Determining oxygenation of imaged tissue

Abstract

The concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in imaged tissue comprises receiving a measured attenuation spectrum 601 and fitting the measured attenuation spectrum to a model atten

Description

BACKGROUND During or after surgery, it is often helpful to assess the level of perfusion in specific parts of the patient's body. Perfusion is the delivery of blood through the capillary bed into biological tissue. If tissue is well perfused, then it has high oxygen saturation, i.e. it has a good blood supply which is delivering more oxygen than it needs. If tissue is poorly perfused, then is has low oxygen saturation, i.e. it has a poor blood supply which is not delivering sufficient oxygen to it. One example surgery for which it is useful to assess tissue perfusion is colorectal tumour surgery. During colorectal tumour surgery, a section of bowel that contains the tumour is removed, and the two open bowel ends reconnected by a connection called anastomosis. Once the anastomosis has been performed, the seal is checked for leakage. Leakage from the inside of the bowel to the outside of the bowel can cause major complications such as sepsis. Although there are different factors that can influence the probability of a leakage in the anastomosis, it is known that poor blood supply (and hence poor perfusion and oxygenation) of the anastomosis increases the risk of anastomosis leakage. Therefore, assessing perfusion after performing an anastomosis is important for identifying anastomosis leakage. It is known to use the bolus passage of indocyanine green (ICG) to confirm the presence of perfusion and the assessment of blood supply after the anastomosis has been completed. ICG is a fluorescent dye that binds tightly to plasma proteins and becomes confined to the vascular for a specific period of time until it is removed from circulation by the liver (it has a half-life of 150-180 seconds). Therefore, it is known to administer ICG intravenously, and once it is in circulation with the blood flow measure its fluorescence. This fluorescence is used as an indication of the blood flow in a specific anatomical tissue at a specific point in time. Whilst useful for providing an instantaneous assessment of perfusion, this ICG fluorescence method is limited to only providing an indication of perfusion at a specific point in time. It cannot be used to provide a continuous measurement of the perfusion of an anatomical tissue. It also does not provide a quantitative measurement of tissue it is invasive in the sense that it requires a chemical (ICG) to be added to the blood stream. SUMMARY OF THE INVENTION According to an aspect of the invention, there is provided a method of determining the concentration of oxyhaemoglobin HbOzand deoxyhaemoglobin Hb in an imaged tissue, the method comprising: receiving a measured attenuation spectrum of the imaged tissue, the measured attenuation spectrum comprising n light attenuation measurement sets of the imaged tissue, each light attenuation measurement set at a different one of n wavelength bands, each light attenuation measurement set comprising a plurality of light attenuation measurements, one for each pixel or group of pixels in the image of the tissue; fitting the received measured attenuation spectrum to a model attenuation spectrum, the model attenuation spectrum comprising a linear combination of a haemoglobin spectrum and a linear background spectrum, the fitting comprising: solving a set of n simultaneous equations, each equation equating a light attenuation measurement set to the model attenuation spectrum for the wavelength band of the light attenuation measurement set, wherein the solving comprises determining values of the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients which causes the measured attenuation spectrum to best fit the model attenuation spectrum. The haemoglobin spectrum may comprise a linear combination of an oxyhaemoglobin HbCh term that varies as a function of the wavelength band of the light attenuation measurement set, and a deoxyhaemoglobin Hb term that varies as a function of the wavelength band of the light attenuation measurement set. The linear background spectrum may comprise a linear combination of a term that varies as a function of the wavelength band of the light attenuation measurement set, and an offset term. The model attenuation spectrum A(At) may be defined as: A (Aj) = pcHboEHbo(Ai) + pcHbsHb(Ai) + .40 + AjAi where p is the pathlength of the light travel between emission by a light source and reception by a light sensor, cHb0 is the concentration of oxyhaemoglobin HbO2, cHb is the concentration of deoxyhaemoglobin Hb, £« / ,0(^-1) is an extinction spectra of oxyhaemoglobin HbO2 at wavelength Ai, £Hb(At) is an extinction spectra of deoxyhaemoglobin Hb at wavelength Ai, >l0 and are background coefficients of the linear background. Solving the set of n simultaneous equations may comprise determining values for the concentration of oxyhaemoglobin HbO2, the concentration of deoxyhaemoglobin Hb, and the background coefficients which causes the measured attenuation spectrum to best fit the model attenuation spectrum by minimising the squared difference of the measured attenuation spectrum and the model attenuation spectrum. The method may further comprise determining haemoglobin saturation from the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb. The method may further comprise outputting the haemoglobin saturation to a surgeon's display. The method may comprise displaying the haemoglobin saturation on the surgeon's display by overlaying the haemoglobin saturation on a video feed from a surgical endoscope also displayed on the surgeon's display. The method may comprise processing the haemoglobin saturation to generate a transparent haemoglobin saturation image, then overlaying the transparent haemoglobin saturation image on the video feed from the surgical endoscope such that the video feed from the surgical endoscope remains visible on the surgeon's display. The method may further comprise determining total haemoglobin concentration from the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb. The method may further comprise outputting the total haemoglo surgeon's display. The method may comprise displaying the total haemoglobin concentration on the surgeon's display by overlaying the total haemoglobin concentration on a video feed from a surgical endoscope also displayed on the surgeon's display. The method may comprise processing the total haemoglobin concentration to generate a transparent total haemoglobin concentration image, then overlaying the transparent total haemoglobin concentration image on the surgical endoscope such that the video feed from the surgical endoscope remains visible on the surgeon's display. The n wavelength bands may be in the visible spectrum. The n wavelength bands may be within the range 520nm to 580nm. n may be greater than 3. The n light attenuation measurement sets may be time multiplexed with respect to each other. The method may further comprise iteratively repeating the method above as to continuously determine concentrations of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in the imaged tissue. The method may further comprise continuously determining haemoglobin saturation and / or total haemoglobin concentration from the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb, and displaying the haemoglobin saturation and / or total haemoglobin concentration on a surgeon's display at a rate of 1 / n the frame rate. The method may further comprise: comparing each light attenuation measurement for each pixel or group of pixels in the image of the tissue to a light attenuation threshold value; and if the light attenuation measurement does not exceed the light atten discarding the determined values of the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels. The method may further comprise interpolating values for the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels from the determined values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels. The method may further comprise: receiving the intensity of each pixel or group of pixels in the image of the tissue; comparing the intensity of the pixel or group of pixels in the image of the tissue toa pixel intensity threshold value; and ifthe intensity of the pixel or group of pixels in the image of the tissue does not exceed the pixel intensity threshold value, discarding the determined values of the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels. The method may further comprise interpolating values for the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels from the determined values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels. The method may further comprise: comparing the determined total haemoglobin concentration of each pixel or group of pixels in the image of the tissue to a total haemoglobin concentration threshold value; and if the determined total haemoglobin concentration of each pixel or group of pixels in the image of the tissue does not exceed the total haemoglobin concentration threshold value, discarding the determined values of the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels. The method may further comprise interpolating values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels from the determined values for the concentration o and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels. BRIEF DESCRIPTION OF THE FIGURES The present invention will now be described by way of example with reference to the accompanying drawings. In the drawings: Figure 1 illustrates the absorption spectra of haemoglobin and water in tissue, and the scattering coefficient of tissue; Figure 2 illustrates the absorption spectra of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in the visible and NIR wavelength ranges; Figure 3 illustrates a light source; Figure 4 illustrates an arrangement of a light source and a light sensor; Figure 5 illustrates a schematic system for processing images of tissue to determine oxygenation information about the imaged tissue; Figure 6 illustrates a method by which the multi-spectral feed from a light sensor may be processed to generate oxygenation information of imaged tissue; and Figure 7 illustrates steps performed by oxygenation determination logic. DETAILED DESCRIPTION The following describes a system and method of determining the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in an imaged tissue from a measured attenuation spectrum of that imaged tissue. The determined concentrations of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb may subsequently be used to determine tissue oxygenation parameters such as haemoglobin saturation and total haemoglobin concentration. These parameters may then be displayed on a surgeon's display. The surgeon may use the displayed tissue oxygenation parameters to assess tissue perfusion. Haemoglobin content and saturation in the capillary bed of tissue can be used to represent oxygenation of the tissue. Haemoglobin molecules have four oxygen-binding sites. A haemoglobin molecule is said to be "saturated" with oxygen when all of its four oxygen binding sites are occupied with oxygen. A saturated haemoglo oxyhaemoglobin HbO2. Figure 1 illustrates the absorption spectrum of haemoglobin 101 and the absorption spectrum of water 102 at typical concentrations found in anatomical tissue. Also illustrated is the scattering coefficient of tissue 103. These spectra are illustrated in both the visible (300-700nm) and near infrared (NIR) (800-1200nm) wavelength bands. The haemoglobin and water spectra are shown magnified by a factor of 40 in the NIR region. The haemoglobin spectrum 101 is of the same order of magnitude to the scattering coefficient of tissue 103 in the visible wavelength bands. However, in the NIR wavelength bands, the magnitude of the haemoglobin spectrum is a factor of 40 smaller than the scattering coefficient of tissue. The spectral features of the haemoglobin spectrum in the visible range are prominent and distinct - a double hump. This is easily distinguishable from the scattering coefficient. Conversely, in the NIR range, there are no such easily identifiable spectral features of the haemoglobin spectrum. It is more comparable in shape to that of the scattering coefficient in the NIR range. Thus, the scattering coefficient has a greater distorting effect on an attenuation spectrum of tissue in the NIR range compared to the visible range. Additionally, as noted above, the absorption spectrum of haemoglobin has a magnitude which is a factor of 40 times lower in the NIR range than the visible range. The gradient of the scattering coefficient is relatively small, thus it adds a similar offset to all the absorption values in the visible wavelength bands. This means the individual oxyhaemoglobin and haemoglobin spectra are equally disturbed by the scattering coefficient spectrum, and hence remain distinguishable from each other. The high concentration of haemoglobin in tissue (40-60pmoles / L) and its strong absorption in the visible range causes the haemoglobin protein to dominate the absorption coefficient in the visible range. The method described below utilises this to determine the concentration of oxyhaemoglobin HbO? and deoxyhaemoglobin Hb in an imaged tissue from a measured attenuation spectrum of that imaged tissue. These concentrations may then subsequently be used to determine tissue oxygenation. Figure 2 illustrates the absorption spectra of oxyhaemoglobin HbCh 201 and deoxyhaemoglobin Hb 202 in the visible and NIR wavelength ranges. These spectra are known as specific extinction spectra. They are of similar magnitude in th particular in the 500-600nm range. They have distinct spectral features in the visible range, making them easily separable spectra. Specifically, the deoxyhaemoglobin spectrum has a single hump in the 500-600nm range, whereas the oxyhaemoglobin spectrum has a doublehump in the 500-600nm range. These spectral features are thus overlapping in the 500-600nm range. The characterising differences in the oxyhaemoglobin and deoxyhaemoglobin spectra enable an attenuation spectrum of tissue to be measured for light incident on the tissue in a wavelength band in the visible range, and that attenuation spectrum split into a component which is due to absorption by oxyhaemoglobin and a component which is due to absorption by deoxyhaemoglobin. As described below, this enables the concentration of each of oxyhaemoglobin and deoxyhaemoglobin in the tissue to be calculated. The attenuation spectrum of the tissue may be generated in a non-invasive manner. Specifically, the tissue may be illuminated with light in specific wavelength bands of the visible spectrum. These wavelength bands may be with the range 300-700nm. The wavelength bands may be within the range 500-600nm. The wavelength bands may be within the range 520-580nm. A light sensor, ideally co-located with the source of the illumination, detects light reflected back from the tissue. The negative log of the ratio of the reflected and emitted light is the attenuated light. In the example in which a minimally invasive surgery is being performed, an endoscope is present in the patient to illuminate and image the surgical siteforthe surgeon. The endoscope comprises a light source and a camera. The video feed from the camera is displayed on a surgeon's display. The standard light source of the endoscope may be replaced with a modified light source capable of both illuminating the surgical site and providing suitable emitted light to result in an attenuation spectrum from the tissue which can be used in the method described herein. Alternatively, the standard light source of the endoscope may be retained, and an additional light source capable of providing suitable emitted light to result in an attenuation spectrum from the tissue may be additionally used. An example light source is shown in figure 3. The light source is a pulsed light source 301. It comprises a plurality of LEDs, each of which emits light in a different wavelength band in the visible spectrum. The LEDs can be turned on and off independently. of the different LEDs which are used to generate the attenuation spectrum may be nonoverlapping. These wavelength bands may contiguously cover a band of the visible spectrum. Light from each of these LEDs may be emitted in turn to illuminate the tissue. An additional LED which emits white light may be used in between the emission of light from the different LEDs. This enables easy separadon of the reflected spectra. In the light source shown in figure 3, seven LEDs are used. These may be a single white (W) and six green-illumination LEDs (Gi...G6). The green illumination LEDs are in the wavelength range 520-580nm. For example, Gi to Ge may correspond to 520, 532, 540, 550, 561 and 580nm. Emitted light may be pulsed in the sequence W, Gi, W, G2, W, G3, W, G4, W, G5, W, G6. The light from each LED is focussed into an input fibre bundle 303. The light is optically coupled into the input fibre bundle. Optical elements such as lenses may be used to couple the light into the input fibre bundle. The light from each LED may be filtered using a narrowband filter to generate the desired illumination wavelength band (Gi...G6) before being directed to the input fibre bundle. Each input fibre bundle has the same number of fibres as the number of LEDs. Thus, in the example of figure 3, there are seven fibres in each input fibre bundle. Each input fibre bundle is connected to an input arm of a fibre mixer 304. The fibre mixer 304 is used to rearrange the fibres in the input fibre bundles by taking a fibre from each input fibre bundle and collating those together into an output fibre bundle 305. The fibre mixer 304 has output arms connected to each output fibre bundle 305. Thus, there are the same number of output fibre bundles as there are input fibre bundles. Each output fibre bundle has a fibre for transporting each wavelength band being used to generate the attenuation spectrum plus a fibre for transporting white light. In the example of figure 3, each output fibre bundle has seven fibres. Six fibres are mounted around a single central fibre. The light from the output fibre bundle passes through a collimating lens 306 to direct it uniformly at the tissue. The output fibre bundles 305 are mounted evenly around a central camera lens 401, as shown in figure 4. When light is emitted from each LED sequentially, a fibre in each fibre bundle is illuminated with the wavelength band of that LED. Thus, an even illumination of the field of view is achieved for each wavelength. Thus, the tissue is illuminated by pulsing the different wavelengths from the light source one-by-one in a time-multiplexed manner. In between each wavelength, white light is pulsed. The pulse rate may be synchronised to the camera frame rate. For example, for a master frame rate of 120 frames per second (fps), half the frames are used to generate white light for use in generating the RGB video feed, and the other half of the frames are used to generate the multi-spectral imaging described herein. Thus, the RGB frame rate is half the master frame rate, in this case 60fps. And the multi-spectral frame rate is also half the master frame rate, in this case 60fps. The rate at which the attenuation spectrum can be generated and the concentrations of oxyhaemoglobin and deoxyhaemoglobin determined is then given by: Rate of Hb02 and Hb concentration determination = master frame rate [equation 1] 2n where n is the number of wavelength bands used to illuminate the tissue. Thus, the tissue is illuminated with n different wavelengths in turn by the light source. In the example described above n=6. However, n may be any number equal to or greater than 3. For example, n may be in the range 3 <n <10. A light sensor measures the light reflected back to the camera. This light sensor may be the camera which is already present in the endoscope. In this case, the camera may obtain both the RGB image (from the white light) and each narrow-band image (from the narrow-band wavelengths) resulting from the pulsed illumination of the tissue. Alternatively, a standard camera may be retained to obtain the RGB image, and an additional light sensor may be used to obtain the narrow-band images. The additional light sensor may be in the endoscope. Alternatively, the additional light sensor may be proximal to, but separate from, the endoscope. The light sensor receives light reflected back from the tissue. Some light that was emitted by the light source will have been absorbed by the tissue, and some light will have been scattered within the tissue. The remainder of the light is reflected back and received by the light sensor. An image of the tissue may be generated using the reflected light rece for each illuminated wavelength. Thus, in each pulse sequence in which the tissue is illuminated by each of the non-white n LEDs once in turn, a time multiplexed hypercube image may be generated which has n images of the illuminated tissue, one image for each illuminated wavelength. In the example of figure 3, a hypercube of six images is generated for each pulse sequence iteration. In the example in which the light source pulses at a rate of 60fps, the hypercube is generated at a rate of 60 / n fps. Thus, for six illuminated wavelengths, the hypercube is generated at a rate of lOfps. Figure 5 illustrates an example schematic system for processing images received from the camera and light sensor so as to determine oxygenation information about the imaged tissue, which may subsequently be displayed on the surgeon's display. The images representing the reflected light spectra from the tissue are fed from the light sensor 501 to a control system 502 outside the body. The control system may also receive the RGB video feed from the camera 503. As mentioned above, the camera 503 and light sensor 501 may be the same component. The control system comprises an image processor 504 for processing the images received by the camera. The control system also comprises a memory 505 coupled to the processor 504. The memory 505 stores computer executable instructions. The processor 504 executes the computer executable instructions in order to perform the processing steps described herein. The memory 505 may be a non-transitory computer readable medium with the computer executable instructions stored on it which the processor reads and performs to execute the processing steps described herein. The image processor may perform standard image processing techniques on the RGB video feed received from the camera 503, such as to correct the white balance. The control system 502 also includes oxygenation determination logic 506. The oxygenation determination logic receives the measured multispectral images from the light sensor of the reflected light at each illuminated wavelength. The control system also comprises a memory 508 coupled to the oxygenation determination logic 506. The memory 508 stores computer executable instructions. The oxygenation determination logic 506 । executable instructions in order to perform the steps described herein. The memory 508 may be a non-transitory computer readable medium with the computer executable instructions stored on it which the oxygenation determination logic 506 reads and performs to execute the steps described herein. As will be described in more detail below, the oxygenation determination logic 506 processes the received measurement data to work out oxygenation information about the imaged tissue. The oxygenation determination logic 506 outputs this oxygenation information to the image processor 504. The image processor may combine the oxygenation information with the processed RGB video feed, and output the combined video feed to the surgeon's display 507. Figure 6 illustrates a method by which the multi-spectral feed from the light sensor 501 may be processed to generate oxygenation information of the imaged tissue. Figure 7 illustrates, in more detail, steps performed by the oxygenation determination logic. Initially, the measured attenuation spectrum of the imaged tissue is calculated according to the following equation: Attenuated light = — Log10 (reflected lls^t\ u emitted light ) [equation 2] Equation 2 may be calculated on a pixel-by-pixel basis to generate the whole attenuation spectrum for each wavelength band (Gi...G6). The measured attenuation spectrum comprises n light attenuation measurement sets of the imaged tissue. This may be presented as the hypercube described above, which comprises n images, each image being one of the light attenuation measurement sets. Each of these light attenuation measurement sets is at a different one of the n wavelength bands. Each light attenuation measurement set comprises a plurality of light attenuation measurements, for example one for each pixel of the image of the tissue. In the example of the hypercube images, each hypercube image comprises a light attenuation measurement set comprising light attenuation measurements which are pixel values. The oxygenation determination logic receives the measured attenuate of figure 6. The attenuation measurement sets are accumulated in memory 701 until all n attenuation measurement sets from a single pulse sequence have been received. This takes a finite time since the tissue is illuminated with the pulses in the pulse sequence sequentially, and thus the n attenuation measurement sets are measured sequentially in time by the light sensor. Suitably, memory 701 is contiguous and stores the hypercube images as individual spectra elements comprising individual pixel values. Each hypercube object may be time indexed. Once all n attenuation measurement sets have been received and stored in memory 701, the hypercube is determined to be complete. The method then moves to step 602 at which the oxygenation determination logic 506 fits the received measured attenuation spectrum to a model attenuation spectrum stored in model attenuation spectrum memory 704. The model attenuation spectrum may comprise a linear combination of a haemoglobin spectrum and a linear background spectrum. Suitably, the haemoglobin spectrum of the model attenuation spectrum comprises a linear combination of an oxyhaemoglobin HbOzterm that varies as a function of the wavelength band of the light attenuation measurement set, and a deoxyhaemoglobin Hb term that varies as a function of the wavelength band of the light attenuation measurement set. The linear background spectrum accounts for scattering of light in the tissue, light reflected by the tissue but at an angle not captured by the light sensor, and light received by the light sensor from other sources. The linear background spectrum may comprise a linear combination of a term that varies as a function of the wavelength band of the light attenuation measurement set, and an offset term. Mathematically, the model attenuation spectrum may be defined as: A^^ = pcHbo£Hbo{Xi') + pcHbEHb(At) + Ao + A^ [equation 3] where p is the pathlength of the light travelled between emission by a light source and reception by a light sensor (which can be assumed to be the same for HbO2 and Hb, and independent of wavelength), cHb0 is the concentration of oxyhaemoglobin HbO2, cHb is the concentration of deoxyhaemoglobin Hb, EHb0(Xi) is an extinction spectra of oxyhaemoglobin H bO2 at wavelength A,, is an extinction spectra of deoxyhaemoglobin Hb at wavelength Aj, and / l0 and At are background coefficients of the linear backgrour to be 1.8mm. The relationship between the attenuation and absorption coefficient is not linear in tissue. This is due to the amount of light which is scattered by the tissue. The absorption and scattering coefficients cannot be separated from a measurement of intensity only. Thus, only the haemoglobin absorption spectra are used in the following, with the scattering spectrum treated as part of the background spectrum. The model attenuation spectrum does not fit to the spectra of HbO2 and Hb highly accurately. However, over a narrow wavelength band in the visible spectrum, which is the range used to illuminate the tissue as described above, the inventors have found sufficiently accurate values for the concentration of HbO2 and Hb can be determined if the measured attenuation spectrum is spectrally fitted to the model attenuation spectrum of equation 3. Optionally, the oxygenation determination logic 506 may first spatially smooth the attenuation measurement sets to remove noise, so as to improve the signal to noise ratio. A kernel may be used to do this. The kernel may filter the attenuation measurement sets according to a commonly known method. For example, the kernel may comprise a matrix which is convoluted with the image to remove noise from the image, thereby smoothening structures in the image. At step 703, the oxygenation determination logic then solves a set of n simultaneous equations. Each equation equates a light attenuation measurement set to the model attenuation spectrum for the wavelength band of the light attenuation measurement set. For example, for a pulse sequence having 6 non-white pulses, a hypercube of 6 images is generated. Each image of the hypercube is equated to a model attenuation spectrum as shown in equation 3. Thus, 6 simultaneous equations are generated. For the pulse sequence mentioned above: W, Gi, W, G2, W, G3, W, G4, W, G5, W, Ge, the simultaneous equations are: Measured attenuation spectrum^G^ = pcHb0£Hb0(Gy) + pcHbEHb(Gy) + ^4q + -^1^1 Measured attenuation spectrum(G2) = pCHbo£Hbo^2) + f)CmP'n + Ao + 41G2 Measured attenuation spectrum^G3) = pcHb0£Hb0(G3) + pcHbeHb(G3) + Ao + ArG3 Measured attenuation spectrum(G4) = pcuhoEHhotG^ + pcHb Measured attenuationspectrum^G^ = pcnhoEHho^G^ + pcHbEHb(G5) + Ao + A±G5 Measured attenuation spectrum^G^ = pcnhoEnhotGs) + pcHbEHb(G(,) + Xo + 4XG6 [equations 4] Since the measurements for the pulse sequence were taken in quick succession (and in total over, for example, lOfps), the concentrations of HbO2 and Hb and background coefficients are assumed to have remained constant over that period. Thus, the values of cHb0, cHb, Ao and are treated to be the same for all equations 4 above. The oxygenation determination logic solves the set of n simultaneous equations by determining values for the concentrations of HbO2 and Hb and the background coefficients which cause the measured attenuation spectrum to best fit the model attenuation spectrum. Any suitable mathematical best-fit technique may be used. For example, the squared difference between the measured and modelled attenuation spectra may be minimised. In this way, the oxygenation determination logic algorithmically estimates the concentration of oxyhaemoglobin and the concentration of deoxyhaemoglobin in the imaged tissue from the sensor data imaged by the light sensor. For each pulsed sequence, the oxygenation determination logic estimates a value for the concentration of oxyhaemoglobin and a value for the concentration of deoxyhaemoglobin for each pixel of the imaged tissue. Oxygen saturation, also known as haemoglobin saturation, is the percentage of total haemoglobin binding sites which are occupied with oxygen. At steps 604 / 705, the oxygenation determination logic may determine the haemoglobin / oxygen saturation from the determined concentrations of oxyhaemoglobin and deoxyhaemoglobin. For example, using the following equation: Oxygen saturation (SO2) = —x 100% [equations] cHbo + cHb Thus, the oxygenation determination logic may, for each pulse sequence, determine a haemoglobin / oxygen saturation value for each pixel of the imaged tissue. Total haemoglobin concentration is the sum of the concentrations ( deoxyhaemoglobin. At steps 606 / 706, the oxygenation determination logic may determine the total haemoglobin concentration. For example, using the following equation: Total haemoglobin concentration (HbT) = cHb0 + cHb [equation 6] Thus, the oxygenation determination logic may, for each pulse sequence, determine a total haemoglobin concentration value for each pixel of the imaged tissue. The oxygenation determination logic may output the haemoglobin saturation values and / or the total haemoglobin concentration values for the pixels of the imaged tissue to the image processor 504. At steps 605 and 607, the image processor 504 outputs the haemoglobin saturation values and / or total haemoglobin concentration values for the pixels of the imaged tissue to the surgeon's display 507. To do so, the image processor 504 may generate an image of the haemoglobin saturation for that pulse sequence. Similarly, the image processor 504 may generate an image of the total haemoglobin concentration for that pulse sequence. The image processor 504 may generate these image(s) by assigning a colour mapping to the haemoglobin saturation values and / or total haemoglobin concentration values for each pixel in the image. Thus, an image of the haemoglobin saturation would appear as different colours to represent different saturation levels. For example, the colours may vary from blue to red, with blue representing low saturation levels and red representing high saturation levels. An image of the total haemoglobin concentration would appear as different colours to represent different total haemoglobin concentration. The image processor 504 may output the image(s) of the haemoglobin saturation and / ortotal haemoglobin concentration to the surgeon's display separately to outputting the RGB video feed to the surgeon's display. In this example, the surgeon may be able to view the haemoglobin saturation and / or total haemoglobin concentration images alongside the RGB video feed. The haemoglobin saturation and / ortotal haemoglobin concentration images may overlap the RGB video feed but are not incorporated into the RGB video feed. The image processor 504 may overlay the image(s) of the haemoglobir haemoglobin concentration on the RGB video feed and output the combination for display on the surgeon's display. In this case, the anatomical features of the surgical site and the surgical instruments displayed in the video feed remain visible in the combined feed output to the surgeon's display. This may be achieved by alpha-blending the haemoglobin saturation and / or total haemoglobin concentration with the RGB video feed. Alpha-blending is a known process in which the overlying image (in this case likely the haemoglobin saturation and / or total haemoglobin concentration) is made transparent such that the underlying (in this case RGB) image is visible. With alpha-blending, the colour mapping of each of the haemoglobin saturation and / or total haemoglobin concentration and the RGB feed is not modified. Instead, the transparency level of the overlying image is modified such that the underlying image is visible through the overlying image. Alternatively, the colour mapping of the overlying haemoglobin saturation and / or total haemoglobin concentration and / or the colour mapping of the underlying RGB video feed may be modified such that the anatomical features of the surgical site and the surgical instruments displayed in the video feed remain visible in the combined feed output to the surgeon's display. With this technique, the overlying image need not be made transparent. However, in a further alternative, both the alpha-blending and colour-mapping modification techniques may be combined to generate the images output to the surgeon's display. Thus, the surgeon may continue to use the combined feed to view the surgical site. If desired, the surgeon may continue to perform surgery using the colour-enhanced image of the surgical site on the surgeon's display. The light source may illuminate the tissue with subsequent pulse sequences. These pulse sequences may be contiguous. In the example described above in which the light source pulses at a rate of 60fps, and there are 6 non-white wavelength bands, the pulse sequence lasts lOfps. Thus, new pulse sequences occur at a rate of lOfps. The light sensor receives reflected light back from the tissue, which is fed to the control system 502. The control system may iteratively perform the method described herein to continuously determine concentrations of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in the imaged tissue. Since the measured attenuation spectrum is received by the control system at a rate of master frame rate / 2n (as described with respect to equation 1), the oxygenation determination logic may determine sequential haemoglobin saturation values and sequential total haemoglobin values at a rate of master frame rate / 2n. Thus, the image processor n of the haemoglobin saturation values and / or the total haemoglobin values at a rate of master frame rate / 2n to the surgeon's display. The RGB feed is generated at a rate of master frame rate / 2. Thus, in the example in which the image(s) of the haemoglobin saturation values and / or the total haemoglobin values are overlaid on the RGB feed, the overlay is updated at a rate of 1 / n of the RGB rate. So, in the example above, the RGB feed is updated every 60fps, and the overlay is updated every lOfps. The control system may perform one or more checks prior to outputting the haemoglobin saturation and / or total haemoglobin concentration to the surgeon's display. These checks may include one or more of the following: 1. Comparing the intensity of each pixel or group of pixels in the RGB image of the tissue to a pixel intensity threshold value. If the intensity of each pixel or group of pixels exceeds the pixel intensity threshold value, then the control system outputs the haemoglobin saturation and / or total haemoglobin concentration to the surgeon's display. If the intensity of a pixel or group of pixels does not exceed the pixel intensity threshold value, then the determined values for the concentration of oxyhaemoglobin and deoxyhaemoglobin and background coefficients forthat pixel or group of pixels are discarded. 2. Comparing the light attenuation measurement for each pixel or group of pixels in the imaged tissue to a light attenuation threshold value. If the light attenuation measurement of each pixel or group of pixels exceeds the light attenuation threshold value, then the control system outputs the haemoglobin saturation and / or total haemoglobin concentration to the surgeon's display. If the light attenuation measurement of a pixel or group of pixels does not exceed the light attenuation threshold value, then the determined values for the concentration of oxyhaemoglobin and deoxyhaemoglobin and background coefficients forthat pixel or group of pixels are discarded. 3. Comparing the determined total haemoglobin concentration for each pixel or group of pixels in the imaged tissue to a total haemoglobin concentration threshold value. If the determined total haemoglobin concentration of each pixel or group of pixels exceeds the total haemoglobin concentration threshold value, then the control system outputs the haemoglobin saturation and / or total haemoglob surgeon's display. If the determined total haemoglobin concentration for a pixel or group of pixels does not exceed the total haemoglobin concentration threshold value, then the determined values for the concentration of oxyhaemoglobin and deoxyhaemoglobin and background coefficients for that pixel or group of pixels are discarded. These checks may be performed by the oxygenation determination logic or by the image processor. If, as a result of one or more of these checks, the determined values for the concentration of oxyhaemoglobin and deoxyhaemoglobin and background coefficients for a pixel or group of pixels are discarded, then the discarded values may be replaced as follows. A value for the concentration of oxyhaemoglobin may be interpolated from the determined values for the concentration of oxyhaemoglobin of neighbouring pixels. A value for the concentration of deoxyhaemoglobin may be interpolated from the determined values for the concentration of deoxyhaemoglobin of neighbouring pixels. A value for each of the background coefficients may be interpolated from the determined values for the background coefficients of neighbouring pixels. These interpolated values may then be used in place of the originally determined values in the subsequent calculations, for example of haemoglobin saturation and / or total haemoglobin concentration. Alternatively, one or more of the determined values for the concentration of oxyhaemoglobin, the concentration of deoxyhaemoglobin, and the background coefficients may be replaced with a default replacement value. The default replacement value may correspond to a colour of, for example, black. The method described herein to determine the concentration of oxyhaemoglobin and deoxyhaemoglobin in an imaged tissue is non-invasive. The process is label-free. In other words, no drug or other product is required to be injected or ingested. No radiation imaging such as X-ray is required. In the case of a minimally invasive surgery, the apparatus used can all be incorporated within an endoscope which is already present for the surgery. The methods described herein enable a continuous measure of tis: anatomical tissue to be determined. By repeatedly illuminating the tissue with light from the light source, and continually sensing the light reflected back, the method described herein to determine the concentration of oxyhaemoglobin and deoxyhaemoglobin in the imaged tissue can be iteratively performed. Thus, a continuous measure of the haemoglobin saturation and / or total haemoglobin concentration values can be generated. For example, as described above, these values can be generated at a rate of frame rate / 2n, where n is the number of non-white wavelength bands in the pulse sequence. Thus, the subsequently determined haemoglobin saturation and / or total haemoglobin concentration in the microvasculature can be used as a continuous measurement of tissue oxygenation. This measure is not only continuous, but also quantitative. Unlike the ICG fluorescence method described in the background section which only provides a snapshot in time of the presence of perfusion, the methods described herein enable a quantitative measurement of tissue oxygenation to be continually generated and outputted to the surgeon's display. The haemoglobin saturation and / or total haemoglobin concentration values output to the surgeon's display have high spatial resolution. The combination of the high spatial resolution, and continuous and quantitative measures of oxygenation assist surgeons in their decisionmaking during surgery. For example, in the anastomosis example discussed in the background section, the continuous and quantitative images output to the surgeon's display with high spatial resolution enable the surgeon to see the level of tissue perfusion and identify any leakages with more certainty than the known ICG fluorescence method. The methods described herein are flexible in the sense that the wavelengths used in the pulse sequence can be easily modified by changing the filters used in the light source. No change to the main hardware is required. The applicant hereby discloses in isolation each individual feature described herein and any combination of two or more such features, to the extent that such features or combinations are capable of being carried out based on the present specification as a whole in the light of the common general knowledge of a person skilled in the art, irrespective of whether such features or combinations of features solve any problems disclosed herein, and without limitation to the scope of the claims. The applicant indicates that invention may consist of any such individual feature or combination of features. In view of the foregoing description it will be evident to a person skilled in the art that various modifications may be made within the scope of the invention. 5

Claims

1. A method of determining the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb in an imaged tissue, the method comprising:receiving a measured attenuation spectrum of the imaged tissue, the measured attenuation spectrum comprising n light attenuation measurement sets of the imaged tissue, each light attenuation measurement set at a different one of n wavelength bands, each light attenuation measurement set comprising a plurality of light attenuation measurements, one for each pixel or group of pixels in the image of the tissue;fitting the received measured attenuation spectrum to a model attenuation spectrum, the model attenuation spectrum comprising a linear combination of a haemoglobin spectrum and a linear background spectrum, the linear background spectrum comprises a linear combination of a term that varies as a function of the LO wavelength band of the light attenuation measurement set, and an offset term, theC\J fitting comprising:solving a set of n simultaneous equations, each equation equating a light attenuation measurement set to the model attenuation spectrum for the T“ wavelength band of the light attenuation measurement set, wherein theCMsolving comprises determining, for each pixel or group of pixels in the image of the tissue, values of the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb and background coefficients which causes the measured attenuation spectrum to best fit the model attenuation spectrum.

2. A method as claimed in claim 1, wherein the haemoglobin spectrum comprises a linear combination of an oxyhaemoglobin HbCh term that varies as a function of the wavelength band of the light attenuation measurement set, and a deoxyhaemoglobin Hb term that varies as a function of the wavelength band of the light attenuationmeasurement set.

3. A method as claimed in any preceding claim, wherein the model attenuation spectrum is defined as:A C^i) = PcHbO£HbO&-i) + PcHb£Hb^i) + + ^1^-1Where p is the pathlength of the light travel between emission by a light source and reception by a light sensor, cHb0 is the concentration of oxyhaemoglobin HbO2, cHb is the concentration of deoxyhaemoglobin Hb, £Hb0(Ai) is an extinction spectra of oxyhaemoglobin HbO2 at wavelength Ab is an extinction spectra ofdeoxyhaemoglobin Hb at wavelength A / , j40 and A± are background coefficients of the linear background.

4. A method as claimed in any preceding claim, wherein solving the set of n simultaneous equations comprises determining values for the concentration of oxyhaemoglobin HbO2, the concentration of deoxyhaemoglobin Hb, and the background coefficients LO which causes the measured attenuation spectrum to best fit the model attenuationCXI*■ M spectrum by minimising the squared difference of the measured attenuation spectrumCM and the model attenuation spectrum.1—5. A method as claimed in any preceding claim, further comprising determining haemoglobin saturation from the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb.

6. A method as claimed in claim 5, further comprising outputting the haemoglobin saturation to a surgeon's display.

7. A method as claimed in claim 6, comprising displaying the haemoglobin saturation on the surgeon's display by overlaying the haemoglobin saturation on a video feed from a surgical endoscope also displayed on the surgeon's display.

8. A method as claimed in claim 7, comprising processing the haemoglobin saturation to generate a transparent haemoglobin saturation image, then overlaying the transparent haemoglobin saturation image on the video feed from the surgicalendoscope such that the video feed from the surgical endoscope remains visible onthe surgeon's display.

9. A method as claimed in any preceding claim, further comprising determining totalhaemoglobin concentration from the concentration of oxyhaemoglobin HbCh anddeoxyhaemoglobin Hb.

10. A method as claimed in claim 9, further comprising outputting the total haemoglobinconcentration to a surgeon's display.

11. A method asclaimed in claim 10, comprising displaying the total haemoglobinconcentrationon the surgeon's display by overlaying the total haemoglobinconcentrationon a video feed from a surgical endoscope also displayed on thesurgeon's display.LD12. A method as claimed in claim 11, comprising processing the total haemoglobinconcentration to generate a transparent total haemoglobin concentration image, thenoverlaying the transparent total haemoglobin concentration image on the surgicalendoscope such that the video feed from the surgical endoscope remains visible onthe surgeon's display.

13. A method as claimed in any preceding claim, wherein the n wavelength bands are inthe visible spectrum.

14. A method as claimed in claim 13, wherein the n wavelength bands are within the range520nm to 580nm.

15. A method as claimed in any preceding claim, wherein n >3.

16. A method as claimed in any preceding claim, wherein the n light attenuationmeasurement sets are time multiplexed with respect to each other.

17. A method as claimed in any preceding claim, further comprising iteratively repeating the method of claim 1 so as to continuously determine concentrations of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb in the imaged tissue.

18. A method as claimed in claim 17, further comprising continuously determining haemoglobin saturation and / or total haemoglobin concentration from the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb, and displaying the haemoglobin saturation and / or total haemoglobin concentration on a surgeon's display at a rate of 1 / n the frame rate.

19. A method as claimed in any preceding claim, further comprising: comparing each light attenuation measurement for each pixel or group of pixels in the image of the tissue to a light attenuation threshold value; andif the light attenuation measurement does not exceed the light attenuationLO threshold value, discarding the determined values of the concentration ofCM oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for£\j that pixel or group of pixels.

20. A method as claimed in claim 19, further comprising interpolating values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels from the determined values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels.

21. A method as claimed in any preceding claim, further comprising:receiving the intensity of each pixel or group of pixels in the image of the tissue;comparing the intensity of the pixel or group of pixels in the image of the tissue to a pixel intensity threshold value; andif the intensity of the pixel orgroup of pixels in the image of the tissue does not exceed the pixel intensity threshold value, discarding the determined values of the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels.

22. A method as claimed in claim 21, further comprising interpolating values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for that pixel or group of pixels from the determined values for the concentration of oxyhaemoglobin HbCh and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels.

23. A method as claimed in any preceding claim, further comprising:comparing the determined total haemoglobin concentration of each pixel or group of pixels in the image of the tissue to a total haemoglobin concentration threshold value; andif the determined total haemoglobin concentration of each pixel or group ofpixels in the image of the tissue does not exceed the total haemoglobin concentrationthreshold value, discarding the determined values of the concentration of oxyhaemoglobin HbChand deoxyhaemoglobin Hb and background coefficients forthat pixel or group of pixels.

24. A method as claimed in claim 23, further comprising interpolating values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and backgroundcoefficients for that pixel or group of pixels from the determined values for the concentration of oxyhaemoglobin HbO2 and deoxyhaemoglobin Hb and background coefficients for neighbouring pixels.

Images