Tissue imaging method and system

Abstract

The present invention relates to the technical field of fluorescence analysis, and in particular to a tissue imaging method and system. The method comprises: a fluorescence excitation unit irradiates a target tissue with a single excitation light to form a light spot on the target tissue, wherein the target tissue comprises a tumor tissue and a non-tumor tissue; a signal collection unit collects a fluorescence signal having a waveband of 1100 nm to 1700 nm and filtered by a filter set, and sends same to a computer processing unit, wherein the fluorescence signal is autofluorescence generated after the target tissue is irradiated with the light spot; and the computer processing unit generates a first fluorescence image on the basis of the received fluorescence signal, and obtains, on the basis of the signal intensity of a fluorescence signal at each pixel point position on the first fluorescence image, a second fluorescence image indicating a tumor tissue region and a non-tumor tissue region. The beneficial effects of the present invention are that the signals are not easily interfered by other factors, and the generated images have high signal-to-noise ratio, and have strong specificity and accuracy.

Description

A tissue imaging method and system Technical Field

[0001] This invention relates to the field of fluorescence analysis technology, and more particularly to a tissue imaging method and system. Background Technology

[0002] Malignant tumors (cancer) have become one of the major factors affecting human health and lifespan, and their clinical treatment currently relies heavily on surgical resection. However, due to the lack of real-time, highly accurate, and specific intraoperative imaging technologies, surgeons still face significant challenges in accurately identifying intact tumor boundaries during surgery. Studies have shown that the different optical properties of malignant tumors and normal tissues, such as elastic scattering, stimulated Raman scattering, light reflection, and autofluorescence, can be used to identify malignant tumor tissues. Among these, autofluorescence imaging technology can achieve real-time tumor imaging by differentiating the differences in endogenous fluorescent substances between diseased and normal tissues, aided by endoscopes or wide-field imaging systems. However, existing autofluorescence imaging systems do not have high specificity for identifying malignant tumors, and their signals are easily interfered with by endogenous pigments such as heme.

[0003] Therefore, there is an urgent need for a tissue image processing technology that can clearly display images of non-tumor and tumor tissues, with a high signal-to-noise ratio and good specificity. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a tissue imaging method and system, which solves the technical problems of low fluorescence signal specificity and easy interference from other endogenous pigments in the autofluorescence of tissue cells, as well as the technical problems of low signal-to-noise ratio and easy interference from background noise.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0008] In a first aspect, embodiments of the present invention provide a tissue imaging method, comprising:

[0009] S01. The fluorescence excitation unit uniformly irradiates the target tissue with excitation light, forming a light spot on the target tissue, the light spot covering the target tissue; the target tissue includes tumor tissue and non-tumor tissue; the excitation light is a single excitation light of any wavelength selected from 740nm to 780nm or 800 to 810nm;

[0010] S02, the signal collection unit collects the fluorescence signal in the 1100nm-1700nm band after filtering by the filter group and sends it to the computer processing unit; the fluorescence signal is the autofluorescence generated by the target tissue after being irradiated by the light spot;

[0011] S03. The computer processing unit generates a first fluorescence image based on the received fluorescence signal, and obtains a second fluorescence image indicating the tumor tissue region and the non-tumor tissue region based on the signal intensity of the fluorescence signal at each pixel position in the first fluorescence image; in the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel position in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel position in the tumor tissue region is greater than 1.

[0012] Optionally, the power density of the excitation light is used in the range of 5 mW / cm². 2 ~25mW / cm 2 ;

[0013] The size range of the light spot is 100cm. 2 ~300cm 2 .

[0014] Optionally, S01 further includes:

[0015] The exposure time of the excitation light is adjusted to a range of 200ms to 500ms.

[0016] Optionally, the filter group includes a 900nm long-pass filter, a 950nm long-pass filter, a 1000nm long-pass filter, and an 1100nm long-pass filter;

[0017] In the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is greater than 2.

[0018] Optionally, the wavelength of the excitation light is 760 nm, and the power density is used in the range of 5 mW / cm². 2 ~15mW / cm 2 ;

[0019] The size range of the light spot is 200cm. 2 ~300cm 2 ;

[0020] In the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 4 to 10.

[0021] Optionally, S03 includes:

[0022] S03-1. The computer processing unit generates a first fluorescence image based on the received fluorescence signal;

[0023] S03-2, The computer processing unit compares the signal intensity of the fluorescence signal at each pixel position in the first fluorescence image with a first threshold, and obtains a process image that indicates the tumor tissue area, non-tumor tissue area and background area based on the comparison result;

[0024] S03-3, The computer processing unit enhances the signal of the tumor tissue region and the non-tumor tissue region in the process image, eliminates the signal of the background region, and obtains a second fluorescence image in which the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 6 to 10.

[0025] Optionally, S03-2 includes:

[0026] The first fluorescence image is grayscaled to obtain a grayscale matrix of the first fluorescence image; and a process image indicating the tumor tissue region, non-tumor tissue region and background region is obtained based on the grayscale matrix and a first threshold; the grayscale matrix is ​​a matrix composed of grayscale values ​​obtained after grayscaled processing of each pixel position in the first fluorescence image.

[0027] Optionally, the first threshold is obtained based on the grayscale matrix and Formula 1;

[0028] Formula 1 is:

[0029] Among them, I1~I n I0 is the gray value of all pixels in the gray-level matrix arranged from smallest to largest, and I1 is the gray value of the pixel with the smallest gray value in the gray-level matrix. n I represents the gray value at the pixel location with the highest gray value in the gray-scale matrix. i I is the larger of the two adjacent gray values ​​with the greatest difference. i-1 It is the smaller gray value among the two adjacent gray values ​​with the greatest difference.

[0030] Optionally, the non-tumor tissue includes normal tissue and / or non-malignant lesion tissue.

[0031] In a second aspect, embodiments of the present invention provide a tissue imaging system, comprising:

[0032] The fluorescence excitation unit is used to emit excitation light into the target tissue, forming a light spot on the target tissue;

[0033] A filter array is used to filter the fluorescence signal, which is the autofluorescence produced by lipofuscin in the target tissue after being irradiated by a light spot;

[0034] The signal collection unit is used to collect the filtered fluorescence signal and send it to the computer processing unit;

[0035] A computer processing unit is used to obtain a second fluorescence image that indicates the tumor tissue area and the non-tumor tissue area based on the received fluorescence signal;

[0036] The angle between the optical path of the signal collection unit receiving the fluorescence signal and the optical path of the single excitation light generated by the fluorescence excitation unit is between 0° and 90°, and the computer processing unit is communicatively connected to the signal collection unit.

[0037] When the fluorescence excitation unit emits excitation light toward the target tissue, the system performs the tissue imaging method described above.

[0038] (III) Beneficial Effects

[0039] The beneficial effects of this invention are: The tissue imaging method of this invention uses excitation light with a wavelength range of 740nm–780nm or 800nm–810nm, and sets the power density of the excitation light to 5mW / cm². 2 ~25mW / cm 2 This method excites the target tissue to produce autofluorescence and collects the autofluorescence in the near-infrared II region. Compared with existing technologies, its signal is less susceptible to interference from other endogenous pigments, while reducing background signal interference, resulting in a higher signal-to-noise ratio. Furthermore, the generated image shows a clear signal difference between non-tumor tissue and tumor tissue, with distinct boundaries, thus exhibiting strong specificity and accuracy. Attached Figure Description

[0040] Figure 1 is a flowchart of a tissue imaging method according to Embodiment 1 of the present invention;

[0041] Figure 2, from left to right, shows the bright field image of liver tissue No. 1 according to Embodiment 3 of the present invention, the fluorescence image of liver tissue No. 1 in the visible light region after being excited by excitation light with a wavelength of 475 nm, the fluorescence image of liver tissue No. 1 in the visible light region after being excited by excitation light with a wavelength of 630 nm, the fluorescence image of liver tissue No. 1 in the visible light region after being excited by excitation light with a wavelength of 760 nm, and the pathological H&E section image of liver tissue No. 1.

[0042] Figure 3, from left to right, shows the bright field image of liver tissue No. 2 according to Embodiment 3 of the present invention, the fluorescence image of liver tissue No. 2 in the visible light region after being excited by excitation light with a wavelength of 475 nm, the fluorescence image of liver tissue No. 2 in the visible light region after being excited by excitation light with a wavelength of 630 nm, the fluorescence image of liver tissue No. 2 in the visible light region after being excited by excitation light with a wavelength of 760 nm, and the pathological H&E section image of liver tissue No. 2.

[0043] Figure 4, from left to right, shows the bright field image of liver tissue No. 3 according to Embodiment 3 of the present invention, the fluorescence image of liver tissue No. 3 in the visible light region after being excited by excitation light with a wavelength of 475 nm, the fluorescence image of liver tissue No. 3 in the visible light region after being excited by excitation light with a wavelength of 630 nm, the fluorescence image of liver tissue No. 3 in the visible light region after being excited by excitation light with a wavelength of 760 nm, and the pathological H&E section image of liver tissue No. 3.

[0044] Figure 5 shows the autofluorescence quantitative images of tumor tissue and adjacent normal tissue in the visible light region of 9 human liver tissues according to Example 3 of the present invention.

[0045] Figure 6 shows the autofluorescence quantitative images of tumor tissue and adjacent normal tissue in the near-infrared region 1 in nine human liver tissues according to Example 3 of the present invention.

[0046] Figure 7 shows the autofluorescence quantitative images of tumor tissue and adjacent normal tissue in the near-infrared 2 region in nine human liver tissues according to Example 3 of the present invention.

[0047] Figure 8 shows the ratio of autofluorescence intensity of normal tissue and tumor tissue in the visible light region, near-infrared region I, and near-infrared region II of liver tissue specimens corresponding to 9 patients according to Example 3 of the present invention.

[0048] Figure 9 is a graph showing the change in the ratio of autofluorescence intensity of normal tissue and tumor tissue in the near-infrared II region under different wavelengths of excitation light according to Embodiment 4 of the present invention.

[0049] Figure 10, from left to right, shows a bright field image of a liver tissue from Example 4 of the present invention and near-infrared II autofluorescence images under excitation light of 700 nm, 760 nm, 850 nm and 900 nm.

[0050] Figure 11, from left to right, shows a bright field image of a liver tissue from Example 4 of the present invention and near-infrared II autofluorescence images under excitation light of 700 nm, 760 nm, 850 nm and 900 nm.

[0051] Figure 12 is a graph showing the change in the ratio of autofluorescence intensity of normal tissue and tumor tissue in the near-infrared II region under different filters according to Embodiment 4 of the present invention.

[0052] Figure 13, from left to right, shows a bright field image of a liver tissue from Example 4 of the present invention and an autofluorescence image obtained by collecting fluorescence signals of 850–1700 nm, 1100–1700 nm, and 1300–1700 nm under 760 nm excitation light.

[0053] Figure 14, from left to right, shows a bright field image of a liver tissue from Example 4 of the present invention and an autofluorescence image obtained by collecting fluorescence signals of 850–1700 nm, 1100–1700 nm, and 1300–1700 nm under 760 nm excitation light.

[0054] Figure 15 is a graph showing the change in the ratio of autofluorescence intensity of normal tissue and tumor tissue in the near-infrared II region of human liver tissue under different excitation light power densities according to Embodiment 4 of the present invention.

[0055] Figure 16 is a graph showing the signal-to-noise ratio variation of human liver tissue under different excitation light power densities according to Embodiment 4 of the present invention.

[0056] Figure 17 is a graph showing the change of the ratio of autofluorescence intensity of normal tissue and tumor tissue of human liver tissue in the near-infrared II region with exposure time according to Embodiment 4 of the present invention.

[0057] Figure 18 is a graph showing the change of signal-to-noise ratio of normal and tumor tissues of human liver tissue in the near-infrared II region with exposure time according to Embodiment 4 of the present invention.

[0058] Figure 19 shows the near-infrared 2 region autofluorescence intensity distribution of non-cirrhotic normal liver tissue and well-differentiated hepatocellular carcinoma tissue, moderately differentiated hepatocellular carcinoma tissue, and poorly differentiated hepatocellular carcinoma tissue in 12 specimens according to Example 5 of the present invention.

[0059] Figure 20, from left to right, shows a bright field view, a near-infrared 2-zone autofluorescence view, and a pathological H&E section view of the liver tissue from a typical liver tissue specimen resected from a cirrhotic patient according to Embodiment 5 of the present invention.

[0060] Figure 21 shows the near-infrared 2 region autofluorescence intensity distribution of cirrhotic liver tissue, well-differentiated hepatocellular carcinoma tissue, moderately differentiated hepatocellular carcinoma tissue, and poorly differentiated hepatocellular carcinoma tissue in 18 specimens according to Example 5 of the present invention.

[0061] Figure 22, from left to right, shows the fluorescence microscopy images of a normal liver tissue section in the visible light region, the near-infrared region I, and the near-infrared region II of the present invention, according to Embodiment 6 of the present invention.

[0062] Figure 23, from left to right, shows the fluorescence microscopy of tumor tissue sections in the visible light region, near-infrared region I, and near-infrared region II according to Embodiment 6 of the present invention.

[0063] Figure 24, from left to right, shows a bright field image of the patient's liver tissue during surgery in a living patient according to Embodiment 7 of the present invention, and a view of the patient's liver tissue during surgery in the near-infrared 2 region.

[0064] Figure 25, from left to right, shows a bright field image of the isolated liver tissue according to Embodiment 7 of the present invention and a view of the isolated liver tissue in the near-infrared 2 region.

[0065] Figure 26 shows the ratio of autofluorescence intensity between normal tissue and tumor tissue in Figures 24 and 25.

[0066] Figure 27 shows the H&E slices of the liver tissue from the patients in Figures 24 and 25.

[0067] Figure 28, from left to right, shows a typical colorectal cancer liver metastasis specimen in bright field, an autofluorescence view in near-infrared II region, and a pathological H&E section view of the typical colorectal cancer liver metastasis specimen according to Embodiment 8 of the present invention.

[0068] Figure 29, from left to right, shows a bright field view, an autofluorescence view in the near-infrared II region, and a pathological H&E section view of the typical intrahepatic cholangiocarcinoma specimen according to Embodiment 8 of the present invention.

[0069] Figure 30, from left to right, shows a typical focal nodular hyperplasia tissue in bright field, an autofluorescence view in near-infrared II, and a pathological H&E section view of the typical focal nodular hyperplasia tissue according to Embodiment 8 of the present invention.

[0070] Figure 31, from left to right, shows a bright field view of a typical inflammatory pseudotumor tissue according to Embodiment 8 of the present invention, an autofluorescence view in the near-infrared II region, and a pathological H&E section view of the typical inflammatory pseudotumor tissue.

[0071] Figure 32 is a graph showing the change in sensitivity of liver tissue in the near-infrared II region, near-infrared I region, and visible light region according to Embodiment 8 of the present invention as a function of specificity. Detailed Implementation

[0072] To better explain and facilitate understanding of this invention, the relevant proprietary terms are explained below.

[0073] NCR: The ratio of the fluorescence intensity of adjacent normal tissue to that of autofluorescence; TNR: The ratio of the fluorescence intensity of tumor tissue to that of normal tissue;

[0074] p: represents the p-value in statistics, which is the probability of observing a sample statistic or a more extreme value under the condition that the null hypothesis is true; J: is the Youden index, which is generally used to evaluate the performance of binary classifiers. Its formula is J = true positive rate + true negative rate - 1.

[0075] ROC: Working characteristics, which can be plotted using sensitivity and specificity through Graphpad software or the built-in program of Origin software;

[0076] AUC: Area under the curve, commonly used to measure prediction accuracy;

[0077] Visible light region: The spectral region with wavelengths between approximately 400nm and 700nm that can be perceived by the average human eye; Near-infrared region I refers to the spectral region with wavelengths between approximately 700nm and 900nm; Near-infrared region II refers to the spectral region with wavelengths between approximately 1000nm and 1700nm.

[0078] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0079] Example 1

[0080] This embodiment provides a tissue image processing method based on the analysis of the tailing signal of the autofluorescence of endogenous lipofuscin in the near-infrared II band, as shown in Figure 1, including:

[0081] S01. The fluorescence excitation unit uniformly irradiates the target tissue with a single 760nm excitation light. The single excitation light is a specific wavelength excitation light, forming a 200cm² excitation pattern on the target tissue. 2 ~300cm 2 The light spot completely covers the target tissue; the target tissue includes both tumor and non-tumor tissues, and the power density of a single excitation light ranges from 5 mW / cm². 2 ~25mW / cm 2 The preferred value is 5mW / cm 2 ~15mW / cm 2 The exposure time of the single excitation light was adjusted to 200ms.

[0082] S02, the signal collection unit collects fluorescence signals with a wavelength range of 1100nm to 1700nm after being filtered by the filter group and sends them to the computer processing unit; the fluorescence signal is the autofluorescence generated by the lipofuscin in the target tissue after being irradiated by a spot formed by a single laser.

[0083] S03-1. The computer processing unit generates a first fluorescence image based on the received fluorescence signal;

[0084] S03-2. The first fluorescence image is grayscaled to obtain a grayscale matrix of the first fluorescence image; and a process image indicating the tumor tissue area, non-tumor tissue area and background area is obtained according to the grayscale matrix and a first threshold; the grayscale matrix is ​​a matrix composed of grayscale values ​​obtained after grayscaled processing of each pixel position in the first fluorescence image.

[0085] The first threshold is obtained based on the grayscale matrix and Formula 1;

[0086] Formula 1 is:

[0087] Among them, I1~I n I0 is the gray value of all pixels in the gray-level matrix arranged from smallest to largest, and I1 is the gray value of the pixel with the smallest gray value in the gray-level matrix. n I represents the gray value at the pixel location with the highest gray value in the gray-scale matrix. i I is the larger of the two adjacent gray values ​​with the greatest difference. i-1 It is the smaller gray value among the two adjacent gray values ​​with the greatest difference.

[0088] S03-3, The computer processing unit enhances the signal of the tumor tissue region and the non-tumor tissue region in the process image, and eliminates the signal of the background region, so that the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is greater than 1; preferably, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is greater than 2; preferably, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 4 to 10; preferably, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 6 to 10.

[0089] This invention provides a tissue image processing method based on near-infrared II autofluorescence signals to reduce the interference of endogenous pigments such as blood and bile on tissue autofluorescence signals, thereby improving the signal-to-noise ratio, sensitivity, and specificity of tissue imaging.

[0090] Example 2

[0091] This embodiment provides a tissue image processing system that analyzes the tailing signal of the autofluorescence of endogenous lipofuscin in the near-infrared band, including:

[0092] The fluorescence excitation unit is used to emit a single excitation light of any wavelength in the range of 740nm~780nm or 800nm~810nm to form a light spot on the target tissue, and the light spot completely covers the target tissue; the light beam generated therefrom irradiates the biological tissue and excites the endogenous lipofuscin of the tissue to produce autofluorescence.

[0093] The fluorescence excitation unit includes: an excitation source, a coupler, an optical fiber, and a beam expander module; wherein, the excitation source is used to generate a single excitation light of any wavelength within the range of 740nm~780nm or 800nm~810nm, preferably 760nm; the coupler is used to guide the single excitation light emitted by the excitation source into the optical fiber; the optical fiber is used to focus the single excitation light onto the target tissue; the beam expander module forms a focused spot on the target tissue from the single excitation light focused by the optical fiber; the excitation source is one of a krypton ion laser, a fixed laser, a semiconductor laser, a gas laser, a fiber laser, or a dye laser; the operating power density of the excitation source is 5mW / cm². 2 ~25mW / cm 2 Preferably, it is 5mW / cm 2 ~15mW / cm 2 The beam expander module is a plano-concave lens that transmits more than 90% of near-infrared light; the angle between the beam path generated by the fluorescence excitation unit and the path of the fluorescence emitted by the endogenous lipofuscin entering the signal collection unit is between 0 and 90°, preferably between 0 and 45°.

[0094] A filter group is used to filter fluorescence signals. The wavelength range of the fluorescence signals passing through the filter group is the near-infrared II band wavelength. The fluorescence signals are autofluorescence generated by the target tissue after being irradiated with a single excitation light. The filter group can pass fluorescence in the 850nm to 1700nm band, preferably long-pass filters of 1100nm, 1000nm, 950nm, and 900nm.

[0095] The signal collection unit includes an imaging objective lens, a signal detection subunit, and a signal transmission subunit. The imaging objective lens is used to collect fluorescence signals filtered by the filter group. The signal detection subunit is used to convert the fluorescence signals collected by the imaging objective lens into electrical signals. The signal transmission subunit is used to send the electrical signals to the computer processing unit.

[0096] The signal detection subunit includes an InGaAs detector with an effective wavelength range of 850nm to 1700nm.

[0097] The computer processing unit amplifies the received electrical signal and converts it from analog to digital; it preprocesses the digital signal and reconstructs it into a first fluorescence image; it inputs the first fluorescence image into a pre-set network model, which is a U-shaped network model. This U-shaped network model adopts the U-Net network architecture and introduces a gated function (GFF) module, outputting a probability matrix for each pixel belonging to tissue and background; when the first fluorescence image is input into the pre-set network model, and each time the probability value of a pixel position is output, the tissue loss function is calculated using Formula 2, which is:

[0098] Among them, y j Let j be the true value of the j-th pixel in a target organization with a total of N pixels. Let c be the estimated value of the j-th pixel in a target organization with a total of N pixels, and e be the value of the j-th pixel. -3 w1 and w2 are pre-set weights, and w1+w2=1. D(x) is the distance between the true value of a pixel and its probability value, and P(x) is the probability value output after the pixel's predicted value is processed by the pre-set softmax function.

[0099] The computer processing unit adjusts the parameters in the U-shaped network model in real time according to the loss function.

[0100] Formula 3 is used to determine whether a single pixel is part of an organization. Formula 3 is:

[0101] Where Z represents the result of determining whether a single pixel is part of an organization, 1 indicates organization, 0 indicates background, and z represents the probability value of whether a single pixel in the probability matrix is ​​part of an organization.

[0102] The organization is segmented from the background based on the organization type of each pixel.

[0103] The segmented first fluorescence image is converted to grayscale using Formula 4 to generate a grayscale matrix for each pixel; Formula 4 is as follows:

[0104] Among them, H x H represents the original grayscale value at pixel x, and H is the grayscale value at pixel x after grayscale conversion. max H represents the grayscale value at 98% of the positions of each pixel in the first fluorescence image after grayscale sorting. min The gray value is the gray value at the 2% position after sorting each pixel position in the first fluorescence image by gray level.

[0105] The tissue type is determined based on Formula 5 and the gray values ​​of each pixel in the gray-scale matrix of the first fluorescence image. Formula 5 is:

[0106] Where H is the grayscale value of pixel x after grayscale conversion, A is the abnormal tissue type, B is the normal tissue type, and L1 is the tissue type judgment threshold.

[0107] The tissue type determination threshold L1 is calculated based on the gray values ​​of each pixel in the gray-scale matrix of the first fluorescence image, including:

[0108] The computer processing unit sorts the gray values ​​of each pixel in the gray-scale matrix of the first fluorescence image from smallest to largest, denoted as I1~I n I1 is the gray value of the pixel with the smallest gray value in the gray-level matrix. n This represents the gray value of the pixel with the highest gray value in the gray-scale matrix.

[0109] The computer processing unit calculates the tissue type determination threshold L1 according to a pre-set formula, wherein the formula is:

[0110] Among them, L max I is the difference between the largest adjacent gray values ​​of each pixel in the gray-level matrix after sorting the gray values ​​from smallest to largest. i I is the larger of the two adjacent gray values ​​with the greatest difference. i-1 It is the smaller gray value among the two adjacent gray values ​​with the greatest difference.

[0111] The computer processing unit adds a label to each pixel in the first fluorescence image based on the grayscale matrix, probability matrix, and tissue type of each pixel location, generating a process image. The label includes whether the pixel location belongs to a tissue, and if so, the tissue type and grayscale value of that pixel location.

[0112] Based on the label of each pixel in the process image, the process image is divided into a tumor tissue region, a non-tumor tissue region, and a background region. Signal enhancement processing is performed on the tumor tissue region and the non-tumor tissue region according to the gray value of each pixel, and signal elimination processing is performed on the background region to generate a second fluorescence image with a boundary line for display.

[0113] This embodiment provides an infrared tissue autofluorescence analysis system that uses a single excitation light in the range of 740nm–780nm or 800nm–810nm to uniformly irradiate isolated or in vivo exposed liver tissue, with a laser power density of 5mW / cm². 2 ~25mW / cm 2 The system excites endogenous lipofuscin in the tissue to produce autofluorescence and collects fluorescence in the 1100nm–1700nm wavelength range, ultimately generating a fluorescence image with a clear boundary between cancerous and normal tissue. The generated image shows a bright signal from normal tissue and a weak signal from tumor tissue. This signal difference between normal and tumor tissues effectively delineates the tumor boundary and provides a high signal-to-noise ratio. Furthermore, because lipofuscin is specifically reduced in tumor tissue and its signal is not easily interfered with by other factors, the system exhibits high specificity and accuracy when used for tumor detection.

[0114] Example 3

[0115] This embodiment provides a tissue imaging system, including:

[0116] The fluorescence excitation unit is used to emit a single excitation light with a wavelength range of 740nm~780nm or 800nm~810nm to form a light spot on the target tissue, and the light spot completely covers the target tissue;

[0117] A filter array is used to filter the fluorescence signal. The wavelength range of the fluorescence signal passing through the filter array is the near-infrared II band wavelength. The fluorescence signal is the autofluorescence generated by the target tissue after being irradiated with a single excitation light.

[0118] The signal collection unit is used to collect the fluorescence signal filtered by the filter group in real time and convert the collected fluorescence signal into an electrical signal.

[0119] The computer processing unit is used to generate a first fluorescence image based on the received electrical signal, and to process the first fluorescence image in real time to generate a fluorescence image with a boundary line for display.

[0120] Near-infrared II autofluorescence imaging was performed on liver tissue removed from nine patients without cirrhosis. Imaging parameters: excitation wavelength 760 nm, filter group using a combination of 850 nm, 1000 nm, and 1100 nm long-pass filters, and exposure time 200 ms.

[0121] The signal collection unit is communicatively connected to the computer processing unit through the signal transmission subunit. The computer processing unit includes a signal display unit for displaying processing results and a signal processing unit for signal processing. The computer processing unit is communicatively connected to the background database.

[0122] Simultaneously, a conventional visible / near-infrared I-region imaging system was used, with a silicon charge-coupled device (SCD) detector, to perform autofluorescence imaging of the tissue in the visible and near-infrared I regions. The visible region imaging parameters were set as follows: the excitation light was LED light filtered through a 475±20nm bandpass filter, the emission filter was a 540±20nm bandpass filter, and signals were collected in the 520-560nm range.

[0123] The near-infrared imaging parameters were set as follows: the excitation light was LED light filtered by a 630±20nm bandpass filter, the emission filter was an 800nm ​​longpass filter, and the signal in the 800-900nm range was collected.

[0124] The obtained images were processed using ImageJ software, and pseudocoloring was applied. Autofluorescence quantification was performed on regions of interest (ROIs) in cancerous tissue and adjacent liver tissue images. The imaged tissues were fixed with paraformaldehyde, then dehydrated, encapsulated in paraffin blocks, and sectioned in paraffin. Finally, the sections were stained with hematoxylin and eosin (H&E) using hematoxylin and eosin.

[0125] Three representative hepatocellular carcinoma (HCC) tissues from different individuals were selected and designated as HCC tissue 1, HCC tissue 2, and HCC tissue 3, as shown in Figures 2, 3, and 4. In the visible light region, the autofluorescence of HCC tissues showed significant individual differences; some tissues exhibited stronger autofluorescence than adjacent liver tissues, while others showed weaker autofluorescence, and the fluorescence within the same HCC tissue was not uniform. However, in the near-infrared regions I and II, the autofluorescence of high-, moderate-, and low-differentiated HCC tissues was lower than that of adjacent normal liver tissue, and the fluorescence within the tissues was relatively uniform. In the near-infrared region II, the difference in autofluorescence between normal and HCC tissues was the greatest, with the clearest boundary between them, which matched the boundary obtained from H&E sections.

[0126] Figures 5, 6, and 7 show the quantitative autofluorescence results of nine hepatocellular carcinoma tissues and adjacent normal liver tissues in the visible, near-infrared I, and near-infrared II regions, respectively. The average autofluorescence intensity of the cancerous tissues in the visible region was approximately 1.61 times that of the normal liver tissues. Due to significant inter-individual variability in the autofluorescence intensity of cancerous tissues in the visible region, statistical analysis showed no significant difference in autofluorescence intensity between cancerous and normal tissues. However, in the near-infrared I and II regions, the average autofluorescence intensity of the adjacent normal liver tissues was higher than that of the cancerous tissues, with intensity differences reaching 2.19 times (p = 0.0039) and 8.98 times (p = 0.0039), respectively.

[0127] By analyzing the autofluorescence intensity of paired specimens, the ratio of autofluorescence intensity of normal tissue to tumor tissue in each patient's corresponding specimen, i.e., the contrast ratio, was calculated. As shown in Figure 8, the average contrast ratio (NCR) provided by near-infrared II autofluorescence imaging was 9.24 ± 1.15 (mean ± sem), significantly higher than that provided by visible light region (0.90 ± 0.21, p < 0.0001) and near-infrared I region (2.31 ± 0.37, p < 0.0001) autofluorescence imaging. The contrast ratio of near-infrared II autofluorescence imaging of patient specimens was approximately 7.25 times and 2.47 times higher than that of ICG-based near-infrared I region (tumor / normal tissue fluorescence intensity ratio TNR = 1.12) and II region (TNR = 2.66), respectively. Furthermore, under near-infrared II autofluorescence imaging, the NCR of each specimen exceeded the widely accepted and easily judged contrast threshold for clinical surgeons, which is generally 2.0. In contrast, under visible light autofluorescence imaging, the NCR value of each specimen was below 2.0; while under near-infrared 1 autofluorescence imaging, nearly half of the specimens had an NCR below 2.0.

[0128] This embodiment utilizes lipofuscin as an endogenous marker for tumors, thereby obtaining a highly specific imaging technique for malignant tumors, especially liver cancer. Lipofuscin is a cellular metabolic waste product, typically a mixture of lipids, cross-linked proteins, small amounts of polysaccharides, and heavy metals. Through extensive experiments and research, the inventors of this application have creatively discovered that because lipofuscin is a substance that is difficult to further degrade after being produced by lysosomes, it continuously accumulates in normal cells in the late stages of cell division, such as hepatocytes, cardiomyocytes, and nerve cells, but can be gradually diluted by rapidly dividing cancer cells. Utilizing this characteristic of lipofuscin, this invention selects lipofuscin as a highly specific marker for malignant tumors, especially liver cancer.

[0129] The inventors of this application discovered through experimental research that by utilizing the broad excitation spectrum of lipofuscin and its residual emission in the near-infrared band, especially in the near-infrared II region, interference from most endogenous fluorescent substances in biological tissues, such as NADH, FAD, vitamin A, elastin, and collagen, as well as interference from endogenous pigments, such as bile pigments and heme, can be avoided, resulting in a clearer image. Specifically, biological tissues are excited using long-wavelength monochromatic excitation light, with the excitation wavelength range being 740–780 nm or 800–810 nm, preferably 760 nm. At this excitation wavelength, most endogenous fluorescent substances cannot be excited to fluoresce. Furthermore, the absorption of bile pigments and heme in this wavelength range is significantly weaker than in the visible light region. Therefore, collecting the near-infrared autofluorescence signal of lipofuscin generated at the excitation wavelength can significantly amplify the difference between normal and tumor tissues, generating a clear image with bright signals from normal tissues and weak signals from tumor tissues, i.e., tumor dark imaging. This signal difference between normal and tumor tissues can be used to delineate tumor boundaries and provide a high signal-to-noise ratio. Furthermore, since lipofuscin is specifically reduced in tumor tissue and its signal is not easily interfered with by other factors, the method has high specificity and accuracy when used to detect tumors.

[0130] Example 4

[0131] This embodiment provides an infrared tissue autofluorescence analysis system. Based on Embodiment 3, the monochromatic laser used in the fluorescence excitation unit is replaced with an ultrafast laser system oscillator with a continuous wavelength. The excitation wavelength is switched every 5 nm within the 660 nm-980 nm range, and the light power density is controlled at 5 mW / cm² by adjusting the size of the excitation light spot. 2 Near-infrared II autofluorescence imaging was then performed on five fresh hepatocellular carcinoma tissue specimens from resected patients, placed on a black plate, with an exposure time set to 200 ms. ImageJ software was used to quantify the autofluorescence of the regions of interest (ROIs) in the cancerous tissue and adjacent liver tissue in the images, and the NCR value of each specimen at different excitation wavelengths was calculated.

[0132] Subsequently, seven freshly excised hepatocellular carcinoma tissue specimens were collected and imaged using a near-infrared II autofluorescence imaging system. A 760nm laser was selected as the light source, with power density and exposure time kept constant. Long-pass filters of 850nm, 900nm, 950nm, 1000nm, 1100nm, 1200nm, and 1300nm were used for imaging. ImageJ software was used to perform autofluorescence quantification on the regions of interest (ROIs) in the cancerous tissue and adjacent liver tissue, and the NCR value of each specimen at different excitation wavelengths was calculated.

[0133] Figure 9 shows the variation curves of the average ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region under different excitation wavelengths. The results indicate that the ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region is the highest under 760 nm excitation light, at 6.63 ± 0.21. Furthermore, as the excitation wavelength exceeds 800 nm, the ratio gradually decreases. When the excitation wavelength exceeds 855 nm, the ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region is less than 2, and the technique's ability to distinguish tumors is almost lost.

[0134] Figures 10 and 11 show two representative hepatocellular carcinoma tissues, representing near-infrared II autofluorescence images generated from hepatocellular carcinoma tissues from two different individuals under different excitation light irradiation. When the excitation light is 760 nm, the contrast between normal liver tissue and cancerous tissue is highest, and their boundary is clearest. However, when the excitation light wavelength is switched to 700 nm, 850 nm, or 900 nm, the boundary between normal and cancerous tissue gradually blurs and becomes indistinguishable.

[0135] Figure 12 shows the variation curves of the average ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region under different emission filters. The results indicate that when using a 1100 nm long-pass filter to collect autofluorescence signals in the 1100 nm–1700 nm band, the ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region reaches its maximum, subsequently decreasing with increasing filter wavelength. Figures 13 and 14 show the near-infrared II autofluorescence images generated by two representative hepatocellular carcinoma tissues from two different individuals under 760 nm excitation light. When collecting signals in the 1100 nm–1700 nm band, the contrast between normal hepatocytes and cancerous tissue is highest, and both are most clearly visible. However, when collecting signals in the 850 nm–1700 nm band, the signal contrast between liver tissue and cancerous tissue decreases, and the boundaries of cancerous tissue in the image become blurred. When collecting signals in the 1300nm–1700nm band, the autofluorescence signals of liver tissue and cancerous tissue nearly disappeared, and the boundary between the two could not be distinguished in the image.

[0136] Seven fresh human tumor tissue specimens were placed on black plates and imaged using an infrared tissue autofluorescence analysis system provided in this embodiment. The imaging parameters were as follows: excitation wavelength 760 nm; power density 5 mW / cm². 2 10mW / cm 2 15mW / cm 2 and 25mW / cm 2Exposure times were set to 50ms, 100ms, 200ms, 300ms, 500ms, and 1000ms, respectively. ImageJ software was used to process the obtained autofluorescence images, calculating the ratio of autofluorescence intensity in normal to tumor tissues. Simultaneously, the intensity ratio of autofluorescence in normal tissues to that in the black plate was calculated as the signal-to-noise ratio.

[0137] Figure 15 shows that the average ratio of normal to tumor tissue autofluorescence intensity in the near-infrared II region increases with increasing excitation power density, while the ratio decreases when the power density increases to 15 mW / cm². 2 At this point, the ratio gradually tends to constant. Similarly, the signal-to-noise ratio of the autofluorescence image of the specimen increases with increasing excitation light power density, as shown in Figure 16. When the power density increases to 20 mW / cm², the signal-to-noise ratio increases. 2 Over time, the ratio gradually tends to constant. This is mainly because the signal detected by the camera is limited, and the signal generated by excessively high power density has become saturated. Considering 20mW / cm²... 2 With 15mW / cm 2 At the specified power density, the ratios of autofluorescence intensity in the near-infrared II region of the specimens (normal / tumor tissues) are similar, while at 20 mW / cm²... 2 Higher Joule heating is easily generated under light. Therefore, 15 mW / cm² is selected. 2 This is the optimal excitation light power density.

[0138] Figures 17 and 18 show the curves of the ratio of autofluorescence intensity in the near-infrared II region of normal / tumor tissue and the signal-to-noise ratio (SNR) as a function of exposure time, respectively. The results show that the ratio of autofluorescence intensity in the near-infrared II region of normal / tumor tissue gradually decreases with increasing exposure time. However, the SNR of the autofluorescence images of the specimens shows a trend of first increasing and then decreasing. The SNR reaches its highest value within the exposure time range of 200 ms to 500 ms. Therefore, 200 ms was ultimately selected as the optimal exposure time for near-infrared II autofluorescence imaging of tissues.

[0139] Example 5

[0140] Experiments were conducted on tissue specimens from different types of hepatocellular carcinoma.

[0141] Twelve hepatocellular carcinoma specimens of different types, obtained intraoperatively from 12 patients without cirrhosis (one case of well-differentiated hepatocellular carcinoma, nine cases of moderately differentiated hepatocellular carcinoma, and two cases of poorly differentiated hepatocellular carcinoma), were imaged on black plates using an infrared tissue autofluorescence analysis system provided in this embodiment. The imaging parameters were standardized as follows: excitation wavelength 760 nm, power density 15 mW / cm². 2The exposure time was 200 ms, and a combination of emission filters at 850 nm, 1000 nm, and 1100 nm long-pass filters was used. ImageJ software was used to process the obtained autofluorescence images, and the autofluorescence intensities of normal liver tissue and tumor tissue were calculated.

[0142] Figure 19 shows the near-infrared spectral density (NIRS) autofluorescence intensity distribution of normal tissue and well-differentiated, moderately differentiated, and poorly differentiated hepatocellular carcinoma (HCC) tissues in 12 specimens. The results showed that the NIRS autofluorescence intensities of well-differentiated, moderately differentiated, and poorly differentiated HCC tissues were similar, all relatively weak, and significantly lower than the NIRS autofluorescence intensity of adjacent normal liver tissue (p<0.0001). This significant difference in autofluorescence intensity between tissues can help clinicians identify HCC lesions of different differentiation types from normal liver tissue. Therefore, this method has a good identification effect on HCC tissues of different differentiation types, such as well-differentiated, moderately differentiated, and poorly differentiated HCC tissues.

[0143] Eighteen specimens of different hepatocellular carcinoma types, obtained intraoperatively from 18 patients with cirrhosis (one case of well-differentiated hepatocellular carcinoma, nine cases of moderately differentiated hepatocellular carcinoma, and two cases of poorly differentiated hepatocellular carcinoma), were imaged on black plates using an infrared tissue autofluorescence analysis system provided in this embodiment. The imaging parameters were standardized as follows: excitation wavelength 760 nm, power density 15 mW / cm². 2 The exposure time was 200 ms, and a combination of emission filters at 850 nm, 1000 nm, and 1100 nm long-pass filters was used. ImageJ software was used to process the obtained autofluorescence images, and the autofluorescence intensities of normal liver tissue and tumor tissue were calculated.

[0144] Figure 20 shows bright-field, near-infrared II autofluorescence (NIIR) and H&E section views of a typical hepatocellular carcinoma specimen resected from a patient with cirrhosis. The results indicate that, similar to non-cirrhotic liver tissue, the entire cirrhotic liver tissue exhibits uniform and strong NIIR autofluorescence. However, the hepatocellular carcinoma tissue shows significantly lower autofluorescence than the adjacent liver tissue. This difference clearly delineates the boundary of the cancerous tissue, and this boundary is consistent with the results obtained from H&E sections. Statistical results show that the NIIR autofluorescence intensity of cirrhotic liver tissue in all specimens is significantly higher than that of the corresponding well-differentiated, moderately differentiated, and poorly differentiated hepatocellular carcinoma tissues, as shown in Figure 21. Therefore, this method can also distinguish between cirrhotic liver tissue and hepatocellular carcinoma tissues of different differentiation types, such as well-differentiated, moderately differentiated, and poorly differentiated hepatocellular carcinoma tissues. In practical applications, it can also be used for tumor imaging and resection in patients with cirrhosis.

[0145] Similar to the process described above, this embodiment can also be used to identify liver metastases from colorectal cancer, intrahepatic cholangiocarcinoma, focal nodular hyperplasia, liver abscess, and inflammatory pseudotumors. Typically, in typical liver metastases from colorectal cancer, the entire adjacent liver tissue exhibits uniform and strong autofluorescence, while in metastatic cancer tissue, the autofluorescence is negligible. Similarly, in typical intrahepatic cholangiocarcinoma, the cholangiocarcinoma tissue also exhibits autofluorescence much weaker than that of the adjacent normal liver tissue.

[0146] In clinical practice, some non-malignant lesions, such as focal nodular hyperplasia, liver abscess, and inflammatory pseudotumors, are often difficult to distinguish from liver cancer through preoperative imaging, biopsy, or intraoperative clinical experience. Under near-infrared II autofluorescence imaging, focal nodular hyperplasia and inflammatory pseudotumor tissues exhibit near-infrared II autofluorescence similar to the surrounding normal liver tissue. Therefore, this approach can also differentiate non-malignant lesions, such as focal nodular hyperplasia, liver abscess, and inflammatory pseudotumors, from liver cancer tissue. In practical clinical applications, this can help physicians make more accurate judgments.

[0147] Example 6

[0148] Paraffin sections of liver cancer tissue specimens were dewaxed using xylene and then mounted with coverslips. Microscopic fluorescence imaging of normal liver tissue and cancerous tissue was performed under a 40x objective lens using an upright confocal fluorescence microscope. In the same field of view, the excitation wavelengths used for imaging were 475nm, 630nm, and 760nm, respectively. Switching filters yielded autofluorescence microscopic images of the tissue in the visible light region, near-infrared I, and near-infrared II regions. The images were processed using ImageJ software, and false colors were adjusted.

[0149] Figures 22 and 23 show microscopic fluorescence images of normal and cancerous liver tissue sections under different imaging channels. The results show that in the visible light region, the autofluorescence of normal liver tissue is diffusely distributed throughout the liver parenchyma, while the autofluorescence in the hepatic sinusoids is extremely weak. This fluorescence distribution pattern is consistent with the reported distribution of endogenous fluorescent substances such as NAD(P)H, flavins, and vitamin A. Furthermore, cancerous tissue also exhibits strong visible light autofluorescence, with widespread distribution observed in liver cancer cells. Under this channel, the granular autofluorescence of lipofuscin in liver tissue is difficult to detect due to interference from strong autofluorescence such as flavins. In the near-infrared region, both normal and cancerous liver tissues show diffusely distributed autofluorescence; however, normal liver tissue also exhibits abundant granular autofluorescence of lipofuscin, while cancerous tissue lacks this granular autofluorescence. However, in the near-infrared II region, normal liver tissue exhibits only abundant autofluorescent particles, while cancerous tissue shows no significant autofluorescence. No other endogenous fluorescent signals can be detected in either tissue. This indicates that the autofluorescence signal of lipofuscin in the near-infrared II region can avoid interference from other endogenous fluorescent signals, and its content is high in normal liver tissue but low in cancerous tissue, ultimately leading to high-contrast imaging of liver cancer tissue.

[0150] Example 7

[0151] Application of Near-Infrared II Autofluorescence Imaging Technology in Vial Resection of Patients. Based on preoperative MRI / CT, ​​blood routine tests, and medical history, a volunteer diagnosed with hepatocellular carcinoma was included in the study. The patient underwent preliminary MRI / CT imaging to locate the lesion before surgery. On the day of surgery, a laparotomy was performed, and the liver surface was examined for cancerous tissue using the near-infrared II autofluorescence imaging technology developed in this invention. Based on the clinician's experience and the results of the near-infrared II autofluorescence imaging technology, the cancerous tissue was removed. The completely removed tumor was then imaged using near-infrared II autofluorescence imaging technology. The near-infrared II autofluorescence imaging parameters were: excitation wavelength 760 nm, power density 15 mW / cm². 2 A combination of 850nm, 1000nm, and 1100nm long-pass filters was used. Simultaneously, a color camera was used to collect bright-field information. The obtained autofluorescence images were processed using ImageJ software to calculate the autofluorescence intensity of normal liver tissue and tumor tissue, and to determine their ratio.

[0152] Similar to the results obtained on ex vivo tissues, near-infrared II autofluorescence imaging (NIIR II) also clearly delineated the boundaries of a hepatocellular carcinoma lesion approximately 3 cm in size on a living human liver, as shown in Figure 24. The entire normal liver showed uniform and strong NIIR II autofluorescence except for the cancerous lesion, while the autofluorescence in the cancerous area was negligible. The NIIR II autofluorescence imaging results of the resected liver cancer tissue were consistent with the in vivo imaging results, as shown in Figure 25. By calculating the ROIs at the normal and cancerous tissue sites, the ratio of autofluorescence intensity between normal and tumor tissues in in vivo imaging was 12.17, while the ratio in ex vivo imaging was 10.88, as shown in Figure 26. This further demonstrates that NIIR II autofluorescence imaging can produce comparable results on ex vivo tissues or in living humans. The final pathological section results confirmed that the tissue detected by the imaging technique was hepatocellular carcinoma, as shown in Figure 27.

[0153] Example 8

[0154] Based on Examples 4 and 5, this embodiment standardizes the imaging parameters of the infrared tissue autofluorescence analysis system to: excitation wavelength (760 nm), power density (15 mW / cm²). 2 Exposure time (200ms), emission filter (850nm, 1000nm and 1100nm long-pass filter combination).

[0155] Different types of liver cancer tissue were removed from multiple patients and imaged on an analysis system. These tissues included liver metastases from colorectal cancer, intrahepatic cholangiocarcinoma, breast cancer, gastric adenocarcinoma, gallbladder cancer, and non-malignant tumor lesions. ImageJ software was used to process the obtained autofluorescence images, calculating the autofluorescence intensity of normal liver tissue and tumor tissue, and the ratio of normal to cancerous or diseased tissue autofluorescence intensity.

[0156] Figure 28 shows the bright-field and near-infrared II autofluorescence (NIR II) views and corresponding H&E section views of a typical colorectal cancer liver metastasis specimen. The tumor boundary in this specimen is difficult to distinguish with the naked eye, but NIR II autofluorescence imaging clearly depicts the boundary between the colorectal cancer liver metastasis tissue and the adjacent liver tissue. The entire adjacent liver tissue shows uniform and strong autofluorescence, but the autofluorescence in the metastatic cancer tissue is negligible. Figure 29 shows the bright-field and near-infrared II autofluorescence (NIR II) views and corresponding H&E section views of a typical intrahepatic cholangiocarcinoma specimen. The results show that the cholangiocarcinoma tissue also exhibits autofluorescence much weaker than the adjacent normal liver tissue. The tumor boundary depicted by NIR II autofluorescence imaging matches the H&E section results well. Clinically, some non-malignant lesions, such as focal nodular hyperplasia, liver abscess, and inflammatory pseudotumors, are often difficult to distinguish from liver cancer through preoperative imaging, biopsy, or intraoperative clinical experience. Figures 30 and 31 show the bright-field and near-infrared II autofluorescence (NIR) views and corresponding H&E slice views of typical focal nodular hyperplasia and inflammatory pseudotumors. Under NIR II autofluorescence imaging, focal nodular hyperplasia and inflammatory pseudotumors exhibit NIR II autofluorescence similar to the surrounding normal liver tissue, but significantly different from the signal of hepatocellular carcinoma tissue. This technique can distinguish malignant tumor lesions with low autofluorescence signals from bright normal liver tissue and non-malignant lesions, thereby guiding accurate resection of malignant tumors.

[0157] In summary, the infrared tissue autofluorescence analysis system based on the above imaging parameters can accurately distinguish between malignant and non-malignant liver tumor tissues (non-malignant tumor tissues include normal tissue, hepatitis, fatty liver, focal nodular hyperplasia, inflammatory pseudotumor, and liver abscess, etc.) and delineate clear boundaries to assist in the detection of malignant liver tumors.

[0158] Furthermore, combining the pathologically validated autofluorescence quantitative results of ROIs (Representative Areas of Infrared Imaging) in the near-infrared II, I, and visible light regions from 30 patients with hepatocellular carcinoma, 10 patients with liver metastases, 6 patients with intrahepatic cholangiocarcinoma, and 9 patients with non-malignant lesions, probability distributions were calculated and ROC curves for different imaging regions were plotted. ROC curves were obtained by varying the intensity threshold and calculating the sensitivity and specificity for each threshold. For near-infrared II autofluorescence imaging, sensitivity = (number of samples with autofluorescence intensity below the threshold in the tumor region II) / (total number of samples). Specificity (1 - false positive rate) = (number of non-cancerous tissue samples with autofluorescence intensity above the threshold) / (total number of non-cancerous tissue samples).

[0159] Figure 32 shows the ROC curves of autofluorescence imaging techniques in the near-infrared II, I, and visible light regions for diagnosing various tumors. The results show that the accuracy (area under the curve, AUC) of near-infrared II autofluorescence imaging for diagnosing malignant tumors is 99.8%, sensitivity is 97.8%, and specificity is 98.4%. The accuracy, sensitivity, and specificity of near-infrared I autofluorescence imaging are 80.3%, 82.6%, and 67.2%, respectively, while those of visible light autofluorescence imaging are 68.2%, 71.7%, and 65.6%, respectively. Generally, the area under the ROC curve (AUC) is between 1.0 and 0.5; the closer to 1, the better the diagnostic effect. AUCs between 0.5 and 0.7 have lower accuracy, AUCs between 0.7 and 0.9 have some accuracy, and AUCs above 0.9 have high accuracy. Therefore, it can be said that the diagnostic effect of near-infrared II autofluorescence imaging technology is significantly better than that of conventional visible light region (excitation wavelength: 455-495nm, emission wavelength range: 520-560nm) and near-infrared I region autofluorescence (excitation wavelength: 660-650nm, emission wavelength range: 800-900nm) imaging technology.

[0160] It's important to clarify that cirrhosis and liver cancer are different conditions, and their affected areas also differ. Cirrhosis typically involves the entire liver tissue becoming cirrhotic, while a liver cancer is only a lesion within the liver. This malignant tumor may be located in one or more parts of the normal liver tissue, or it may be located in a specific area within the cirrhotic tissue. Currently, there are difficulties in distinguishing between cirrhosis and liver cancer. However, during treatment, doctors often need to precisely remove the malignant tumor lesion while preserving as much normal or cirrhotic liver tissue as possible.

[0161] According to the conclusions in Example 4, under excitation light of 760 nm, the average ratio of normal to tumor tissue autofluorescence intensity in the near-infrared region II was the highest. Furthermore, when the excitation light wavelength exceeded 800 nm, this ratio gradually decreased. When the excitation light wavelength exceeded 855 nm, the ratio of normal to tumor tissue autofluorescence intensity in the near-infrared region II was less than 2, indicating that the ability to distinguish tumors was almost nonexistent. Additionally, when collecting signals in the 1100 nm–1700 nm band, the contrast between normal hepatocytes and cancerous tissue in the image was the highest, and both were the clearest. However, when collecting signals in the 850 nm–1700 nm band, the signal contrast between liver tissue and cancerous tissue decreased, and the boundary of cancerous tissue in the image became blurred. When collecting signals in the 1300 nm–1700 nm band, the autofluorescence signals of liver tissue and cancerous tissue almost disappeared, making it impossible to distinguish their boundary in the image.

[0162] In-depth research revealed that when analyzing hepatocellular carcinoma specimens from cirrhotic patients using excitation light of 740nm–780nm or 800nm–810nm and collecting autofluorescence at 1100nm–1300nm, similar to non-cirrhotic liver tissue, the entire cirrhotic liver tissue exhibited uniform and strong near-infrared II autofluorescence, as shown in Figure 20. In contrast, hepatocellular carcinoma tissue showed significantly lower autofluorescence than adjacent liver tissue. This difference also clearly delineated the boundaries of the cancerous tissue; that is, under excitation light of any wavelength within the 740nm–780nm or 800nm–810nm range, the autofluorescence intensity of cirrhotic tissue in the near-infrared II region was significantly higher than that of the corresponding well-differentiated, moderately differentiated, and poorly differentiated hepatocellular carcinoma tissues.

[0163] As can be seen, this method can distinguish between cirrhotic liver tissue and malignant liver tumor tissue; however, malignant liver tumors do not necessarily develop from cirrhosis, but may also be metastasized from cancers in other locations.

[0164] Therefore, in-depth research is needed on the correlation between cholangiocarcinoma, metastatic malignant tumors, and non-malignant lesions in the liver. Based on the above research (i.e., Figures 28, 29, 30, and 31 in Example 8), it was found that using excitation light of 740nm–780nm or 800nm–810nm and collecting autofluorescence at 1100nm–1300nm can not only distinguish between cirrhotic tissue and malignant liver tumor tissue, but also distinguish between malignant tumor tissue and non-malignant tumor tissue from cancer metastases in other locations (non-malignant tumor tissue here includes normal tissue, as well as lesions such as hepatitis, fatty liver, focal nodular hyperplasia, inflammatory pseudotumor, and liver abscess). Furthermore, as shown in Figure 32, using the excitation light in this study and collecting autofluorescence in the 1100nm–1300nm band significantly improves the accuracy, sensitivity, and specificity of detecting malignant tumors in different types of patients compared to techniques based on other excitation light and detection ranges (including the visible and near-infrared regions).

[0165] In summary, the infrared tissue autofluorescence analysis system provided in this embodiment, after irradiating liver tissue with excitation light of any wavelength between 740nm and 780nm or between 800nm ​​and 810nm, collects the generated autofluorescence and images it in the near-infrared II region. The imaging results show a clear boundary between non-malignant tumor tissue and malignant tumor tissue, and can distinguish between liver malignant tumor tissue (here, liver malignant tumor tissue includes malignant tumor tissue developed from cirrhosis, liver malignant tumor tissue unrelated to cirrhosis, and malignant tumor tissue metastasized from other locations) and lesion tissue (here, lesion tissue includes cirrhotic tissue, hepatitis, fatty liver, focal nodular hyperplasia, inflammatory pseudotumor, and liver abscess, etc.) and delineate a clear boundary to assist in the detection of liver malignant tumors.

Claims

1. A tissue imaging method, characterized in that, include: S01. The fluorescence excitation unit uniformly irradiates the target tissue with excitation light, forming a light spot on the target tissue, the light spot covering the target tissue; the target tissue includes tumor tissue and non-tumor tissue; the excitation light is a single excitation light of any wavelength selected from 740nm to 780nm or 800nm ​​to 810nm; the power density of the excitation light is used in the range of 5mW / cm². 2 ~25mW / cm 2 The size of the light spot ranges from 100cm. 2 ~300cm 2 ; S02, the signal collection unit collects the fluorescence signal in the 1100nm-1700nm band after filtering by the filter group and sends it to the computer processing unit; the fluorescence signal is the autofluorescence generated by the target tissue after being irradiated by the light spot; S03. The computer processing unit generates a first fluorescence image based on the received fluorescence signal, and obtains a second fluorescence image that indicates the tumor tissue area and non-tumor tissue area based on the signal intensity of the fluorescence signal at each pixel position in the first fluorescence image. In the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is greater than 1.

2. The tissue imaging method according to claim 1, characterized in that, S01 further includes: The exposure time of the excitation light is adjusted to a range of 200ms to 500ms.

3. The tissue imaging method according to claim 1, characterized in that, The filter group includes a 900nm long-pass filter, a 950nm long-pass filter, a 1000nm long-pass filter, and an 1100nm long-pass filter; In the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is greater than 2.

4. The tissue imaging method according to claim 1, characterized in that, The excitation light has a wavelength of 760 nm and a power density range of 5 mW / cm². 2 ~15mW / cm 2 ; The size range of the light spot is 200cm. 2 ~300cm 2 ; In the second fluorescence image, the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 4 to 10.

5. The tissue imaging method according to claim 1, characterized in that, S03 includes: S03-1. The computer processing unit generates a first fluorescence image based on the received fluorescence signal; S03-2, The computer processing unit compares the signal intensity of the fluorescence signal at each pixel position in the first fluorescence image with a first threshold, and obtains a process image that indicates the tumor tissue area, non-tumor tissue area and background area based on the comparison result; S03-3, The computer processing unit enhances the signal of the tumor tissue region and the non-tumor tissue region in the process image, eliminates the signal of the background region, and obtains a second fluorescence image in which the ratio of the signal intensity of the fluorescence signal at any pixel location in the non-tumor tissue region to the signal intensity of the fluorescence signal at any pixel location in the tumor tissue region is 6 to 10.

6. The tissue imaging method according to claim 5, characterized in that, S03-2 includes: The first fluorescence image is grayscaled to obtain a grayscale matrix of the first fluorescence image; and a process image indicating the tumor tissue region, non-tumor tissue region and background region is obtained based on the grayscale matrix and a first threshold; the grayscale matrix is ​​a matrix composed of grayscale values ​​obtained after grayscaled processing of each pixel position in the first fluorescence image.

7. The tissue imaging method according to claim 6, characterized in that, The first threshold is obtained based on the grayscale matrix and Formula 1; Formula 1 is: Among them, I1~I n I0 is the gray value of all pixels in the gray-level matrix arranged from smallest to largest, and I1 is the gray value of the pixel with the smallest gray value in the gray-level matrix. n I represents the gray value at the pixel location with the highest gray value in the gray-scale matrix. i I is the larger of the two adjacent gray values ​​with the greatest difference. i-1 It is the smaller gray value among the two adjacent gray values ​​with the greatest difference.

8. The tissue imaging method according to any one of claims 1 to 7, characterized in that, The non-tumor tissues include normal tissues and / or non-malignant lesion tissues.

9. A tissue imaging system, characterized in that, include: The fluorescence excitation unit is used to emit excitation light into the target tissue, forming a light spot on the target tissue; A filter array is used to filter the fluorescence signal, which is the autofluorescence produced by lipofuscin in the target tissue after being irradiated by a light spot; The signal collection unit is used to collect the filtered fluorescence signal and send it to the computer processing unit; A computer processing unit is used to obtain a second fluorescence image that indicates the tumor tissue area and the non-tumor tissue area based on the received fluorescence signal; The angle between the optical path of the signal collection unit receiving the fluorescence signal and the optical path of the single excitation light generated by the fluorescence excitation unit is between 0° and 90°, and the computer processing unit is communicatively connected to the signal collection unit. When the fluorescence excitation unit emits excitation light toward the target tissue, the system performs the tissue imaging method according to any one of claims 1 to 8.

Images