectrum for each of the laser speckle images, step 220 may include the following steps: processing the spatial power spectrum to define a time-varying value for each of a plurality of spatial frequencies or spatial frequency ranges; performing a Fourier-based transform on each of the time-varying values ​​to generate a plurality of frequency power spectra; and for each frequency power spectrum, defining the amplitude of each of the plurality of frequencies or frequency ranges of the frequency power spectrum as a blood flow velocity indicator.

[0098] Examples of Fourier based transforms are well known in the art and include the Fast Fourier Transform or the Short-time Fast Fourier Transform.

[0099] The advanced approach incorporates time-varying data into each metric produced. Therefore, each metric can further carry important and clinically interpretable information about blood flow velocity in the imaged region. This provides valuable data for increasing contextual understanding of the blood flow velocity metric, for example, for further processing by an analysis system or for presentation for individual understanding.

[0100] Although the spatial power spectrum is derived from laser speckle images captured over a period of time (i.e., a non-instantaneous time step), the time step of the relevant temporal information contained in a series / sequence of spatial power spectra is much larger than the capture time (e.g., on the order of 100x or more).

[0101] The method 200 may include a step 230 of displaying the blood flow velocity indicator defined in step 220, for example on a user interface. Methods for controlling a user interface to display data, such as a blood flow velocity indicator, are well known in the art.

[0102] Step 230 may or may, for example, include displaying (e.g., on a user interface) a spatial power spectrum containing a blood flow velocity indicator. The method recognizes that the spatial power spectrum can be visualized as a bar graph or line graph. This allows comparison of blood flow velocity distributions between different perfusion states (e.g., within the same patient) or between patients. If this step is performed, this effectively provides a visual representation of the blood flow velocity indicator.

[0103] It will be appreciated that, at least to ensure consistent interpretability of information generated using spatial frequency transforms, the resolution of the laser speckle image in terms of surface distance (e.g., measured in mm or m) should be known. This may be particularly relevant to camera-based systems in which the optical sensor may be located at varying distances from the surface.

[0104] The pixel spacing (eg in mm) is preferably used to define the sampling frequency of the spatial frequency transform.

[0105] The surface range resolution of the laser speckle image may be estimated using various techniques, including triangulation or identifying reference markers in the image. Alternatively, the surface range resolution may be predefined or set, for example by ensuring that the laser speckle image is captured from the same distance from the / each object, for example using a spacer. This method provides a way to calibrate the frequency content across measurements.

[0106] In order to improve the accuracy of the blood flow velocity information, the one or more laser speckle images preferably include a plurality of laser speckle images with different exposure times. Images recorded with different exposure times will contain blood flow velocity information with different dynamic ranges.

[0107] In general, the power of the spatial power spectrum generated by process 210 thereby produces a multivariate marker for blood velocity distribution and perfusion of the tissue. In particular, the power (serving as a blood velocity indicator) effectively produces a set of ordered amplitudes, the order being defined by the spatial frequency associated with each power.

[0108] As previously described, one or more pre-processing steps and / or one or more post-processing steps may be performed. In particular, to improve interpretability, the blood flow velocity indicator (eg, spatial power spectrum) may be further processed to produce other measurement results or outputs.

[0109] Figure 3 Some example post-processing steps are shown that can be used to process blood flow velocity indicators to produce different types of processed data. Figure 3 A method 300 is shown that includes various methods for converting a blood flow velocity indicator into a variable of some target quantity.

[0110] Two methods are shown, one represented by step 310 and the other represented by step 320. One or both of these steps may be included in method 300.

[0111] One method is illustrated by step 310, which includes processing a spatial power spectrum to determine one or more characteristics of the spatial power spectrum. The spatial power spectrum includes a blood flow velocity indicator, for example, each value of the spatial power spectrum is a blood flow velocity indicator.

[0112] One suitable example of a characteristic of a spatial power spectrum includes a cutoff spatial frequency or an angular spatial frequency of the spatial power spectrum. Another suitable characteristic is the gradient or slope of the decaying attenuation (decay) from the cutoff spatial frequency to higher frequencies. The cutoff spatial frequency and the gradient / slope of the decay in the spectrum are indicators of the amount and velocity distribution of scatterers in the imaging region.

[0113] Figure 4 The relationship between the velocity of the scatterer and the spatial power spectrum is intuitively shown.

[0114] Figure 4 Two waveforms are shown, a first waveform 410 providing a baseline condition and a second waveform 420 providing a condition where the flow of scatterers (e.g., red blood cells) in the illuminated region of interest is increased compared to the first waveform. Each waveform depicts amplitude or power A (y-axis) versus spatial frequency f (x-axis).

[0115] Thus, each waveform provides a visual representation of spatial frequency data or a spatial power spectrum.

[0116] When illuminating a region of interest with light, each illuminating photon has a certain probability of striking one or more scatterers moving at one or more different speeds in the illuminated region of interest. This probability, and therefore the number of photons that will encounter only slower scatterers and / or the number that will encounter fast scatterers, will depend on the available velocity profile. The velocity profile will therefore define the blurring of the laser speckle image (i.e., the attenuation of higher spatial frequency components) and the frequency dependence of the attenuation. This can be observed and quantified using a cutoff spatial frequency or angular spatial frequency and the degree of attenuation.

[0117] In particular, as the number of faster scatterers increases, the probability of a photon hitting a fast scatterer also increases. Therefore, the probability of smaller spatial frequencies being attenuated increases. This effectively shifts the cutoff spatial frequency of the spatial power spectrum to a lower spatial frequency. As the flow of scatterers (e.g., red blood cells) in the illumination area increases, the first cutoff spatial frequency f of the first waveform 410 increases. c1 How much is greater than the second cutoff spatial frequency f of the second waveform 420? c2 , which illustrates this effect.

[0118] However, while this shift in the cutoff spatial frequency to lower spatial frequencies is caused by the addition of some fast scatterers, it could also be caused by an increase in the number of medium-speed scatterers. The difference between these scenarios can be unraveled by looking at the slope of the decay, which, for example, can be different in different scenarios.

[0119] The faster the scatterers, the faster the high spatial frequencies will decorrelate. This means that as the ratio of fast to slow scatterers increases, the slope (e.g., of the waveform) will be steeper because the attenuation of higher powers increases with the increase in faster scatterers. The relationship between this ratio and the steepness or gradient of the waveform can be a non-linear relationship.

[0120] therefore, Figure 4 An advantage of determining the characteristics of the spatial power spectrum is demonstrated, as well as the advantage of the spatial power spectrum itself (e.g., for personal viewing for evaluation).

[0121] Back to Figure 3 The method 300 may further include a step 315 of displaying the determined characteristics of the spatial power spectrum on a user interface. Methods for controlling a user interface to display data are well known in the art.

[0122] Step 320 illustrates another method for processing the blood flow velocity indicator.

[0123] Step 320 includes processing the one or more blood flow velocity indicators using a computer implemented function or mapping function to determine values ​​of one or more parameters of the subject.

[0124] Example parameters that may be derived, predicted or determined using a computer-implemented method include a blood flow velocity distribution or a distribution parameter thereof. Thus, the computer-implemented function may be a function that maps a value of a blood flow velocity indicator into a blood flow velocity distribution.

[0125] Other example parameters that can be derived, predicted or determined using computer-implemented methods include one or more clinical parameters, such as pathological prognosis (eg, predicted chance of reduced ischemia or wound healing). The pathological prognosis can be the prognosis of a predetermined or known pathology of a subject.

[0126] For example, the computer-implemented function may be a machine learning algorithm. A first example of a machine learning algorithm is configured to process a blood flow velocity indicator to generate or determine a blood flow velocity distribution or a distribution parameter thereof. A second example of a machine learning algorithm is configured to process a blood flow velocity indicator to generate or determine one or more values ​​of one or more clinical parameters (e.g., pathological prognosis).

[0127] A machine learning algorithm is any self-training algorithm that processes input data to produce or predict output data. For a first example of a machine learning algorithm, the input data includes a blood flow velocity indicator, and the output data includes a blood flow velocity distribution and / or its distribution parameters. For a second example of a machine learning algorithm, the output data instead includes one or more values ​​of one or more clinical parameters.

[0128] Suitable machine learning algorithms for the present invention will be apparent to those skilled in the art. Examples of suitable machine learning algorithms include decision tree algorithms and artificial neural networks. Other machine learning algorithms such as logistic regression, support vector machines, or naive Bayes models are suitable alternatives. The precise decision about which machine learning algorithm to choose may depend, at least in part, on the type of data to be processed or generated (e.g., whether the output data is a continuous variable or a categorical variable). The appropriate selection will be apparent to those skilled in the art of machine learning algorithms.

[0129] The structure of artificial neural networks (or neural networks for short) is inspired by the human brain. Neural networks consist of layers, each layer includes multiple neurons. Each neuron includes a mathematical operation. In particular, each neuron can include a different weighted combination of a single type of transformation (e.g., the same type of transformation, sigmoid, etc., but with different weights). During the processing of input data, the mathematical operation of each neuron is performed on the input data to produce a numerical output, and the output of each layer in the neural network is sequentially fed to the next layer. The last layer provides the output.

[0130] Methods for training machine learning algorithms are well known. Typically, such methods include obtaining a training data set, including training input data entries and corresponding training output data entries. An initialized machine learning algorithm is applied to each input data entry to generate a predicted output data entry. The error between the predicted output data entry and the corresponding training output data entry is used to modify the machine learning algorithm. The process can be repeated until the error converges and the predicted output data entry is sufficiently similar to the training output data entry (e.g., ±1%). This is generally referred to as a supervised learning technique.

[0131] For example, in the case where the machine learning algorithm is formed by a neural network, the mathematical operation (weighting) of each neuron can be modified until the error converges. Known methods for modifying neural networks include gradient descent, back propagation algorithms, etc.

[0132] To generate a first example of a machine learning algorithm, the training input data entries of the training data set correspond to an example set of blood flow velocity indicators, and the training output data entries of the training data set correspond to a blood flow velocity distribution or a distribution parameter thereof.

[0133] One method for generating such a training data set is to use Monte Carlo simulation. In particular, different simulated laser speckle images can be generated (using Monte Carlo techniques) using a tissue model and a reference true value blood flow velocity distribution and / or its distribution parameters. The simulated laser speckle images can then be processed (using the above techniques) to generate a set of blood flow velocity indicators for each simulated laser speckle image. Each set of blood flow velocity indicators serves as a training input data entry. Each reference true value blood flow velocity distribution and / or its distribution parameters used to simulate the simulated laser speckle image serves as a training output data entry.

[0134] The technique provides a simple and easily implemented end-to-end learnable model. In particular, despite the lack of an analytical model for the underlying functional relationship, the method can be used to approximate the results of such an unknown functional relationship.

[0135] To generate the second example of the machine learning algorithm, the training data set is a first training data set including a plurality of sets of blood flow velocity indicators as training input data items, and corresponding values ​​of one or more clinical parameters as training output data items (for patients for whom the blood flow velocity indicators for the corresponding training input data items were generated). The values ​​of the one or more clinical parameters may be generated by a skilled and / or experienced clinician and / or based on historical medical data.

[0136] In some examples, the second example of the machine learning algorithm includes using two machine learning algorithm processes. Each machine learning algorithm process is another example of the machine learning algorithm. The first machine learning algorithm process can be the same as the first example of the machine learning algorithm previously described and can be trained as described above. The second machine learning algorithm process can use another machine learning algorithm to process the output of the first example of the machine learning algorithm to predict the value of one or more clinical parameters.

[0137] In this method, the training data set may again include multiple sets of blood flow velocity indicators as training input data items, and corresponding values ​​of one or more clinical parameters as training output data items. For this scenario, the second example of training the machine learning algorithm may include modifying only the second machine learning algorithm process without modifying the already trained first machine learning algorithm process. This increases the difficulty of training the second machine learning algorithm process.

[0138] The method 300 may further comprise a step 325 of displaying the determined value(s) of the parameter(s) on a user interface. Methods for controlling a user interface to display data are well known in the art.

[0139] It will be appreciated that step 320 may be repeated for different parameters. Thus, a plurality of machine learning algorithms may process (at least) the blood flow velocity information to produce value(s) of the parameter(s) for the subject under study.

[0140] The above-described method generates processed data (eg, blood flow velocity distribution information and / or value(s) of one or more clinical parameters) using only blood flow velocity indicators derived from laser speckle images.

[0141] However, further information may be used to generate processed data.

[0142] A suitable example of such further information are subject properties, such as properties of the subject, such as age, comorbidities, subcutaneous fat content, amount of chromophores and / or different macromorphological properties, such as the presence and / or absence of scar / wound tissue.

[0143] It will be appreciated that the scattering properties of the illuminated region of interest of the object will affect or bias the baseline appearance of the projected speckle pattern (e.g., regardless of the actual blood flow velocity). The scattering properties depend on or are dependent on properties of the object, such as those previously described. Thus, the blood flow velocity indicator produced by the proposed method is related to the object characteristics. Using such object characteristics in determining one or more parameters of the object allows these parameters to be taken into account, for example, as control variables.

[0144] Therefore, a set of one or more subject characteristics may be further used to determine the value(s) of one or more parameters (e.g., blood flow velocity information and / or clinical parameters) of the subject. This provides a more accurate and reliable method for determining such value(s).

[0145] In a simple example, the set of one or more object characteristics may be included in the input data provided to a computer-implemented method for generating one or more parameters of an object. For example, a machine learning method for determining the value(s) of one or more parameters may be trained to receive as input a blood flow velocity parameter and the set of one or more object characteristics. In this case, each training input data entry of a training data set used to train the machine learning algorithm may also include a set of one or more object characteristics of the object associated with the training input data entry.

[0146] In a more complex example, a different machine learning method can be trained for each of a plurality of historical object clusters, e.g., each machine learning algorithm can be trained using data derived from a corresponding single historical object cluster. Each cluster of historical objects can share similar values ​​of one or more object properties. One or more object properties of the current object can be processed to identify the most similar historical object cluster. The machine learning algorithm trained for / using the historical object cluster can then be used to process (at least) the blood flow velocity indicator for the current object to produce (one or more) values ​​of one or more parameters of the object.

[0147] Another suitable example of further information is information derived from a time-varying signal representing temporal variations of blood flow in the region of interest. A suitable example of a time-varying signal is a photoplethysmographic signal or photoplethysmogram, e.g. obtainable from a PPG sensor such as a finger-worn pulse oximeter. Another example of a time-varying signal is a speckle plethysmographic signal.

[0148] While spatial frequency data can be used to assess mean velocity distribution, time-varying signals can capture pulsatile aspects along the temporal dimension. This provides additional clinically useful information for assessing and determining output data.

[0149] Figure 5 A method 500 is shown according to an embodiment that utilizes a time-varying signal to generate output data.

[0150] Method 500 includes executing the previously described Figure 2 Method 200 is described.

[0151] The method 500 further comprises a step of obtaining a time-varying signal 510. The time-varying signal represents the temporal variation of blood flow in the region of interest. Suitable examples are known to those skilled in the art, some of which have been described above.

[0152] The method 500 further comprises a step 520 of transforming the time varying signal into the frequency domain. Step 520 may be performed using any established transform technique, for example any suitable Fourier based transform, such as a Fast Fourier Transform or a Short-time Fast Fourier Transform.

[0153] The method 500 also includes a step 530 of defining a set of frequency amplitudes. The set of frequency amplitudes contains amplitudes of different frequency components or frequency ranges of the time-varying signal. In one example, the set of frequency amplitudes includes a value / amplitude for each of a plurality of predetermined frequencies. In another example, for each of a plurality of predetermined frequency ranges, the set of frequency amplitudes includes a sum or average (i.e., combined) value / amplitude of the amplitudes within the predetermined frequency range.

[0154] The method 500 also includes a step 540 of determining one or more parameters of the illuminated region of interest using the blood flow velocity index and the defined set of frequency amplitudes.

[0155] The one or more parameters may include, for example, blood flow velocity distribution information and / or clinical parameters (eg, prognosis, etc.).

[0156] Methods for processing data to generate such one or more parameters have been previously described in the context of processing blood flow velocity indicators alone, for example using machine learning algorithms. These methods can be suitably adapted to process additional information as input, for example, a defined set of frequency amplitudes as well as blood flow velocity indicators.

[0157] In this case, each training input data entry of the training data set used to train the machine learning algorithm may also include a set of frequency amplitudes derived from the time-varying signal.

[0158] Preferably, the time-varying signal obtained in step 510 is captured within at least an exposure time for generating the laser speckle image.

[0159] In order to improve the signal-to-noise ratio of the spatial power spectrum and / or blood flow velocity information, and the accuracy of any output data generated using such information, the content of the laser speckle image(s) may be controlled or defined.

[0160] Consider two general approaches for defining the content of the laser speckle image(s).Of course, depending on the embodiment, both, one, or neither of these approaches may be performed.

[0161] The first method processes, masks or filters the obtained laser speckle image(s) to produce one or more processed laser speckle images for generating spatial frequency data. Thus, in the first method, content is extracted from the obtained laser speckle image(s). In other words, one or more pre-processing steps are applied to generate laser speckle image(s) processed using spatial frequency transform.

[0162] The second method controls how and / or when to capture (one or more) laser speckle images. Therefore, in the second method, the operation of the laser speckle device is controlled. The second method can, for example, control the exposure time of (one or more) laser speckle images and / or the time to capture (one or more) laser speckle images.

[0163] Figure 6 A method 600 is shown for performing an approach according to a first method for defining the content of laser speckle image(s).

[0164] The method 600 includes a step 610 of obtaining an initial laser speckle image. The initial laser speckle image contains a speckle pattern generated by reflection of light illuminating a region of interest of an object. The initial laser speckle image contains the laser speckle image and has a higher resolution than the laser speckle image.

[0165] The method 600 also includes a step 620 of processing the initial laser speckle image to identify one or more portions of the initial laser speckle image that represent a region of interest.

[0166] For example, step 620 may be performed by identifying any portion of a heterogeneous region representing an object in the initial laser speckle image as a heterogeneous portion, the heterogeneous region including one or more capillaries and one or more of an artery and a vein; and excluding the identified heterogeneous portion of the initial laser speckle image from one or more portions representing a region of interest in the identified initial laser speckle image.

[0167] For example, detection of heterogeneous regions can be performed by analyzing texture features at different locations in the initial speckle image. Clustering or segmentation into subsets of homogeneous regions can then allow sub-region selection, for example, discarding or excluding regions with potentially large veins in favor of sub-regions dominated by skin tissue.

[0168] The method 600 also includes a step 630 of extracting one or more portions of the identified initial laser speckle image to define a laser speckle image.

[0169] Method 600 provides a mechanism to manually or automatically mask regions of interest.

[0170] More specifically, an initial laser speckle image (e.g., recorded by an optical sensor) will cover a specific area. Within this area, a region of interest can be selected. This selection can be performed manually (e.g., using spline interpolation control points) or (preferably) automatically.

[0171] In case of heterogeneous surfaces such as large blood vessels crossing the observation area, the mask can have functional relevance. This can lead to misinterpretations or inaccuracies, since spectral analysis of small patches always assumes a periodic continuation of the observed pattern. Therefore, it would be advantageous to detect and exclude / mask severe heterogeneities from the laser speckle image(s).

[0172] Figure 7 A method 700 for performing an approach according to the second method for defining the content of (one or more) laser speckle images is shown. In particular, Figure 7 Techniques for controlling the time at which laser speckle image(s) are captured and / or the capture time of laser speckle image(s) are shown.

[0173] The method 700 comprises a step 710 of obtaining a time-varying signal. The time-varying signal represents the temporal variation of blood flow in the region of interest. Suitable examples are known to those skilled in the art, some of which have been described above.

[0174] Step 710 is similar to previously described step 510. Step 710 may be performed by obtaining a time-varying signal from a monitoring device such as a PPG sensor (eg, a finger-worn pulse oximeter).

[0175] The method 700 also includes a step 720 of identifying a time window for capturing (one or more) laser speckle images. In particular, the time-varying signal can be processed to identify the pulsatility (exponential) or periodicity of the time-varying signal. The periodicity of the time-varying signal can be used to set the time window for capturing the laser speckle image.

[0176] For example, the identified time window may be a time window corresponding to the periodicity of the time-varying signal (eg, periodicity due to motion or (one or more) heartbeats), eg, a time window whose length is proportional to the period of the time-varying signal.

[0177] As another example, the identified time window may be a time window that includes a known or predetermined number of cycles or a predetermined proportion of cycles of the time-varying signal.

[0178] As yet another example, the identified time window may be a time window that includes the same number of time periods (time period ratio) as a previous time window during which a previous or earlier laser speckle image was captured. This ensures consistency in measuring or obtaining (one or more) laser speckle images.

[0179] The method further comprises a step 730 of obtaining or capturing one or more laser speckle images in the identified time window. Step 730 may be performed by appropriately controlling the laser speckle device.

[0180] The time-varying signal obtained in step 710 may be further used to determine one or more parameters, for example, as previously described with reference to Figure 5 described.

[0181] The method 700 may include processing the time-varying signal to determine the capture time of the laser speckle image 740. In particular, the time-varying signal may be processed using a frequency transform process (e.g., a Fourier-based transform) to identify one or more frequency components (e.g., one or more main frequencies) of the time-varying signal.

[0182] Step 730 may be modified to include obtaining the laser speckle image(s) having an exposure time that matches one of the frequency components or the dominant frequencies (eg, the inverse of one of the frequency components or the dominant frequencies).

[0183] For example, this approach can ensure that multiple laser speckle images are captured at the same point during a periodic cycle of the object, such as a cardiac cycle.

[0184] In some examples, step 720 may be omitted. Step 730 may be modified accordingly.

[0185] The above-described means and methods generate output data (e.g., blood flow velocity indicators or data derived therefrom) from a single set of one or more laser speckle images. To provide improved clinical insight, two sets of output data may be generated, wherein a first set of output data is generated prior to clinical interrogation or stimulation, and a second set of output data is generated after clinical interrogation or stimulation. Example stimuli include thermal stimulation or occlusive stimulation.

[0186] In this way, any of the previously described methods can be repeated on either side of the clinical stimulation. The output data generated by the repeated methods is preferably provided or displayed on a user interface. This provides useful insights into the regulatory capacity of the tissue.

[0187] Figure 8 is a processing system according to an embodiment (eg, previously referred to as Figure 1 Schematic diagram of a processing system 150 described herein.

[0188] As shown, processing system 800 may include a data processor 820, a memory 864, and a communication module 868. These elements may communicate with each other, directly or indirectly, such as via one or more buses.

[0189] The data processor 820 may include a central data processor (CPU), a digital signal processor (DSP), an ASIC, a controller, an FPGA, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein. The data processor 820 may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in combination with a DSP core, or any other such configuration. In some embodiments, the processor is a distributed processing system, such as formed by a group of distributed processors.

[0190] The memory 864 may include a cache memory (e.g., a cache memory of the data processor 820), a random access memory (RAM), a magnetoresistive RAM (MRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a solid-state memory device, a hard drive, other forms of volatile and non-volatile memory, or a combination of different types of memory. In an embodiment, the memory 864 includes a non-transitory computer-readable medium. The non-transitory computer-readable medium can store instructions. For example, the memory 864 or the non-transitory computer-readable medium can have a program code recorded thereon, which includes instructions for causing the processing system 800 or one or more components of the processing system 800 (particularly the data processor 820) to perform the operations described herein. For example, the processing system 800 can perform the operations of the method 200.

[0191] Instructions 866 may also be referred to as code or program code. The terms "instructions" and "code" should be broadly interpreted to include any type of computer-readable statements. For example, the terms "instructions" and "code" may refer to one or more programs, routines, subroutines, functions, processes, etc. "Instructions" and "code" may include a single computer-readable statement or many computer-readable statements. The memory 864 having the code recorded thereon may be referred to as a computer program product.

[0192] Communications module 868 may include any electronic circuitry and / or logic circuitry to facilitate direct or indirect data communications between processing system 800 and any external device.

[0193] In this regard, the communication module 868 includes an input interface 810 and an output interface 830 for the processing system 800. The input interface 810 can be configured to receive input data from the data processor, such as one or more laser speckle images and an optional time-varying signal. The output interface 830 can be configured to provide output data from the data processor, such as a determined blood flow velocity index, a parameter generated by processing the blood flow velocity index, etc.

[0194] In some cases, the communications module 868 facilitates direct or indirect communications between various elements of the processing system 800 , such as using a bus or the like.

[0195] It should be understood that the disclosed methods are preferably computer-implemented methods. Therefore, the concept of a computer program is also proposed, which includes computer program code for implementing any described method when the program is run on a processing system such as a computer or a group of distributed processors.

[0196] Different parts, lines or blocks of the code of the computer program according to the embodiment can be executed by a processing system or a computer to perform any method described herein. In some alternative embodiments, the functions marked in the block diagram or flow chart may not appear in the order marked in the figure. For example, depending on the functions involved, the two consecutive frames shown can actually be executed substantially simultaneously, or these frames can sometimes be executed in reverse order.

[0197] The present disclosure proposes a computer program (product) comprising instructions, which, when executed by a computer or a processing system, cause the computer or the processing system to perform (the steps of) any method described herein. The computer program (product) may be stored on a non-transitory computer-readable medium.

[0198] Similarly, a computer-readable (storage) medium comprising instructions which, when executed by a computer or a processing system, cause the computer or processing system to perform (the steps of) any of the methods described herein is also proposed. A computer-readable data carrier stored on the previously described computer program (product) is also proposed. A data carrier signal carrying the previously described computer program (product) is also proposed.

[0199] By studying the drawings, the disclosure and the claims, those skilled in the art can understand and implement changes to the disclosed embodiments when practicing the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the attributive "one" or "an" does not exclude multiple. A single processor or other unit can implement the functions of several items recorded in the claims. The fact that certain measures are recorded in mutually different dependent claims does not indicate that the combination of these measures cannot be advantageous. The computer program can be stored / distributed on a suitable medium, such as an optical storage medium or solid-state medium provided together with or as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems. If the term "suitable for" is used in a claim or specification, it should be noted that the term "suitable for" is intended to be equivalent to the term "configured to". Any figure marks in the claims should not be interpreted as limiting the scope.

Claims

1. A computer-implemented method for processing one or more laser speckle images, each laser speckle image being an image containing a speckle pattern produced by reflection of light illuminating a region of interest of an object, the computer-implemented method comprising: processing the one or more laser speckle images using a spatial frequency transform to generate spatial frequency data; and A plurality of blood flow velocity indices are defined using the spatial frequency data, each blood flow velocity indice indicating or representing an amount of blood moving at a corresponding different blood flow velocity within the region of interest.

2. The computer-implemented method of claim 1 , wherein: The step of processing the one or more laser speckle images to produce spatial frequency data comprises converting the one or more laser speckle images into a spatial power spectrum, the spatial power spectrum identifying, for each of a plurality of spatial frequencies, a power of the spatial frequency in the one or more laser speckle images; and Each power in the spatial power spectrum serves as a corresponding blood flow velocity indicator.

3. The computer-implemented method of claim 2, wherein: The step of processing the laser speckle image to generate spatial frequency data comprises: converting each laser speckle image into an initial spatial power spectrum, the initial spatial power spectrum identifying, for each of a plurality of spatial frequencies, a power of the spatial frequency in the laser speckle image; defining a plurality of spatial frequency ranges; and The spatial power spectrum is generated by combining, for each of the plurality of spatial frequency ranges, powers of spatial frequencies in each initial spatial frequency spectrum that fall within the spatial frequency range to generate a power for the spatial power spectrum.

4. The computer-implemented method of any one of claims 1 to 3, further comprising processing one or more of the blood flow velocity indicators using a computer-implemented method to determine one or more values ​​of one or more characteristics of the object, wherein The one or more characteristics include at least one of: blood flow velocity distribution; clinical parameters such as pathological prognosis.

5. The computer-implemented method of claim 4, wherein: The one or more characteristics include at least one pathological prognosis.

6. A computer-implemented method according to claim 4 or 5, wherein: The computer implemented model is a machine learning model.

7. The computer-implemented method according to any one of claims 1 to 6, further comprising: obtaining a time-varying signal representing the time variation of blood flow in the region of interest; transforming the time-varying signal into the frequency domain; defining a set of frequency amplitudes, the set of frequency amplitudes comprising amplitudes of different frequency components or frequency ranges of the time-varying signal; and One or more parameters of the illuminated region of interest are determined using the blood flow velocity index and a defined set of frequency amplitudes.

8. The computer-implemented method of any one of claims 1 to 7, further comprising a laser speckle image definition process, the process comprising, for each laser speckle image: obtaining an initial laser speckle image, the initial laser speckle image being an image containing a speckle pattern generated by reflection of light irradiating a region of interest of an object, wherein the initial laser speckle image contains the laser speckle image and has a higher resolution than the laser speckle image; processing the initial laser speckle image to identify one or more portions of the initial laser speckle image representing the region of interest; and One or more portions of the identified initial laser speckle image are extracted to define the laser speckle image.

9. The computer-implemented method of claim 8, wherein: The step of processing the initial laser speckle image comprises: identifying any portion of the initial laser speckle image representing a heterogeneous region of the object as a heterogeneous portion, the heterogeneous region comprising one or more of an artery and a vein and one or more capillaries; and The identified heterogeneous portion of the initial laser speckle image is excluded from the one or more identified portions of the initial laser speckle image representing the region of interest.

10. The computer-implemented method of claim 8 or 9, wherein: The step of processing the initial laser speckle image includes identifying one or more portions of the initial laser speckle image containing one or more desired text features of the initial laser speckle image as the portion of the initial laser speckle image representing the region of interest.

11. A computer-implemented method according to any one of claims 1 to 10, wherein: The one or more laser speckle images include a plurality of laser speckle images having different exposure times.

12. A computer program product comprising code which, when executed by a processor circuit, causes the processor circuit to perform the steps of the method according to any one of claims 1 to 11.

13. A non-transitory computer readable medium or data carrier comprising or carrying the computer program product according to claim 12.

14. A processing system for processing one or more laser speckle images, each laser speckle image being an image containing a speckle pattern produced by reflection of light illuminating a region of interest of an object, the processing system being configured to: processing the one or more laser speckle images using a spatial frequency transform to generate spatial frequency data; and A plurality of blood flow velocity indices are defined using the spatial frequency data, each blood flow velocity indice indicating or representing an amount of blood moving at a corresponding different blood flow velocity within the region of interest.

15. A laser speckle imaging system, comprising: The processing system according to claim 14; as well as A laser speckle device is configured to capture the one or more laser speckle images.

Images