CN118714971A - Systems and methods for tissue analysis using remote PPG - Google Patents

Abstract

A tissue analysis system and method processes a hyperspectral image to classify tissue types within the image. The presence or level of tissue oxygenation is estimated for each tissue type as well as PPG perfusion. The estimated tissue oxygenation and PPG perfusion information are combined to provide a tissue state for each tissue type. The system and method use remote PPG sensing.

Description

Systems and methods for tissue analysis using remote PPG

Technical Field

The present invention relates to the analysis of tissue using photoplethysmography (PPG) analysis.

Background Art

Microcirculation is the circulation of blood in the smallest blood vessels present in the vasculature embedded within organ tissues. The structure of the microvasculature is extremely complex because there are so many capillaries in the tissues, where the individual capillaries are so coiled that the blood flow at any given point in any given capillary can be in any direction. For this reason, it can be said that its overall function becomes average. That is, there is an average blood flow rate through each tissue capillary bed, an average capillary pressure within the capillaries, and an average rate of mass transfer between the capillary blood and the surrounding interstitial fluid. This is the perfusion of the tissue.

Organ perfusion is a key indicator of injury and disease, which can include inflammation, stagnant or stopped blood flow, and all pathologies that can lead to global tissue hypoxia and organ dysfunction. Perfusion monitoring can be used to assess these microvascular injuries and many other injuries, such as the progress of healing of burned skin or wounds or restoration of perfusion downstream of vascular lesions, and necrosis in diabetic patients (e.g., foot ulcers, sepsis).

Non-contact PPG imaging is a recent emerging technology that is able to monitor skin perfusion. PPG imaging utilizes readily available cameras and light sources to remotely detect dynamic changes in blood volume beneath the skin and allows the extraction of blood pulsation signals. PPG imaging allows the detailed description of large tissue areas and thus the construction of so-called perfusion maps, which is a great advantage over contact PPG. PPG imaging has been shown to be able to detect skin perfusion disturbances, such as stimulation, temperature changes, and even flow obstruction during pressure measurements, and has even been evaluated for peripheral arterial disease assessment.

An amplitude map representing the amplitude of the PPG signal for each image sensing pixel can be constructed. In addition, a delay map providing a measure of the average time delay between the PPG signal wave for each pixel and the reference PPG signal can also be constructed. Due to small differences in the microcirculatory bed (such as resistance and elasticity of the blood vessels) and different arterial branches supplying the recorded tissue, there are small delays in the arrival of blood pulsations.

For the purposes of the present application involving the processing of contactless PPG signals, perfusion may be defined as the amplitude of the PPG signal of the tissue, ie the amplitude of the pulsatility extracted from the tissue.

It is well known that perfusion levels are not the same everywhere in the body. For example, in the applicant's study, PPG imaging of the hands and feet was studied and it was found that the perfusion level on the palms was higher than on the soles of the feet. This is also true for organs such as the intestine. The small intestine and colon have different functions, are far apart and are supplied by different arterial branches, and therefore have different perfusion levels.

If there are different tissue types in a single image of a remote PPG camera, then general PPG image analysis will average out these differences. For example, if an anastomosis is being performed (e.g., after removing a portion of the intestine) and perfusion levels are measured via PPG imaging, there are different intestines in the image. Image analysis will also not show whether the perfusion differences are physiological or due to pathology (e.g., ischemia, inflammation).

For example, during an open bowel resection, a portion of the colon or small intestine is surgically removed to remove a tumor. First, the surgeon moves the intestine away from the surrounding organs and membranes. The intestine is clamped on both sides of the tumor and blood flow to the section is interrupted. The surgeon will then resect the diseased portion of the intestine and sew the two clamped sides together, which is called an anastomosis. Intestinal anastomosis is performed to create communication between two previously distant portions of the intestine. This procedure restores intestinal continuity after the pathology affecting the intestine is removed. If the anastomosis does not heal properly due to inadequate perfusion, a leak occurs. Leaks are dangerous and can lead to serious infections and require hospitalization and reoperation. These leaks occur in 5% to 10% of patients who undergo a bowel resection. Here, objective perfusion parameters can help identify inadequate healing.

In particular, it would be desirable to be able to measure perfusion in both parts of the anastomosis separately, as perfusion levels are not expected to be the same due to anatomical differences between different tissue types.

Another measure of interest for determining tissue or organ health is tissue oxygenation levels.

Therefore, it would be of interest to be able to monitor perfusion and oxygenation of different tissue types separately to provide a more complete assessment of tissue status information.

US2020/138360 discloses the use of multispectral imaging to measure oxygenation. It also discloses the use of PPG and oxygenation to provide tissue classification.

Zaunseder, Sebastian et al., "Spatio-temporal analysis of blood perfusion by imaging photoplethysmography" (Progress in Biomedical Optics and Imaging, SPIE-Int. Soc. for Optical Engineering, Vol. 10501, February 20, 2018 (XP060099995)) discloses a method for analyzing blood perfusion using remote PPG. A perfusion velocity index is calculated for each pixel of the PPG delay map. The distribution of all indices gives an indication of perfusion.

Summary of the invention

The invention is defined by the claims.

According to an example according to one aspect of the present invention, a system for tissue analysis is provided, comprising:

A processor adapted to receive a hyperspectral image captured by a hyperspectral image sensor and process the hyperspectral image to:

classifying tissue types and segmenting the hyperspectral image into one or more regions corresponding to different tissue types;

Estimate the presence or level of tissue oxygenation for each tissue type;

deriving a PPG perfusion map based on PPG amplitude levels at one or more wavelengths of the hyperspectral image sensor; and

The estimated tissue oxygenation and PPG perfusion information from the PPG perfusion map are combined to derive tissue perfusion and oxygenation status for each tissue type.

The system utilizes remote PPG sensing to derive PPG perfusion maps from PPG amplitude levels. A hyperspectral image sensor is used so that different images contain different spectral contents.

By imaging at different spectral contents, the tissue oxygenation level can be determined. The oxygenation level can be a binary indication (oxygenated or not oxygenated), or it can be the oxygenation level.

Hyperspectral images also enable identification of different tissue types, such as tissue of different organs or different regions of an organ. This image segmentation can be performed from only one wavelength image, or it can use multiple wavelength images. For each individual tissue type, oxygenation information as well as PPG perfusion information are combined to provide a tissue state that takes both into account.

For example oxygenation is obtained from the analysis of multiple wavelengths. Different tissue types can have different perfusions as well as different oxygenations at steady state, so more accurate information is obtained by separating the images into different tissue types and treating them separately. By using hyperspectral imaging to first classify the tissue types (e.g. organs), the classification information can be used to determine a standard perfusion level at steady state. This helps to better assess whether a specific tissue is well perfused. The oxygenation level is extracted only after the first classification step.

The processor may be adapted to classify the tissue type using a machine learning algorithm.This performs image based segmentation and classification.

The tissue types include, for example, tissues for different organ types or tissues for different regions of an organ.

The hyperspectral image, for example, includes a set of 2D images, each 2D image being at a different wavelength in a set of wavelengths.

The processor is for example adapted to estimate the tissue oxygenation using spectral information from hyperspectral images for at least two different wavelengths and to derive the PPG perfusion map from at least one of the 2D images. Various methods are known for extracting blood oxygen saturation from diffuse reflectance spectra and from hyperspectral cameras.

The processor may be further adapted to derive a PPG delay map based on the relative PPG delays between different image regions.

The PPG delay map is obtained from the relative delay of pulsating blood flow reaching the tissue. The delay information is used for different image areas, i.e., image data associated with different pixels of the image sensor.

The PPG delay map can be used as part of the classification of tissue types. For example, different tissue regions can have different delay times. Therefore, the delay information can also assist in tissue classification. For example, a first tissue region can have a delay time within a first range that does not overlap with a delay time within a second range of a second tissue type. Each range can be considered to be a mean plus or minus a multiple (e.g., 1) standard deviation (i.e., a statistical range rather than a range that includes all values).

Thus, different tissue regions have different delay characteristics. This may occur because different tissue types are supplied by different arteries and therefore have different paths and therefore path lengths to the heart. This may occur in the case of different parts of the intestine, and these parts may become adjacent during an open bowel resection. There are resulting intestinal anastomoses with communication between previously distant parts of the intestine.

The relative delay includes, for example, a delay relative to a reference PPG signal, which is an average delay time period for all image regions of a frame of image data. The delay is calculated as the average time delay between the PPG signal wave of each pixel and the reference PPG signal over a specific time period.

The length of the time period includes at least a heartbeat cycle, and a reference PPG signal is obtained as a spatial average of all pixel values of a video (or series of images) as a time-dependent signal modulated at the heart rate. A signal that varies frame by frame over time is created.

Tissue perfusion and oxygenation status include, for example, one of the following:

above normal pulse and normal oxygenation;

higher than normal pulse and lower than normal oxygenation;

normal pulsation and normal oxygenation;

normal pulse and subnormal oxygenation;

below normal pulse and normal oxygenation;

Subnormal pulsation and subnormal oxygenation.

Thus, the tissue status can be obtained taking into account both perfusion and oxygenation levels.

The system, for example, further comprises a hyperspectral image sensor for capturing the hyperspectral image.

The present invention also provides a computer-implemented tissue analysis method, comprising:

receiving a hyperspectral image sensor image;

classifying tissue types and segmenting the hyperspectral image into one or more regions corresponding to different tissue types;

Estimate the presence or level of tissue oxygenation for each tissue type;

deriving a PPG perfusion map based on PPG amplitude levels at one or more wavelengths of the hyperspectral image sensor; and

The estimated tissue oxygenation and PPG perfusion information from the PPG perfusion map are combined to derive tissue perfusion and oxygenation status for each tissue type.

Classifying tissue type is preferably performed using a machine learning algorithm.

The tissue types include, for example, tissues for different organ types or tissues for different regions of an organ.

The hyperspectral image preferably comprises a set of 2D images, each 2D image being at a different wavelength in a set of wavelengths, and wherein the method comprises:

estimating the tissue oxygenation using spectral information from the hyperspectral images for at least two different wavelengths; and

The PPG perfusion map is derived from at least one of the 2D images.

The method may further comprise deriving a PPG delay map based on the PPG relative delays between different image sensor regions. The PPG delay map may be used to classify tissue types.

The invention also provides a computer program comprising computer program code means adapted to implement the method defined above when said program is run on a computer. The invention also provides a processor programmed with the computer program.

These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the invention and in order to more clearly show how it may be implemented, reference will now be made, by way of example only, to the accompanying drawings, in which:

Fig. 1 shows frames of an acquired video of the intestine and a representation of a global PPG signal derived from the video;

FIG2 shows individual PPG signals for individual regions of interest;

FIG3 shows a PPG perfusion map;

FIG4 shows a magnification of a PPG signal extracted from two regions of interest from FIG2 and a PPG delay graph;

Figure 5 shows a hypercube of a hyperspectral camera;

FIG6 shows a system for tissue analysis; and

FIG. 7 illustrates a computer-implemented tissue analysis method.

DETAILED DESCRIPTION

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

It should be understood that the detailed description and specific examples are intended for illustration purposes only and are not intended to limit the scope of the invention when indicating exemplary embodiments of the device, system and method. These and other features, aspects and advantages of the device, system and method of the present invention will be better understood from the following description, the appended claims and the accompanying drawings. It should be understood that the drawings are merely schematic and are 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 parts.

The present invention provides a tissue analysis system and method that processes hyperspectral images to classify tissue types within the image. The presence or level of tissue oxygenation is estimated for each tissue type as well as PPG perfusion. The estimated tissue oxygenation and PPG perfusion information are combined to provide a tissue state for each tissue type. The system and method use remote PPG sensing.

Before describing the system and method of the present invention, the known operation of a remote PPG imaging system and known image processing methods will first be described.

Remote PPG imaging enables determination of tissue perfusion from captured images of tissue of interest (e.g., skin or even tissue beneath the skin). Remote PPG typically uses ambient light used as a broadband white light source and analyzes diffuse and specular reflections for different color components. Remote PPG imaging can be used to construct PPG amplitude maps and PPG delay maps.

For this purpose, a camera or series of cameras (at one or more wavelengths) captures a video of a tissue area (e.g., skin) at a distance. The measurement derived from the video is a remote PPG image, which provides a contactless measurement of the pulse signal by analyzing subtle changes in skin color (or organ color) (i.e., at different wavelengths of light).

The use of remote PPG for inflammation detection has been proposed. It has also been proposed in European Patent Application No. 20214955.5 to measure perfusion based on PPG amplitude levels and information about the distribution of PPG phases, such as the standard deviation or interquartile range of the phase map.

By extracting the pulse signal at each individual location of the skin area (corresponding to each pixel of the camera), a spatial map of the pulse signal can be derived, which shows both amplitude and delay. Thus, this perfusion map represents the amplitude and delay of the PPG signal for each pixel and therefore each location of the skin.

Figure 1(a) shows a representation of a frame of an acquired video of the intestine. By spatially averaging the pixel values in a region of interest (ROI) 10, a signal can be derived from the video, which is modulated with the heart rate. This signal is shown in Figure 1(b) as the normalized intensity of the PPG signal with respect to time.

The PPG signal in Figure 1(b) represents the average value of the entire region of interest.

However, it is also possible to capture separate signals from smaller regions of interest, as shown in regions 1 and 2 in Figure 1(a).

FIG2 shows separate PPG signals, as a first signal 20 for a first region of interest and a second signal 22 for a second region of interest. The two PPG signals are modulated at the same heart rate frequency, but show different amplitudes. By extracting the amplitude from the PPG signal of each separate region of tissue (i.e., from each corresponding pixel of the captured video), a PPG perfusion map can be obtained, as shown in FIG3 .

Figure 3 is a black and white representation. However, the PPG perfusion may be color coded, for example assigning redder colors to areas with higher perfusion and bluer colors to areas with lower perfusion. The PPG amplitude map thus represents the amplitude of the PPG signal for each pixel and therefore each location of the skin or other tissue being analyzed.

Additional valuable information can be obtained by analyzing the PPG signal.

Figure 4(a) shows a zoom-in of PPG signals extracted from two regions of interest from Figure 2. Even though the signals are extracted simultaneously from the same organ, the pulsation arrives slightly earlier in one region than in the other. Due to small differences in the microcirculatory bed (such as the resistance and elasticity of the blood vessels, and the different arterial branches supplying the recorded tissues), there are small delays in the arrival of blood pulsations.

The delay of the PPG signal of each pixel relative to the reference signal can be extracted and used to build a delay map. A delay map is shown in Figure 4.

Since the delay between signals is not always constant but varies during acquisition, an average delay is used between the signals at each pixel.

The average delay represents the average time delay between the PPG signal of each pixel and the reference PPG signal. In this way, a time signal is established. The length of the PPG signal should at least include the heartbeat cycle, so it is subject-dependent.

The average delay is the delay value assigned to each pixel. As can be seen from the graph of Figure 4, there are arrival time differences between the peaks of the PPG signal. Therefore, for each pixel, the average of these delays (relative to the reference signal) is assigned. Finally, a delay map is created where each pixel contains the average delay between the PPG signal of that pixel and the reference PPG signal.

By acquiring a video stream, a series of images are obtained over time, for example at a frame rate of 20 frames per second. To calculate the global PPG signal over time, the pixel values of frame 1 are spatially averaged so that the 2D pixel set produces one value. The pixels of frame 2 are then averaged, and so on.

Finally, a PPG signal is obtained over time (with the same length as the video that has been acquired), wherein each value of the signal is the spatial average of one corresponding frame of the acquired video. By way of example, an image frame comprises, for example, 968×728 pixels.

Thus, the PPG signal of each pixel is compared with a global PPG signal used as a reference. Thus, the value assigned to each pixel "n" of the delay map represents the average delay between the PPG signal in pixel "n" and the global PPG signal used as a reference. Thus, the value of the delay of each pixel in the PPG delay map represents the average delay (in terms of average time shift) between the PPG signal of that pixel and the global PPG signal. For example, the delay is calculated by using a lock-in amplification algorithm.

The delay map thus provides a measure of the average time delay between the PPG signal wave and the reference PPG signal for each pixel for at least one heartbeat cycle.The reference PPG signal is the global PPG signal of Fig. 1(b) extracted from the entire region of interest of the intestine.

Since the reference signal is extracted from the entire region of interest, the PPG signal from a given location of the image is likely to be in phase with the reference.

Similar to the amplitude graph, Figure 4(b) is a black and white representation. However, the PPG delay graph can be color-coded. For example, the Hue, Saturation, Value (HSV) color system is used because it works well with the periodicity of the PPG signal.

To extract the amplitude and delay plots, a lock-in amplification method can be used.

Details on how to calculate the PPG map using the lock-in amplification method can be found in:

(i) Lai M, Shan C, Ciuhu-Pijlman C, Izamis ML. Perfusion monitoring by contactless photoplethysmography imaging (2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019) April 8, 2019 (pp. 1778-1782) IEEE); and

(ii)Lai M, Dicorato CS, de Wild M, Verbakel F, Shulepov S, Groen J, Notten M, Lucassen G, van Sambeek MR, Hendriks BH. Evaluation of a non-contact Photo-Plethysmographic Imaging (iPPG) system for peripheral arterial disease assessment (Medical Imaging 2021: Biomedical Applications in Molecular, Structural, and Functional Imaging February 15, 2021 (Volume 11600, Page 116000F). International Society for Optics and Photonics).

Since remote PPG technology has shown great advantages for remotely and non-invasively assessing skin-level perfusion, PPG imaging can be converted to organ perfusion assessment for detecting perfusion of microvasculature tissue beds beneath the organ surface without any modification to the current acquisition setup.

Remote PPG sensing can be performed using a standard image sensor that receives reflected ambient light.

However, there is also the option of using hyperspectral imaging. Hyperspectral imaging (HSI) is an emerging imaging modality for medical applications that offers great potential for non-invasive disease diagnosis and surgical guidance. The goal of hyperspectral imaging is to collect a three-dimensional dataset of spatial and spectral information, called a hypercube.

As shown in Figure 5, a hypercube is a three-dimensional data set 50 that includes a two-dimensional image 52 at each wavelength in a set of wavelengths. Figure 5 also shows the reflectivity curves (ie, spectral signatures) of the pixels in each image.

The present invention utilizes hyperspectral imaging so that tissue in the field of view can be classified and different tissue types, such as different organs in the abdominal region, can be segmented.

The advantages of hyperspectral imaging will now be explained. A conventional RGB camera has sensors that are sensitive to the red bell-shaped wavelength range, the green range and the blue. The spectral signature of an object cannot be identified because there are only 3 points (1 red, 1 green, 1 blue) that are not even specific to any individual wavelength. HSI allows the acquisition of multiple images of the same object at once, where each image represents a narrow wavelength range of the electromagnetic energy spectrum, i.e. a narrow spectral band. HSI cameras can acquire several spectral bands, not only in the visible range, but also in the UV and IR range. For example, HSI sensors that acquire 25 bands in the IR range and 16 bands in the visible range are known. There are also HIS sensors with much more bands.

Due to the large number of wavelengths, the spectral characteristics of the recorded tissue can be extracted and the tissue can be identified and classified. Classification based on the spectral characteristics of tissue can be much more accurate using HSI than using RGB images.

Tissue oxygenation can be calculated by combining in the model several wavelengths of interest acquired with an HSI camera. An RGB camera has red, green and blue channels that are sensitive to a much wider WL range, so extracting oxygenation from a normal RGB camera is not possible. The use of an HSI camera enables determination of tissue classification, tissue oxygen levels and processing of PPG maps. An HSI camera has, for example, at least 10 spectral bands, for example at least 20 spectral bands.

Therefore, the presence or level of tissue oxygenation can be determined for each tissue type, as well as the PPG amplitude levels measured separately for different tissue types. Perfusion measurements only require one wavelength of the hyperspectral image, but multiple wavelengths can be used to determine perfusion. The estimated tissue oxygenation from the PPG perfusion and the PPG perfusion information are then combined to derive the tissue perfusion and oxygenation status for each tissue type.

Figure 6 shows a system 60 for tissue analysis. The system includes one or more hyperspectral imaging cameras 62 and a processor 64 adapted to receive hyperspectral images and process the images.

Image processing classifies tissue types (e.g., tissue 1 and tissue 2) and segments the hyperspectral image into one or more regions corresponding to different tissue types. The presence or level of tissue oxygenation is estimated for each tissue type, and the PPG amplitude level is obtained from the PPG perfusion map (at one or more wavelengths of the hyperspectral camera) for each tissue type. The estimated tissue oxygenation and PPG perfusion information are combined to derive tissue perfusion and oxygenation status for each tissue type.

This information is output, for example, by the display 66, which may provide a PPG graph (amplitude graph and/or delay graph) over time or as a PPG signal.

Tissue classification uses, for example, machine learning algorithms such as support vector machines (SVM) and convolutional neural networks (CNN).

The field of view of the hyperspectral camera can include all organs inside the abdomen or only parts of the internal organs. By classifying the different organs present in the field of view, comparison of perfusion between different organs is prevented, which may have different perfusion levels due to their different functions and which may even be supplied by different arterial branches. For each classified tissue type, a standard perfusion level at steady state can be obtained for comparison with the derived tissue perfusion state.

Oxygenation analysis gives an indication of the amount of oxygen in the tissue, while PPG imaging extracts the perfusion of the tissue in terms of blood pulsation. This is an indication of the blood flowing inside the organ.

Oxygenation and perfusion (PPG pulsatility) are combined to give a better indication of the state of organ tissue, for example by avoiding necrosis. As shown in FIG6 , the system outputs perfusion and oxygenation information (“P,O”) for different tissue types. The information is displayed graphically as described above, and therefore FIG6 is a simplified representation of the system's functionality.

The combination of oxygenation and perfusion enables the identification of six potential scenarios:

(i) High (above normal) pulsation and oxygenation: This condition may indicate tissue inflammation/irritation with normal oxygenation levels.

(ii) High pulsation and no oxygenation: This condition may indicate tissue inflammation/irritation, but very low oxygen levels (hypoxia).

(iii) Normal pulsation and oxygenation: This is the normal state of a healthy organ.

(iv) Normal pulsation and no oxygenation: The heart is pumping blood to the organs, but the oxygen level is very low (hypoxia).

(v) No pulse and oxygenation: The heart does not pump blood to the organs, but oxygen levels are still adequate.

(vi) Pulseless and non-oxygenated: This is the worst condition, where blood flow is blocked and oxygen levels are low.

FIG. 7 illustrates a computer-implemented tissue analysis method 70 .

The method includes receiving a hyperspectral image sensor image at step 72 .

Optionally, in step 74 a PPG delay map is generated.

In step 76 , different tissue types are classified and the hyperspectral image is segmented into one or more regions corresponding to the different tissue types.

Image segmentation may be based solely on image processing of one or more of the hyperspectral images. However, when generating a PPG delay map as shown in FIG7 , different PPG delays for different regions of the image may also be used (e.g., by a machine learning algorithm) as part of segmenting the image into different tissue types.

For example, a range of PPG delay times from a PPG delay map may be used as part of the classification. Different regions may have different global arrival times (relative to a reference or relative to each other) and delay spreads, meaning that they have substantially non-overlapping ranges (e.g., within a number of standard deviations, such as one standard deviation). Thus, these differences may form part of the machine learning classification.

By way of example, organs in the abdominal region may be segmented into spleen 90 , colon 92 , bladder 94 , peritoneum 96 , and small intestine 98 .

In step 78, the presence or level of tissue oxygenation is estimated for each tissue type. Various methods are known for extracting blood oxygen saturation from diffuse reflectance spectra and from hyperspectral cameras. See, for example, "Stratonnikov AA, Loschenov VB. Evaluation of blood oxygen saturation in vivo from diffuse reflectance spectra. Journal of biomedical optics. 2001 Oct; 6(4): 457-67".

In step 80 , a PPG perfusion map is derived based on the PPG amplitude levels at one or more wavelengths of the hyperspectral image sensor.

The estimated tissue oxygenation and PPG perfusion information from the PPG perfusion map are used to derive tissue perfusion and oxygenation status for each tissue type, and the tissue status is output in step 82 (eg, by displaying a color-coded image).

The invention may be used to assess the status of an organ during minimally invasive surgery in order to reduce the risk of necrosis or even assess the restoration of proper blood flow to the tissue.

The system's hyperspectral or other imaging sensors may, for example, include a charge coupled device (CCD) to record images and output them in an analog or digital electronic format or signal. These sensors may include cameras, such as hyperspectral cameras. Such imaging sensors or cameras are known in the art and are available from several suppliers.

The system, processor, or processor circuit may have an input interface communicatively coupled to the processor or processor circuit, wherein the input interface is configured to receive an image from the imaging sensor for use by the processor. For example, the image may be received in an analog or digital electronic format or signal.

Functions and methods related to image processing, data processing and/or output or image generation for output to a user as described herein can be implemented in a general-purpose computer, processor, processor circuit. By way of example, suitable processors or processor circuits include general-purpose processors, special-purpose processors, conventional processors, digital signal processors (DSPs), multiple microprocessors, one or more microprocessors associated with a DSP core, controllers, microcontrollers, application-specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) circuits, any other type of integrated circuit (IC) and/or state machines. Such processors can be made with semiconductor technology as known in the art.

The system or processor (processor circuit) preferably includes a memory accessible by the processor.

The system or processor (processor circuit) includes an output interface communicatively coupled to the processor (processor circuit) and configured to output result data and/or images to a user interface or other device for further processing of the results.

The system preferably includes a user interface to be coupled to or coupled to an output interface of the system or processor (processing circuitry). An exemplary user interface includes a display device having an input for coupling to the output interface and receiving result data to generate an image or display of results visible to a user.

Functions and methods related to, for example, image and/or data processing and/or data or image generation as described herein may be implemented in a computer program, software or firmware incorporated in a non-transitory computer-readable storage medium and/or downloadable from a communications network for execution by a general purpose computer, processor or processor circuit.

Thus, a computer program, software or firmware may be stored/distributed on a suitable medium, optical storage medium or solid-state medium provided with or as part of other hardware. Examples of non-transient computer-readable storage media include read-only memory (ROM) (such as electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM)) or random access memory (RAM) (such as dynamic RAM (DRAM) or static RAM (SRAM)), registers, cache memory, semiconductor memory devices, magnetic media (such as internal hard disks and removable disks), magneto-optical media, and optical media (such as CD-ROM disks and digital versatile disks (DVDs)). The medium may be a memory of a system that is accessible by a processor or processor circuit of the system.

Alternatively or additionally, the computer program, software or firmware may be stored and distributed and downloadable from a communications network such as a Wide Area Network (WAN), a Local Area Network (LAN) or a Wireless LAN (WLAN) such as the Internet or via other wired or wireless telecommunications systems such as, for example, a 3G, 4G or 5G network.

The system can operate at a low frame rate (e.g., 20 frames per second or less) and in a short duration (such as 20 seconds or less, such as 10 seconds or less). PPG perfusion can be based on infrared or near infrared or other wavelengths such as green. Multiple wavelengths can be used to obtain a better PPG signal.

The computer program may be stored/distributed on suitable media, such as optical storage media or solid-state media provided together with or as part of other hardware, but the computer program may 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 the claims or the description, it should be noted that the term "suitable for" is intended to be equivalent to the term "configured to". If the term "arranged" is used in the claims or the description, it should be noted that the term "arranged" is intended to be equivalent to the term "system", and vice versa.

Those skilled in the art can understand and implement variations to the disclosed embodiments when practicing the claimed invention by studying the drawings, the disclosure and the claims. In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality.

The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Any reference signs in the claims should not be construed as limiting the scope.

Claims (15)

1. A system for tissue analysis, comprising: A processor (64) adapted to (72) receive a hyperspectral image captured by a hyperspectral image sensor and process the image to: (74) deriving a PPG delay map based on the relative PPG delays between different image regions; (76) classifying tissue types and segmenting the hyperspectral image into one or more regions corresponding to different tissue types using different PPG delays for different regions of the image; (78) estimating the presence or level of tissue oxygenation for each tissue type; (80) deriving a PPG perfusion map based on the PPG amplitude levels at one or more wavelengths of the hyperspectral image sensor; and (82) Combining the estimated tissue oxygenation and PPG perfusion information from the PPG perfusion map to derive tissue perfusion and oxygenation status for each tissue type. 2. The system of claim 1, wherein the processor is adapted to classify the tissue type using a machine learning algorithm. 3. The system according to claim 1 or 2, wherein the tissue types include tissues for different organ types or tissues for different regions of an organ. 4. The system of any one of claims 1 to 3, wherein the hyperspectral image comprises a set of 2D images, each 2D image being at a different wavelength in a set of wavelengths. 5. The system of claim 4, wherein the processor is adapted to estimate the tissue oxygenation using spectral information from hyperspectral images for at least two different wavelengths and to derive the PPG perfusion map from at least one of the 2D images. 6. The system of any one of claims 1 to 5, wherein the relative delay comprises a delay relative to a reference PPG signal, the delay being an average delay period for all image regions of a frame of image data. 7. The system of any one of claims 1 to 6, wherein the tissue perfusion and the oxygenation state comprises one of: above normal pulse and normal oxygenation; higher than normal pulse and lower than normal oxygenation; normal pulsation and normal oxygenation; normal pulse and subnormal oxygenation; below normal pulse and normal oxygenation; Subnormal pulsation and subnormal oxygenation. 8. The system according to any one of claims 1 to 7, further comprising a hyperspectral image sensor for capturing the hyperspectral image. 9. The system according to any one of claims 1 to 8, wherein the processor (64) is further adapted to obtain a standard perfusion level at steady state for each classified tissue type for comparison with the derived tissue perfusion state. 10. A computer-implemented tissue analysis method comprising: (72) receiving a hyperspectral image sensor image; (74) deriving a PPG delay map based on the relative PPG delays between different image regions; (76) classifying tissue types and segmenting the hyperspectral image into one or more regions corresponding to different tissue types using different PPG delays for different regions of the image; (78) estimating the presence or level of tissue oxygenation for each tissue type; (80) deriving a PPG perfusion map based on the PPG amplitude levels at one or more wavelengths of the hyperspectral image sensor; and (82) Combining the estimated tissue oxygenation and PPG perfusion information from the PPG perfusion map to derive tissue perfusion and oxygenation status for each tissue type. The method according to claim 10 , wherein the tissue types include tissues for different organ types or tissues for different regions of an organ. 12. The method of claim 10 or 11, wherein the hyperspectral image comprises a set of 2D images, each 2D image being at a different wavelength in a set of wavelengths, and wherein the method comprises: estimating the tissue oxygenation using spectral information from the hyperspectral images for at least two different wavelengths; and The PPG perfusion map is derived from at least one of the 2D images. 13. The method according to any one of claims 10 to 12, comprising obtaining a standard perfusion level at steady state for each classified tissue type for comparison with the derived tissue perfusion state. 14. A computer program comprising computer program code means adapted to implement the method according to any of claims 10 to 13, when said program is run on a computer. 15. A processor programmed with a computer program according to claim 12.