CN111870231A - Endoscopic tumor blood vessel normalization detection system and detection method - Google Patents
Endoscopic tumor blood vessel normalization detection system and detection method Download PDFInfo
- Publication number
- CN111870231A CN111870231A CN202010686934.0A CN202010686934A CN111870231A CN 111870231 A CN111870231 A CN 111870231A CN 202010686934 A CN202010686934 A CN 202010686934A CN 111870231 A CN111870231 A CN 111870231A
- Authority
- CN
- China
- Prior art keywords
- blood vessel
- tumor
- image
- blood
- vessel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0071—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by measuring fluorescence emission
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0075—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by spectroscopy, i.e. measuring spectra, e.g. Raman spectroscopy, infrared absorption spectroscopy
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0077—Devices for viewing the surface of the body, e.g. camera, magnifying lens
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/02007—Evaluating blood vessel condition, e.g. elasticity, compliance
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/026—Measuring blood flow
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/026—Measuring blood flow
- A61B5/0285—Measuring or recording phase velocity of blood waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Abstract
The invention discloses an endoscopic tumor vessel normalization detection system and a detection method. The method comprises the following steps: firstly, measuring the blood flow velocity of the tumor and the blood vessel shape; then, measuring the density of tumor blood vessels, the node of the blood vessel branch and the degree of blood vessel distortion; and finally, integrating the tumor blood flow velocity, the blood vessel density and the distortion degree through an artificial intelligence analysis module, and judging whether the tumor blood vessels are normalized or not. The invention has the advantages of high space-time resolution, no wound, no radiation, intelligent diagnosis and the like, and guides the application of the antitumor drug in the window period of normalization of blood vessels by monitoring whether the tumor blood vessels are normalized.
Description
Technical Field
The invention relates to the technical field of medical instruments, in particular to an endoscopic tumor blood vessel normalization detection system and a detection method.
Background
Cancer is a serious disease that endangers human life and health, taking more than 800 million people worldwide every year. The world health organization published 2019 in 9 months reports that men 1/5 and women 1/6 worldwide will have cancer during life, and men 1/8 and women 1/11 will die of cancer. Cancer is expected to become the leading cause of death in every country in the 21 st century, and due to the stroke and the relative decline in the mortality rate of coronary heart disease in countries with higher levels of economic development, cancer will become the major and only disease that hinders the increase in life expectancy. According to the latest statistical data, the death of malignant tumor accounts for 23.91% of all the causes of death of Chinese residents, the incidence rate of cancer is on the trend of rising year by year, the cervical cancer, nasopharyngeal carcinoma, intestinal cancer, gastric cancer, esophageal cancer and the like are also rising continuously, and the medical cost caused by the malignant tumor exceeds 2200 hundred million every year.
Traditional cancer treatment modalities, including surgery, radiation therapy, chemotherapy, and targeted therapies, all have different limitations. The operation is the first treatment mode for the patient with tumor which is not transferred, the tumor tissue which is seen by naked eyes can be directly removed, but the operation treatment for the patient with the transfer is not good. The radiotherapy and chemotherapy can kill tumor cells in large area and multiple points, but can also damage normal tissue cells at the same time, and generate adverse reactions with different degrees. The targeted medicine has obvious treatment effect and quick response, but the tumor cells quickly generate drug resistance.
The occurrence, development, invasion and metastasis of tumors are extremely complex processes in which angiogenesis plays a crucial role. In 1971, professor Folkman J first suggested that tumor growth and metastasis were vascular-dependent, and blocking tumor angiogenesis was an effective strategy to halt tumor growth. The theory of anti-angiogenesis therapy for malignant tumor is that the nutrition supply of tumor is cut off by inhibiting tumor neovascularization, thereby achieving the purpose of inhibiting tumor growth. Research on the theoretical basis enables anti-angiogenesis drugs to enter the clinic and achieve curative effects. However, in recent years, a variety of preclinical and clinical studies show that the anti-tumor angiogenesis drug can achieve a therapeutic effect when applied for a short time, and can cause tumor vascular necrosis after long-term use, so as to cause hypoxia and acidic microenvironment, but cause drug resistance when solid tumor cells have reduced sensitivity to radiotherapy and chemotherapy, and even after anti-vascular therapy for several months, patients can have a relapse phenomenon.
Research shows that by reasonably applying the anti-angiogenesis medicine, abnormal tumor vascular system can be repaired before blood vessels are regressed, so that the tumor blood vessels tend to be normal, and the medicine is more effectively transported to tumor cells, thereby improving the sensitivity of tumor tissues to radiotherapy and chemotherapy. The process of normalization of the microvessels is transient, and after transient normalization, the tumor vessels will gradually recover the disordered structure. If a balance is found between tumor angiogenesis and excessive vascular regression and an optimal administration time is available, the hypothesis of a "time window" is provided, i.e., the optimal balance between tumor angiogenesis and excessive vascular regression is reached within the time window, and the tumor blood vessels tend to normalize. The anti-tumor effect can be more effectively exerted by combining radiotherapy and chemotherapy in the time window.
However, different classes of anti-angiogenic drugs act on different types of tumors, often resulting in different ranges of "normalized time windows of blood vessels". Therefore, how to determine the specific time and range of the "window" is the key to further optimize the combination of radiotherapy, chemotherapy and immunotherapy with anti-tumor angiogenesis therapy mode and to exert the optimal synergistic effect.
To date, the "gold standard" for assessing tumor angiogenesis status and normalization of tumor blood vessels remains a histopathological index such as microvessel density, basal membrane thickness, pericyte coverage, and the like. Normalized tumor vessels show tight vascular endothelial cell junctions, increased pericyte coverage, reduced leakage, reduced vascular branching, and relatively normal basement membrane. However, the above-mentioned indexes have the disadvantages of invasiveness, over-dependence on definite materials and one-sidedness, so that the above-mentioned indexes are mainly used for the basic research of exploring the mechanism of tumor angiogenesis and tumor vessel normalization at present, but obviously cannot meet the requirement of clinical evaluation on tumor vessel normalization. Compared with the prior art, the imaging examination has the characteristics of non-invasiveness and comprehensiveness, and has obvious advantages in the aspect of clinical evaluation of tumor vessel normalization.
At present, biomedical imaging methods developed based on different physical effects have different drawbacks in the detection of normalization of tumor vessels. Magnetic Resonance Imaging (MRI) is costly, time consuming, lengthy, a wide variety of imaging equipment and contrast agents, and has stringent requirements for image acquisition and image post-processing techniques. The risks of radiation and contrast agent allergies and kidney injury present in X-ray Computed Tomography (CT) imaging remain difficult to overcome. Nuclear medicine imaging techniques such as SPECT/CT, PET/CT, etc. have limited their clinical application due to high requirements for imaging equipment, technology, and high examination costs. In addition, the accuracy of ultrasonic diagnosis depends on the performance of the ultrasonic equipment and the clinical experience and technical level of the examining physician, which greatly limits the wide application of the ultrasonic equipment. Therefore, the detection method for accurately judging the normalization window period of the tumor blood vessels in a noninvasive, non-radiative and intelligent manner has good application prospect.
Disclosure of Invention
The invention aims to solve the technical problem of providing an endoscopic tumor vessel normalization detection system and method aiming at the defects in the prior art.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the invention provides an endoscopic tumor blood vessel normalization detection system, which comprises:
the multi-mode light source module is used for illuminating the biological tissue to generate a white light image, a special spectrum image and a speckle image;
the endoscopic probe is used for extending into a human body cavity to illuminate and receive image signals;
the image processor is connected with the endoscopic probe and is used for carrying out image processing operations including demosaicing, denoising, defogging and color enhancement on the white light image acquired by the endoscopic probe;
the blood flow detection and analysis module is connected with the endoscopic probe and is used for processing the speckle images acquired by the endoscopic probe through an algorithm to obtain a blood flow velocity image;
the blood vessel morphology analysis module is connected with the blood flow detection analysis module and is used for extracting the tumor blood vessel morphology structure and measuring the tumor blood vessel density and the blood vessel distortion degree;
the artificial intelligence analysis module is connected with the blood vessel morphology analysis module and used for integrating the blood flow velocity, the blood vessel density and the distortion degree of the tumor blood, judging whether the tumor blood vessels are normalized or not and giving a prediction result of the normalization window period of the tumor blood vessels;
and the display module is connected with the output ends of the image processor and the artificial intelligence analysis module and is used for displaying images and prompting the judgment result of whether the tumor blood vessels are normalized.
Further, the multi-modal light source of the present invention comprises: a near-infrared laser light source and an LED light source; wherein the central wavelength range of the near-infrared laser light source is 660 nm-1380 nm; the spectrums of the LED light sources are combined into a white light spectrum, and the LED light sources at least comprise UV light sources with the wavelengths of 400 nm-430 nm.
Furthermore, the endoscopic probe comprises a light transmitting optical fiber bundle, an image sensor, a biopsy forceps channel and a nozzle which are connected with each other; the display module comprises a display and a voice prompter.
Furthermore, the hardware of the blood flow detection and analysis module is composed of a Field Programmable Gate Array (FPGA), a proprietary Asic chip or a high-performance display card; the hardware of the artificial intelligence analysis module is formed into a proprietary Asic chip or a high-performance display card.
The invention provides an endoscopic tumor blood vessel normalization detection method, which comprises the following steps:
and 5, integrating the blood flow speed, the blood vessel density and the distortion degree of the tumor by an artificial intelligence analysis module, judging whether the tumor blood vessels are normalized or not, and giving a voice prompt of a tumor blood vessel normalization window period prediction result.
Further, the specific method for acquiring the tumor blood flow image in the step 1 of the present invention is as follows:
with Ns*NsThe window filter processes the original speckle image to obtain a normalized variance image Vns(x, y), the processing formula is:
wherein N issThe size of the window filter is 5-11, Is(i, j) is the gray value of the pixel in the window filter, i and j are respectively the horizontal and vertical coordinates of the pixel in the window filter,
the mean value of the pixel gray level in the window filter is used, and x and y are the horizontal and vertical coordinates of the newly generated pixel points of the normalized variance image;
calculating a two-dimensional blood flow image of the tumor, wherein the formula is as follows:
wherein, v (x, y) is the blood flow image.
Further, the specific method for detecting and detecting the tumor vascular structure in step 2 of the present invention is as follows:
(a) respectively filtering the blood flow image by using 12 two-dimensional Gabor filters in different directions to obtain a filtered blood flow image Bλ,,θ(x, y), the processing formula is:
Bλ,,θ(x,y)＝∫∫f(u,v)gλ,,θ(x-u,y-v)dudv
wherein f (u, v) is a blood flow image, gλ,,θ(x-u, y-v) is a two-dimensional Gabor filter, the parameter lambda is the reciprocal of the spatial frequency, the parameter is the variance characteristic of the Gabor filter, the value range of the lambda is 0.1-10, the value interval is 0.1, and is lambda/2; the parameter theta is the filtering direction of the Gabor filter, the blood vessel detection in different directions is realized by changing the parameter theta, and the value of theta is N pi/12, wherein N is 0,1, … and 12;
(b) extraction of Bλ,,θMaximum response values max (B) in 12 different directions for each pixel in (x, y)λ,,θ(x, y)) are combined into a new image, i.e. the image of the vessel structure H (x, y).
Further, the specific method of step 3 of the present invention is:
the specific method for extracting the blood vessel density comprises the following steps:
(a) carrying out binarization threshold operation on the blood vessel structure image H (x, y), resetting a pixel value with a pixel gray value larger than 0 in the blood vessel structure image to be 1, and resetting a pixel gray value smaller than or equal to 0 to be 0; a new blood vessel structure image H' (x, y) is obtained again;
(b) counting all pixels with pixel value 1 in H' (x, y), and dividing by the total pixels in the blood vessel structure image to obtain blood vessel density;
the method for calculating the number of the blood vessel branch nodes comprises the following steps:
(a) extracting a new blood vessel central skeleton image of the blood vessel structure image H' (x, y) through a conventional window filter function such as a two-dimensional Gabor filter or a Gaussian filter;
(b) analyzing the communication relationship between each pixel point and a neighborhood pixel point on the blood vessel central skeleton image, and recording a certain blood vessel central skeleton pixel point as a blood vessel branch node when the communication number of the certain blood vessel central skeleton pixel point and the neighborhood blood vessel central skeleton pixel point is more than 3;
(c) traversing the whole blood vessel central skeleton image, obtaining all blood vessel branch nodes in the image, and calculating the total number of the blood vessel branch nodes.
Further, the specific method for calculating the degree of blood vessel distortion in step 4 of the present invention is as follows:
(a) dividing the blood vessel skeleton image into a plurality of sections according to branch nodes; calculating the curvature of each point of each section of the vascular skeleton by using the following formula;
wherein, K (x, y) represents curvature, and x, y represent horizontal and vertical coordinates of each point respectively;
(b) and taking an absolute value of the curvature of each point of a certain section of the blood vessel skeleton, and then carrying out superposition averaging to obtain the distortion degree of the section of the blood vessel.
Further, the method for determining whether tumor blood vessels are normalized and the prediction result of the window period of normalization of blood vessels in step 5 of the present invention comprises:
(a) collecting 1000 tumor speckle images by using a tumor blood vessel normalization detection system, and extracting blood vessel morphological function information of blood flow velocity, blood vessel density, blood vessel diameter and blood vessel distortion degree;
(b) collecting animal tumor tissue samples, and judging whether tumor blood vessels are normalized or not by an immunofluorescence or immunohistochemical method;
(c) establishing a corresponding relation database of blood flow velocity, blood vessel density, blood vessel diameter, blood vessel distortion degree and tumor blood vessel normalization, wherein the training data are blood flow velocity, blood vessel density, blood vessel diameter and blood vessel distortion degree, and the training label is tumor blood vessel normalization or tumor blood vessel non-normalization;
(d) training a deep neural network model by using the database, wherein the deep neural network model comprises VGG16, GooleNet and ResNet deep neural network models;
(e) and (4) predicting a new tumor blood flow image by using the training result, and giving a prompt for whether tumor blood vessels are normalized.
The invention has the following beneficial effects:
the invention is based on the laser speckle blood flow imaging principle and an artificial intelligence analysis system, and can be used for carrying out functional and structural imaging on blood vessels in tumor tissues and further automatically judging whether the tumor blood vessels are normalized. Compared with other existing tumor blood vessel imaging methods, the endoscopic tumor blood vessel normalization detection system provided by the invention has the advantages that:
(1) the adopted blood flow image acquisition method is based on the laser speckle optical imaging principle, can obtain a blood flow perfusion image with high spatial resolution in a non-invasive and non-radioactive manner, and is convenient for extracting characteristic indexes related to tumor blood vessel normalization, such as tumor blood flow velocity, blood vessel density, blood vessel distortion degree and the like. Compared with color Doppler ultrasound, the color Doppler ultrasound has obvious technical advantages.
(2) The invention introduces a convolutional neural network into a tumor blood vessel normalization analysis system, can extract high-dimensional characteristic information which is difficult to obtain by human eyes, and helps to judge and predict the tumor blood vessel normalization window period more accurately.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a schematic diagram of the results of a system according to an embodiment of the present invention;
FIG. 2 is a flow chart of a processing method of an embodiment of the invention;
FIG. 3 is a flowchart of a method for determining normalization of tumor blood vessels according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, the endoscopic tumor blood vessel normalization detection system according to the embodiment of the present invention comprises a multi-modal light source module, an endoscopic probe, an image processor, a blood flow detection and analysis module, a blood vessel morphology analysis module, an artificial intelligence analysis module, and a display module.
The multimode light source consists of near-infrared laser with the central wavelength of 785nm, a red light LED with the central wavelength of 660nm, a green light LED with the central wavelength of 540nm, a blue light LED with the central wavelength of 460nm and a UV light LED with the central wavelength of 410nm, and the five beams of light can be combined through a dichroic mirror or a spectroscope and guided into the endoscopic probe. The multi-modal light source may provide individual laser illumination, but may also provide white light illumination or special spectrum illumination.
The endoscopic probe consists of a light transmitting fiber bundle, an image sensor, a forceps opening, a nozzle and the like. One end of the light transmitting optical fiber bundle is connected with the multi-mode light source and used for guiding the illumination light source into the human body cavity for illumination. The image sensor at the head end of the endoscopic probe is used for receiving image signals and transmitting the image signals to the image processor or the blood flow detection and analysis module through signal lines. If white light illumination or special light illumination is adopted, the image sensor transmits the collected white light image or special light image to the image processor to perform a series of image processing operations such as demosaicing, denoising, defogging, color enhancement and the like; if the laser illumination is adopted, the image sensor transmits the collected speckle images to the blood flow detection and analysis module for blood flow image reconstruction processing.
Then, the blood vessel morphology analysis module extracts characteristic parameters of the blood flow image, wherein the extracted parameters comprise blood flow speed, blood vessel diameter, blood vessel density, the number of blood vessel branch nodes, blood vessel distortion degree and the like. And finally, the artificial intelligence analysis module synthesizes the characteristic parameters to carry out comprehensive decision, judges whether the tumor blood vessels are normalized or not and provides a prediction result of the normalization window period of the blood vessels. The display module is used for displaying and prompting the video image and the prediction result of the blood vessel normalization.
The tumor blood vessel normalization detection method of the embodiment of the invention comprises the following specific steps as shown in figure 2:
and 5, integrating the blood flow speed, the blood vessel density and the distortion degree of the tumor by an artificial intelligence analysis module, judging whether the tumor blood vessels are normalized or not, and giving a voice prompt of a tumor blood vessel normalization window period prediction result.
The specific method for acquiring the blood flow image in the step (1) comprises the following steps:
with Ns*NsThe window filter processes the original speckle image to obtain a normalized variance image Vns(x, y), the processing method is shown as formula (I):
wherein N issThe size of the window filter is generally 5-11, Is(i, j) is the gray value of the pixel in the window filter, i and j are respectively the horizontal and vertical coordinates of the pixel in the window filter,
the mean value of the pixel gray levels in the window filter is used, and x and y are the horizontal and vertical coordinates of the newly generated pixel points of the normalized variance image.
A two-dimensional blood flow image of the tumor is calculated using formula (II), where v (x, y) is the blood flow image.
The specific method for detecting the vascular structure in the step (2) comprises the following steps:
(a) respectively filtering the blood flow image by using 12 two-dimensional Gabor filters in different directions to obtain a filtered blood flow image Bλ,,θ(x, y), the treatment method is shown in formula (III):
Bλ,,θ(x,y)＝∫∫f(u,v)gλ,,θ(x-u,y-v)dudv (Ⅲ)
wherein f (u, v) is a blood flow image, gλ,,θ(x-u, y-v) is a two-dimensional Gabor filter, the parameter λ is the reciprocal of the spatial frequency, the parameter is the variance characteristic of the Gabor filter, and the value range of λ adopted in this embodiment is 0.1 to 10, and the value interval is 0.1, which is λ/2. The parameter θ is the filtering direction of the Gabor filter, and blood vessel detection in different directions can be achieved by changing the parameter θ, in this embodiment, the value of θ is N pi/12, where N is 0,1, …, and 12.
(b) Extraction of Bλ,,θMaximum response values max (B) in 12 different directions for each pixel in (x, y)λ,,θ(x, y)) are combined into a new image, i.e. the image of the vessel structure H (x, y).
The specific method for extracting the blood vessel density in the step (3) comprises the following steps:
(a) carrying out binarization threshold operation on the blood vessel structure image H (x, y), resetting a pixel value with a pixel gray value larger than 0 in the blood vessel structure image to be 1, and resetting a pixel gray value smaller than or equal to 0 to be 0; a new blood vessel structure image H' (x, y) is obtained again;
(b) and counting all pixels with the pixel value of 1 in H' (x, y), and dividing by the total pixels in the blood vessel structure image to obtain the blood vessel density.
The method for calculating the number of the blood vessel branch nodes in the step (3) comprises the following steps:
(a) extracting a new blood vessel central skeleton image of the blood vessel structure image H' (x, y) through a conventional window filter function such as a two-dimensional Gabor filter or a Gaussian filter;
(b) analyzing the communication relation between each pixel point and the neighborhood pixel point on the blood vessel central skeleton image, and when the communication number between a certain blood vessel central skeleton pixel point and the neighborhood blood vessel central skeleton pixel point is more than 3, marking the certain blood vessel central skeleton pixel point as a blood vessel branch node.
(c) Traversing the whole blood vessel central skeleton image, obtaining all blood vessel branch nodes in the image, and calculating the total number of the blood vessel branch nodes.
The method for calculating the blood vessel distortion degree in the step (4) comprises the following steps:
(a) dividing the blood vessel skeleton image into a plurality of sections according to branch nodes; calculating the curvature of each point of each section of the blood vessel skeleton by using a formula (IV);
(b) and taking an absolute value of the curvature of each point of a certain section of the blood vessel skeleton, and then carrying out superposition averaging to obtain the distortion degree of the section of the blood vessel.
The method for determining whether tumor blood vessels are normalized and the prediction result of the window period of normalization of blood vessels in step (5) is shown in FIG. 3:
(a) the system provided by the invention is used for collecting tumor speckle images of 1000 patients and extracting blood vessel morphological function information such as blood flow velocity, blood vessel density, blood vessel diameter, blood vessel distortion degree and the like.
(b) The biopsy channel of the system provided by the invention is used for collecting the biopsy sample of the tumor pathological tissue, and whether the tumor blood vessel is normalized or not is judged by an immunofluorescence or immunohistochemical method.
(c) Establishing a corresponding relation database of blood flow velocity, blood vessel density, blood vessel diameter, blood vessel distortion degree and tumor blood vessel normalization, wherein the training data are blood flow velocity, blood vessel density, blood vessel diameter and blood vessel distortion degree, and the training label is tumor blood vessel normalization or tumor blood vessel non-normalization.
(d) The database is used for training a deep neural network model, and the deep neural network model can be a conventional deep neural network model such as VGG16, GooleNet, ResNet and the like.
(e) And (4) predicting a new tumor blood flow image by using the training result, and giving a prompt for whether tumor blood vessels are normalized.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.
Claims (10)
1. An endoscopic tumor vessel normalization detection system, comprising:
the multi-mode light source module is used for illuminating the biological tissue to generate a white light image, a special spectrum image and a speckle image;
the endoscopic probe is used for extending into a human body cavity to illuminate and receive image signals;
the image processor is connected with the endoscopic probe and is used for carrying out image processing operations including demosaicing, denoising, defogging and color enhancement on the white light image acquired by the endoscopic probe;
the blood flow detection and analysis module is connected with the endoscopic probe and is used for processing the speckle images acquired by the endoscopic probe through an algorithm to obtain a blood flow velocity image;
the blood vessel morphology analysis module is connected with the blood flow detection analysis module and is used for extracting the tumor blood vessel morphology structure and measuring the tumor blood vessel density and the blood vessel distortion degree;
the artificial intelligence analysis module is connected with the blood vessel morphology analysis module and used for integrating the blood flow velocity, the blood vessel density and the distortion degree of the tumor blood, judging whether the tumor blood vessels are normalized or not and giving a prediction result of the normalization window period of the tumor blood vessels;
and the display module is connected with the output ends of the image processor and the artificial intelligence analysis module and is used for displaying images and prompting the judgment result of whether the tumor blood vessels are normalized.
2. The endoscopic tumor vessel normalization detection system according to claim 1, wherein the multi-modal light source comprises: a near-infrared laser light source and an LED light source; wherein the central wavelength range of the near-infrared laser light source is 660 nm-1380 nm; the spectrums of the LED light sources are combined into a white light spectrum, and the LED light sources at least comprise UV light sources with the wavelengths of 400 nm-430 nm.
3. The endoscopic tumor vessel normalization detection system according to claim 1, wherein the endoscopic probe comprises a light transmitting fiber bundle, an image sensor, a biopsy forceps channel and a nozzle which are connected with each other; the display module comprises a display and a voice prompter.
4. The endoscopic tumor vessel normalization detection system according to claim 1, wherein the hardware of the blood flow detection and analysis module is configured as a Field Programmable Gate Array (FPGA), a proprietary Asic chip or a high-performance graphics card; the hardware of the artificial intelligence analysis module is formed into a proprietary Asic chip or a high-performance display card.
5. An endoscopic tumor blood vessel normalization detection method using the endoscopic tumor blood vessel normalization detection system according to claim 1, comprising the steps of:
step 1, an endoscopic probe is inserted into a human body cavity, a laser light source in a multi-mode light source is turned on, a tumor speckle image of an illuminated biological tissue is collected, and a tumor blood flow image is reconstructed from the speckle image;
and 2, turning on a green light LED and a UV light LED light source in the multi-mode light source, wherein the green light LED and the UV light LED light source are in a proportion of 1: 1, performing mixed spectrum illumination according to the spectrum peak ratio, and combining an image processing technology to obtain a tumor blood vessel image;
step 3, extracting blood vessel density, blood vessel diameter and blood vessel branch nodes from the tumor blood vessel image;
step 4, calculating the distortion degree of the blood vessels, comparing the distortion degree with the preset normal tissue blood vessels with the same diameter, and grading the distortion degree of the tumor blood vessels;
and 5, integrating the blood flow speed, the blood vessel density and the distortion degree of the tumor by an artificial intelligence analysis module, judging whether the tumor blood vessels are normalized or not, and giving a voice prompt of a tumor blood vessel normalization window period prediction result.
6. The endoscopic tumor vessel normalization detection method according to claim 5, wherein the specific method for obtaining the tumor blood flow image in step 1 of the method is:
with Ns*NsThe window filter processes the original speckle image to obtain a normalized variance image Vns(x, y), the processing formula is:
wherein N issThe size of the window filter is 5-11, Is(i, j) is the gray value of the pixel in the window filter, i and j are respectively the horizontal and vertical coordinates of the pixel in the window filter,
the mean value of the pixel gray level in the window filter is used, and x and y are the horizontal and vertical coordinates of the newly generated pixel points of the normalized variance image;
calculating a two-dimensional blood flow image of the tumor, wherein the formula is as follows:
wherein, v (x, y) is the blood flow image.
7. The endoscopic tumor vessel normalization detection method according to claim 5, wherein the specific method for detecting and detecting the tumor vessel structure in step 2 of the method comprises:
(a) respectively filtering the blood flow image by using 12 two-dimensional Gabor filters in different directions to obtain a filtered blood flow image Bλ,,θ(x, y), the processing formula is:
Bλ,,θ(x,y)＝∫∫f(u,v)gλ,,θ(x-u,y-v)dudv
wherein f (u, v) is a blood flow image, gλ,,θ(x-u, y-v) is a two-dimensional Gabor filter, the parameter lambda is the reciprocal of the spatial frequency, the parameter is the variance characteristic of the Gabor filter, the value range of the lambda is 0.1-10, the value interval is 0.1, and is lambda/2; the parameter theta is the filtering direction of the Gabor filter, the blood vessel detection in different directions is realized by changing the parameter theta, and the value of theta is N pi/12, wherein N is 0,1, … and 12;
(b) extraction of Bλ,,θEach pixel in (x, y) is in 12 different directionsMaximum response value max (B)λ,,θ(x, y)) are combined into a new image, i.e. the image of the vessel structure H (x, y).
8. The endoscopic tumor vessel normalization detection method according to claim 5, wherein the specific method in step 3 of the method is:
the specific method for extracting the blood vessel density comprises the following steps:
(a) carrying out binarization threshold operation on the blood vessel structure image H (x, y), resetting a pixel value with a pixel gray value larger than 0 in the blood vessel structure image to be 1, and resetting a pixel gray value smaller than or equal to 0 to be 0; a new blood vessel structure image H' (x, y) is obtained again;
(b) counting all pixels with pixel value 1 in H' (x, y), and dividing by the total pixels in the blood vessel structure image to obtain blood vessel density;
the method for calculating the number of the blood vessel branch nodes comprises the following steps:
(a) extracting a new blood vessel central skeleton image of the blood vessel structure image H' (x, y) through a conventional window filter function such as a two-dimensional Gabor filter or a Gaussian filter;
(b) analyzing the communication relationship between each pixel point and a neighborhood pixel point on the blood vessel central skeleton image, and recording a certain blood vessel central skeleton pixel point as a blood vessel branch node when the communication number of the certain blood vessel central skeleton pixel point and the neighborhood blood vessel central skeleton pixel point is more than 3;
(c) traversing the whole blood vessel central skeleton image, obtaining all blood vessel branch nodes in the image, and calculating the total number of the blood vessel branch nodes.
9. The endoscopic tumor vessel normalization detection method according to claim 5, wherein the specific method for calculating the degree of vessel distortion in step 4 of the method is:
(a) dividing the blood vessel skeleton image into a plurality of sections according to branch nodes; calculating the curvature of each point of each section of the vascular skeleton by using the following formula;
wherein, K (x, y) represents curvature, and x, y represent horizontal and vertical coordinates of each point respectively;
(b) and taking an absolute value of the curvature of each point of a certain section of the blood vessel skeleton, and then carrying out superposition averaging to obtain the distortion degree of the section of the blood vessel.
10. The endoscopic tumor vessel normalization detection method according to claim 5, wherein the method for determining whether tumor vessels are normalized and the prediction result of the vessel normalization window period in step 5 comprises:
(a) collecting 1000 tumor speckle images by using a tumor blood vessel normalization detection system, and extracting blood vessel morphological function information of blood flow velocity, blood vessel density, blood vessel diameter and blood vessel distortion degree;
(b) collecting animal tumor tissue samples, and judging whether tumor blood vessels are normalized or not by an immunofluorescence or immunohistochemical method;
(c) establishing a corresponding relation database of blood flow velocity, blood vessel density, blood vessel diameter, blood vessel distortion degree and tumor blood vessel normalization, wherein the training data are blood flow velocity, blood vessel density, blood vessel diameter and blood vessel distortion degree, and the training label is tumor blood vessel normalization or tumor blood vessel non-normalization;
(d) training a deep neural network model by using the database, wherein the deep neural network model comprises VGG16, GooleNet and ResNet deep neural network models;
(e) and (4) predicting a new tumor blood flow image by using the training result, and giving a prompt for whether tumor blood vessels are normalized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010686934.0A CN111870231B (en) | 2020-07-16 | 2020-07-16 | Endoscopic tumor blood vessel normalization detection system and detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010686934.0A CN111870231B (en) | 2020-07-16 | 2020-07-16 | Endoscopic tumor blood vessel normalization detection system and detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111870231A true CN111870231A (en) | 2020-11-03 |
| CN111870231B CN111870231B (en) | 2022-06-03 |
Family
ID=73155520
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010686934.0A Active CN111870231B (en) | 2020-07-16 | 2020-07-16 | Endoscopic tumor blood vessel normalization detection system and detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111870231B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112580437A (en) * | 2020-11-26 | 2021-03-30 | 复旦大学附属中山医院 | Method and device for measuring size of target |
| CN112998651A (en) * | 2021-02-10 | 2021-06-22 | 中国医学科学院北京协和医院 | Application of photoacoustic imaging in breast tumor scoring system and scoring system |
| CN114366045A (en) * | 2022-01-14 | 2022-04-19 | 山东承势电子科技有限公司 | Thoracic aorta vascular morphology feature calculation device |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102197983A (en) * | 2010-03-26 | 2011-09-28 | 富士胶片株式会社 | Electronic endoscope system |
| CN102429650A (en) * | 2011-11-10 | 2012-05-02 | 华中科技大学 | Laser speckle blood flow imaging contrast analytical method |
| CN102842136A (en) * | 2012-07-19 | 2012-12-26 | 湘潭大学 | Optic disc projection location method synthesizing vascular distribution with video disc appearance characteristics |
| CN105748029A (en) * | 2016-02-18 | 2016-07-13 | 深圳开立生物医疗科技股份有限公司 | Imaging system of endoscope |
| CN106413536A (en) * | 2014-05-23 | 2017-02-15 | 柯惠有限合伙公司 | Systems for imaging of blood flow in laparoscopy |
| CN108430306A (en) * | 2015-10-09 | 2018-08-21 | 瓦索普蒂奇医疗公司 | System and method for using laser speckle contrast Imaging fast to check vascular system and particle stream |
| CN109497955A (en) * | 2018-12-18 | 2019-03-22 | 聚品(上海)生物科技有限公司 | Human body spontaneous fluorescent illumination excitation and image processing system and method |
| CN109893102A (en) * | 2019-01-15 | 2019-06-18 | 温州医科大学 | A kind of parser of macular area choroidal capillaries density |
| CN110151108A (en) * | 2019-05-10 | 2019-08-23 | 南京航空航天大学 | Endoscopic laser speckle blood flow blood oxygen imaging system |
| CN111295135A (en) * | 2017-08-28 | 2020-06-16 | 东卡罗莱娜大学 | Multi-spectral physiological visualization (MSPV) using laser imaging methods and systems for blood flow and perfusion imaging and quantification in endoscopic design |
-
2020
- 2020-07-16 CN CN202010686934.0A patent/CN111870231B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102197983A (en) * | 2010-03-26 | 2011-09-28 | 富士胶片株式会社 | Electronic endoscope system |
| CN102429650A (en) * | 2011-11-10 | 2012-05-02 | 华中科技大学 | Laser speckle blood flow imaging contrast analytical method |
| CN102842136A (en) * | 2012-07-19 | 2012-12-26 | 湘潭大学 | Optic disc projection location method synthesizing vascular distribution with video disc appearance characteristics |
| CN106413536A (en) * | 2014-05-23 | 2017-02-15 | 柯惠有限合伙公司 | Systems for imaging of blood flow in laparoscopy |
| CN108430306A (en) * | 2015-10-09 | 2018-08-21 | 瓦索普蒂奇医疗公司 | System and method for using laser speckle contrast Imaging fast to check vascular system and particle stream |
| CN105748029A (en) * | 2016-02-18 | 2016-07-13 | 深圳开立生物医疗科技股份有限公司 | Imaging system of endoscope |
| CN111295135A (en) * | 2017-08-28 | 2020-06-16 | 东卡罗莱娜大学 | Multi-spectral physiological visualization (MSPV) using laser imaging methods and systems for blood flow and perfusion imaging and quantification in endoscopic design |
| CN109497955A (en) * | 2018-12-18 | 2019-03-22 | 聚品(上海)生物科技有限公司 | Human body spontaneous fluorescent illumination excitation and image processing system and method |
| CN109893102A (en) * | 2019-01-15 | 2019-06-18 | 温州医科大学 | A kind of parser of macular area choroidal capillaries density |
| CN110151108A (en) * | 2019-05-10 | 2019-08-23 | 南京航空航天大学 | Endoscopic laser speckle blood flow blood oxygen imaging system |
Non-Patent Citations (1)
| Title |
|---|
| 沈娜: "抗血管生成药物时间窗的研究进展", 《肿瘤预防与治疗》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112580437A (en) * | 2020-11-26 | 2021-03-30 | 复旦大学附属中山医院 | Method and device for measuring size of target |
| CN112998651A (en) * | 2021-02-10 | 2021-06-22 | 中国医学科学院北京协和医院 | Application of photoacoustic imaging in breast tumor scoring system and scoring system |
| CN112998651B (en) * | 2021-02-10 | 2021-08-27 | 中国医学科学院北京协和医院 | Application of photoacoustic imaging in breast tumor scoring system and scoring system |
| CN114366045A (en) * | 2022-01-14 | 2022-04-19 | 山东承势电子科技有限公司 | Thoracic aorta vascular morphology feature calculation device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111870231B (en) | 2022-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111870231B (en) | Endoscopic tumor blood vessel normalization detection system and detection method | |
| US9962090B2 (en) | Reflective mode multi-spectral time-resolved optical imaging methods and apparatuses for tissue classification | |
| US11937973B2 (en) | Systems and media for automatically diagnosing thyroid nodules | |
| US20220142484A1 (en) | Reflective mode multi-spectral time-resolved optical imaging methods and apparatuses for tissue classification | |
| JP5410274B2 (en) | How to characterize an organization | |
| WO2017074505A1 (en) | Reflective mode multi-spectral time-resolved optical imaging methods and apparatuses for tissue classification | |
| US11024031B1 (en) | System and method for diagnosing severity of gastric cancer | |
| US11232570B2 (en) | System and method for diagnosing severity of gastritis | |
| CN111870230A (en) | Multi-parameter tumor blood vessel normalization detection system and detection method | |
| WO2009037464A1 (en) | Imaging apparatus and method | |
| CN113288186A (en) | Deep learning algorithm-based breast tumor tissue detection method and device | |
| CN113143297B (en) | Focus detection system and method | |
| CN111870224B (en) | Tumor blood vessel normalization detection system and detection method | |
| Verikas et al. | Advances in laryngeal imaging | |
| Negied | Infrared thermography-based breast cancer detection—comprehensive investigation | |
| Lixin et al. | Recent advances of hyperspectral imaging application in biomedicine | |
| Leng et al. | Rectal cancer treatment management: deep-learning neural network based on photoacoustic microscopy image outperforms histogram-feature-based classification | |
| Turani et al. | Optical radiomic signatures derived from OCT images to improve identification of melanoma | |
| CN111798426B (en) | Deep learning and detecting system for mitotic image in gastrointestinal stromal tumor of moving end | |
| KR101725648B1 (en) | Method for Analyzing an Ischemic Brain Optically Using an Indocyanine Green | |
| Jayanthi et al. | Diffuse reflectance imaging: a tool for guided biopsy | |
| WO2022193073A1 (en) | Lesion detection system and method | |
| Patil et al. | An Analysis of Techniques for Detecting Dental Care: A Brief Survey | |
| Sagari et al. | A Comprehensive Study of Advances in Oral Cancer Detection Using Image Processing Techniques | |
| Ferrarese et al. | A new paradigm for geometric accuracy prediction in medical image segmentation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |