A fluorescence diagnosis assistance system for Helicobacter pylori

Abstract

The present invention relates to the technical field of medical image processing, and discloses a fluorescence diagnosis assistance system for Helicobacter pylori, comprising: an image acquisition and processing module, configured to obtain optical signals emitted in the stomach under white light and narrowband spectra, and convert the optical signals into hyperspectral images; an image analysis module, configured to identify whether each pixel point is positive, and based on the identification results of all pixel points, determine whether the hyperspectral image is positive to obtain a first determination result; slice the hyperspectral image, and after identifying whether each sub-image is positive, based on the identification results of all sub-images, determine whether the hyperspectral image is positive to obtain a second determination result; based on the first determination result and the second determination result, determine whether the stomach is positive for Helicobacter pylori. Visualization, quantification, and non-invasiveness of endoscopic Hp diagnosis are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and particularly to a fluorescence diagnosis assistance system for Helicobacter pylori. Background Art

[0002] The statements in this part only mention the background art related to the present invention, and do not necessarily constitute the prior art.

[0003] Helicobacter pylori (Hp) infection is one of the most common bacterial infections, and the current infection rate is close to 60%. The prerequisite for eradicating Hp is accurate diagnosis of Hp infection. Endoscopic diagnosis of Hp infection can prompt the treatment of subsequent gastric diseases (such as gastric cancer). At present, the most commonly used method during gastroscopy is the rapid urease test, which can quickly obtain test results through biopsy tissues. However, the rapid urease test also has the following disadvantages: Sampling can cause trauma, and if the patient has abnormal coagulation or other conditions, sampling may lead to bleeding risk; Biopsy sampling is restricted by the site. The negative result of a single tissue is not sufficient to represent the entire pylorus or even the whole stomach, and it is impossible to judge the lesion range and the situation of Hp infection; The result interpretation is judged by medical staff with the naked eye, which is subjective. Therefore, it is of great significance to develop a method for diagnosing Hp infection that is rapid, non-invasive, comprehensive, and objective during gastroscopy.

[0004] Fluorescence imaging diagnosis is a diagnostic technique based on photochemistry. By irradiating the lesion site with excitation light, diagnosis can be carried out according to the tissue spectral characteristics. When the excitation light interacts with the luminescent substance in biological tissue, the excitation light transfers energy to the luminescent substance. After the luminescent substance absorbs the energy, it can transition to a high-energy state. The high-energy state is unstable and will transition to the ground state through various channels, generating fluorescence. Fixing the wavelength of the excitation light and detecting the change of fluorescence intensity with wavelength, the fluorescence spectrum of the substance is obtained. By comparing the fluorescence spectra of the lesion site and the normal site, the diagnosis and localization of the lesion can be achieved. The application of excitation light around 400nm in fluorescence diagnosis is very popular. Due to the limitation of short-wave excitation on its penetration depth in tissues, it is mostly used for superficial diagnosis. With the development of laser technology, laser-induced fluorescence imaging diagnosis can achieve rapid, non-destructive, and accurate detection of lesions, but its development is restricted by diagnostic equipment. Hp can spontaneously produce and accumulate endogenous porphyrins. Porphyrins are a class of macromolecular heterocyclic compounds formed by the α-carbon atoms of four pyrrole-like subunits interconnected by methylene bridges, and can participate in the light-induced reactions in organisms as photosensitizers. Porphyrins can produce red fluorescence when excited by light with a wavelength of about 405nm, and there are two fluorescence peaks at 632nm and 690nm, which can be used for the diagnosis and imaging of the lesion site.

[0005] In recent years, optical endoscopes, especially Narrow Band Imaging (NBI) technology, have been developed for the diagnosis of early gastrointestinal cancer. The light source of the optical endoscope is an LED with a wavelength range of 400 - 700 nm, which can excite the endogenous porphyrin produced by bacteria to emit fluorescence. NBI filters out the broadband spectrum through a filter in a way of pre - processing the endoscope light source, and only retains the narrow - band spectra of 415 nm and 540 nm, which can well display the microvascular structure of the superficial mucosa and the submucosal blood vessels. The wavelength of NBI at 415 nm is very close to the wavelength that excites porphyrin. Previous studies have found that the white - light LED light source and the narrow - band spectra of NBI can excite the endogenous porphyrin of Hp. After the porphyrin absorbs light and reaches the excited state, it can react with oxygen in the environment to generate reactive oxygen species, thereby damaging Hp and achieving a bactericidal effect. Similarly, the endogenous porphyrin of Hp is excited by a white - light endoscope or an NBI endoscope to emit red fluorescence. However, the optical signals captured by traditional digestive endoscope cameras are relatively single, the detection accuracy is not high, and it is difficult to clearly distinguish the tissue background, autofluorescence, and the mixture of other bacteria species solely through these signals. Summary of the Invention

[0006] To solve the deficiencies of the prior art, the present invention provides a fluorescence diagnostic assistance system for Helicobacter pylori, which requires no additional device and only performs gastroscopy under white - light and NBI modes. By using a light source with a specific wavelength to excite the endogenous porphyrin of Hp in the gastric mucosa, the fluorescence signal emitted is converted into a hyperspectral image. Through the combination of pixel - point recognition and sub - image recognition of the hyperspectral image, the visualization, quantification, and non - invasiveness of Hp diagnosis under the endoscope are achieved.

[0007] In the first aspect, the present invention provides a fluorescence diagnostic assistance system for Helicobacter pylori;

[0008] A fluorescence diagnostic assistance system for Helicobacter pylori includes:

[0009] An image acquisition and processing module, which is used to obtain the optical signals emitted in the stomach under white - light and narrow - band spectra and convert the optical signals into a hyperspectral image;

[0010] An image analysis module, which is used to identify whether each pixel point in the hyperspectral image is positive, and based on the recognition results of all pixel points, judge whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain a plurality of sub - images, identify whether each sub - image is positive, and then based on the recognition results of all sub - images, judge whether the hyperspectral image is positive to obtain a second judgment result; based on the first judgment result and the second judgment result, obtain whether the stomach is positive for Helicobacter pylori.

[0011] Furthermore, for the recognition of whether the pixel point is positive, a random forest algorithm is adopted.

[0012] Further, for identifying whether the sub - graph is positive, a deep residual network is adopted.

[0013] Further, the sub - graph is obtained by sampling and slicing the hyperspectral image row - by - row.

[0014] Further, the first judgment result is obtained by adopting a multi - pixel voting mechanism based on the identification results of all pixel points.

[0015] Further, the second judgment result is obtained by adopting a multi - sub - graph voting mechanism based on the identification results of all sub - graphs.

[0016] In a second aspect, the present invention further provides an electronic device, including:

[0017] A memory for non - temporarily storing computer - readable instructions; and

[0018] A processor for running the computer - readable instructions,

[0019] wherein, when the computer - readable instructions are run by the processor, the following steps are executed:

[0020] Obtain the optical signals emitted in the stomach under the action of white light and narrow - band spectra, and convert the optical signals into a hyperspectral image;

[0021] Identify whether each pixel point in the hyperspectral image is positive, and based on the identification results of all pixel points, judge whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain a plurality of sub - graphs, and after identifying whether each sub - graph is positive, based on the identification results of all sub - graphs, judge whether the hyperspectral image is positive to obtain a second judgment result; based on the first judgment result and the second judgment result, obtain whether the stomach is positive for Helicobacter pylori.

[0022] Further, the sub - graph is obtained by sampling and slicing the hyperspectral image row - by - row.

[0023] In a third aspect, the present invention further provides a storage medium that non - temporarily stores computer - readable instructions, wherein when the non - temporary computer - readable instructions are executed by a computer, the following steps are executed:

[0024] Obtain the optical signals emitted in the stomach under the action of white light and narrow - band spectra, and convert the optical signals into a hyperspectral image;

[0025] Identify whether each pixel point in the hyperspectral image is positive, and based on the identification results of all pixel points, determine whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain multiple sub-images, identify whether each sub-image is positive, and then based on the identification results of all sub-images, determine whether the hyperspectral image is positive to obtain a second judgment result; based on the first judgment result and the second judgment result, determine whether the stomach is positive for Helicobacter pylori.

[0026] Further, the sub-images are obtained by slicing the hyperspectral image by row sampling.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] The present invention does not require additional devices and only performs gastroscopy under white light and NBI modes. By exciting the endogenous porphyrin of Hp in the gastric mucosa with a light source of a specific wavelength, converting the emitted fluorescence signal into a hyperspectral image, and combining the identification of pixel points and sub-images of the hyperspectral image, the visualization, quantification, and non-invasiveness of Hp diagnosis under endoscopy are realized.

[0029] Advantages of additional aspects of the present invention will be partially given in the following description or learned through the practice of the present invention. Description of the Drawings

[0030] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0031] Figure 1 It is a schematic diagram of data flow of a fluorescence diagnosis assistance system for Helicobacter pylori in the first embodiment;

[0032] Figure 2 It is a schematic diagram of the spectral feature vector in the first embodiment. Detailed Description of the Embodiment

[0033] It should be noted that the following detailed descriptions are all exemplary and are intended to provide further explanations of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0034] Without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0035] All data acquisition in this embodiment is based on compliance with laws and regulations and user consent, and is a legal application of the data.

[0036] Embodiment 1

[0037] This embodiment provides a fluorescence diagnosis assistance system for Helicobacter pylori.

[0038] The fluorescence diagnosis assistance system for Helicobacter pylori provided in this embodiment converts and amplifies the color signals collected by the endoscopic system into hyperspectral images through calculation, and can clearly distinguish the intensity and coverage area of the fluorescence signals emitted by endogenous porphyrins in gastric mucosa Hp, which can reflect the infection degree and involved sites of Hp.

[0039] The fluorescence diagnosis assistance system for Helicobacter pylori provided in this embodiment does not require additional devices. Only gastroscopy is performed in white light and NBI modes. The endogenous porphyrins in gastric mucosa Hp are excited by a light source with a specific wavelength, and the emitted fluorescence signals are converted into hyperspectral images. By identifying the fluorescence intensity and area of a specific wavelength in the stomach, the visualization, quantification, and non-invasiveness of Hp diagnosis under endoscopy are realized.

[0040] The fluorescence diagnosis assistance system for Helicobacter pylori provided in this embodiment combines narrowband imaging (NBI) technology and hyperspectral imaging technology. The endogenous porphyrins in gastric mucosa Hp are excited by a light source with a specific wavelength, and the emitted fluorescence signals are converted into hyperspectral images. The hyperspectral images are analyzed and identified by a deep learning algorithm.

[0041] The fluorescence diagnosis assistance system for Helicobacter pylori provided in this embodiment includes: a light source and imaging module, an image acquisition and processing module, and an image analysis module.

[0042] The light source and imaging module, as Figure 1 shown, uses an LED light source with a wavelength range of 400 - 700 nm, which can excite the endogenous porphyrins produced by Hp to emit fluorescence; through NBI technology, using narrowband spectra of 415 nm and 540 nm, the microvascular structure of the superficial mucosa and the submucosal blood vessels are optimally displayed.

[0043] The image acquisition and processing module uses a traditional digestive endoscope camera to capture the optical signals emitted in the stomach under the action of the LED light source (white light) and narrowband spectra, and converts the collected color signals (optical signals) into hyperspectral images through a calculation method to clearly distinguish the fluorescence signals emitted by endogenous porphyrins in gastric mucosa Hp.

[0044] The image analysis module uses a random forest algorithm and a ResNet34 deep learning model.

[0045] Random Forest Algorithm: This algorithm is a learning technique based on the integration of decision trees. By integrating multiple decision trees, it improves the stability and accuracy of the model. Although decision trees, as weak classifiers, may have problems such as overfitting, random forests effectively overcome these problems through the integration method.

[0046] ResNet34 Deep Learning Model: To handle image classification tasks, the highly performant ResNet (Deep Residual Network) 34 model is adopted. This model solves the degradation problem in deep neural networks by introducing residual connections. The parameters and structure of the model have been finely tuned.

[0047] Image Analysis Module: First, pixel-level samples are used, with the spectral feature vector of each pixel in the hyperspectral image as the training instances and test data. Random forests are used for spectral signal recognition and classification. Finally, a multi-pixel voting mechanism is adopted, where whether an image is positive is determined by the predicted labels of the majority of pixels. That is, based on the spectral feature vector of the pixels, each pixel point in the hyperspectral image is identified as to whether it is positive, and a multi-pixel voting mechanism is used to determine whether the hyperspectral image is positive, obtaining a first judgment result.

[0048] Among them, as Figure 2 shown, each pixel point in the hyperspectral image corresponds to a curve, and the spectral feature vector of a pixel is a curve corresponding to the pixel point in the hyperspectral image, which is manifested as fluorescence intensity and area.

[0049] Image Analysis Module: To further improve the recognition accuracy and system performance, the hyperspectral image is sampled and sliced row by row to form multiple sub-images, constituting a two-dimensional image dataset. The labeled two-dimensional sub-images are input into the ResNet34 model for supervised training and testing. Here, a multi-sub-image voting mechanism is adopted, and the predicted labels of the majority of sub-images determine the label of the hyperspectral image. That is, the hyperspectral image is sliced to form multiple sub-images, and after identifying whether each sub-image is positive, a multi-sub-image voting mechanism is used to determine whether the hyperspectral image is positive, obtaining a second judgment result.

[0050] Image Analysis Module: Based on the first judgment result and the second judgment result, it is determined whether the stomach is positive for Helicobacter pylori.

[0051] The characteristics of hyperspectral images are related to the field of view captured by the hyperspectral imager. Different fields of view may bring different spectral characteristics. However, for the same patient, even with different fields of view, there are still similar patterns among the hyperspectral images. For negative and positive hyperspectral images, there are specific differences in their spectral distribution characteristics in certain patterns, and these differences are learnable. For the rigor of the results, multiple fields of view are taken for a patient, and the magnification of the field of view is adjusted to be consistent to capture the comprehensive fluorescence spectral distribution characteristics.

[0052] The first judgment result and the second judgment result are a process of mutual verification. When both the first judgment result and the second judgment result of an image are positive / negative, the image is finally considered positive / negative; when the two judgment results are inconsistent, it is necessary to combine the gold standard for Helicobacter pylori diagnosis to determine the infection status of the patient, and then re-take the hyperspectral image at an appropriate angle according to this result and put it into the model for testing again.

[0053] Embodiment 2

[0054] This embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the above one or more computer programs are stored in the memory. When the electronic device runs, the processor executes the one or more computer programs stored in the memory so that the electronic device performs the following steps:

[0055] Obtain the optical signals emitted in the stomach under the action of white light and narrowband spectra, and convert the optical signals into hyperspectral images;

[0056] Identify whether each pixel point in the hyperspectral image is positive, and based on the recognition results of all pixel points, judge whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain multiple sub-images, and identify whether each sub-image is positive, and then based on the recognition results of all sub-images, judge whether the hyperspectral image is positive to obtain a second judgment result; based on the first judgment result and the second judgment result, obtain whether the stomach is positive for Helicobacter pylori.

[0057] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0058] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0059] In the implementation process, each step of the above method may be completed by the integrated logic circuit in the hardware of the processor or the instructions in the form of software.

[0060] The method in the first embodiment can be directly embodied as being executed by the hardware processor, or executed by the combination of the hardware and software modules in the processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0061] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with this embodiment can be implemented by electronic hardware or the combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but this implementation should not be considered to exceed the scope of the present invention.

[0062] Embodiment Three

[0063] This embodiment also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the following steps are performed:

[0064] Obtain the optical signals emitted in the stomach under the action of white light and narrow-band spectra, and convert the optical signals into hyperspectral images;

[0065] Identify whether each pixel point in the hyperspectral image is positive, and based on the recognition results of all pixel points, determine whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain a plurality of sub-images, and after identifying whether each sub-image is positive, based on the recognition results of all sub-images, determine whether the hyperspectral image is positive to obtain a second judgment result; based on the first judgment result and the second judgment result, obtain whether the stomach is positive for Helicobacter pylori.

[0066] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A Helicobacter pylori fluorescence diagnosis auxiliary system, characterized in that: include: The light source and imaging module uses white light with a wavelength range of 400 to 700 nm to stimulate the endogenous porphyrin produced by Helicobacter pylori to emit fluorescence. Through narrow-band imaging technology, the narrow-band spectra of 415 nm and 540 nm are used to display the microvascular structure of the superficial mucosa and the blood vessels in the submucosal layer. An image acquisition and processing module is used to obtain optical signals emitted from the stomach under the action of white light and narrow-band spectrum, and convert the optical signals into hyperspectral images; The image analysis module is used to identify whether each pixel in the hyperspectral image is positive, and based on the identification results of all the pixels, determine whether the hyperspectral image is positive to obtain a first judgment result; slice the hyperspectral image to obtain multiple sub-images, and after identifying whether each sub-image is positive, determine whether the hyperspectral image is positive based on the identification results of all the sub-images to obtain a second judgment result; based on the first judgment result and the second judgment result, determine whether the stomach is positive for Helicobacter pylori; The sub-image is obtained by sampling and slicing the hyperspectral image by rows; The first judgment result is obtained based on the recognition results of all pixels using a multi-pixel voting mechanism; The second judgment result is obtained based on the recognition results of all sub-graphs by using a multi-sub-graph voting mechanism; The first judgment result and the second judgment result are a process of mutual verification. When the first judgment result and the second judgment result of an image are both positive / negative, the image is finally considered to be positive / negative. When the two judgment results are inconsistent, it is necessary to combine the gold standard for Helicobacter pylori diagnosis to determine the patient's infection status, re-shoot the hyperspectral image at an appropriate angle based on this result, and put it into the model for testing again.

2. A Helicobacter pylori fluorescence diagnosis auxiliary system as claimed in claim 1, characterized in that: Whether the pixel point is positive or not is identified by using a random forest algorithm.

3. The Helicobacter pylori fluorescence diagnosis auxiliary system as claimed in claim 1, characterized in that: Whether the sub-image is positive or not is identified using a deep residual network.

4. An electronic device, comprising: a memory for non-transitory storage of computer readable instructions; as well as a processor for executing the computer readable instructions, When the computer readable instructions are executed by the processor, the following steps are performed: White light with a wavelength range of 400 to 700 nm is used to stimulate the endogenous porphyrin produced by Helicobacter pylori to emit fluorescence. Through narrow-band imaging technology, the narrow-band spectra of 415 nm and 540 nm are used to display the microvascular structure of the superficial mucosa and the blood vessels in the submucosal layer. Acquire the optical signals emitted from the stomach under the action of white light and narrow-band spectrum, and convert the optical signals into hyperspectral images; Identify whether each pixel in the hyperspectral image is positive, and determine whether the hyperspectral image is positive based on the identification results of all the pixels to obtain a first determination result; slice the hyperspectral image to obtain a plurality of sub-images, and after identifying whether each sub-image is positive, determine whether the hyperspectral image is positive based on the identification results of all the sub-images to obtain a second determination result; determine whether the stomach is positive for Helicobacter pylori based on the first determination result and the second determination result; The sub-image is obtained by sampling and slicing the hyperspectral image by rows; The first judgment result is obtained based on the recognition results of all pixels using a multi-pixel voting mechanism; The second judgment result is obtained based on the recognition results of all sub-graphs by using a multi-sub-graph voting mechanism; The first judgment result and the second judgment result are a process of mutual verification. When the first judgment result and the second judgment result of an image are both positive / negative, the image is finally considered to be positive / negative. When the two judgment results are inconsistent, it is necessary to combine the gold standard for Helicobacter pylori diagnosis to determine the patient's infection status, re-shoot the hyperspectral image at an appropriate angle based on this result, and put it into the model for testing again.

5. A storage medium characterized by being non-temporary The computer readable instructions are stored in a manner such that when the computer readable instructions are executed, the following steps are performed: The light source and imaging module uses white light with a wavelength range of 400 to 700 nm to stimulate the endogenous porphyrin produced by Helicobacter pylori to emit fluorescence. Through narrow-band imaging technology, the narrow-band spectra of 415 nm and 540 nm are used to display the microvascular structure of the superficial mucosa and the blood vessels in the submucosal layer. Acquire the optical signals emitted from the stomach under the action of white light and narrow-band spectrum, and convert the optical signals into hyperspectral images; Identify whether each pixel in the hyperspectral image is positive, and determine whether the hyperspectral image is positive based on the identification results of all the pixels to obtain a first determination result; slice the hyperspectral image to obtain a plurality of sub-images, and after identifying whether each sub-image is positive, determine whether the hyperspectral image is positive based on the identification results of all the sub-images to obtain a second determination result; determine whether the stomach is positive for Helicobacter pylori based on the first determination result and the second determination result; The sub-image is obtained by sampling and slicing the hyperspectral image by rows; The first judgment result is obtained based on the recognition results of all pixels using a multi-pixel voting mechanism; The second judgment result is obtained based on the recognition results of all sub-graphs by using a multi-sub-graph voting mechanism; The first judgment result and the second judgment result are a process of mutual verification. When the first judgment result and the second judgment result of an image are both positive / negative, the image is finally considered to be positive / negative. When the two judgment results are inconsistent, it is necessary to combine the gold standard for Helicobacter pylori diagnosis to determine the patient's infection status, re-shoot the hyperspectral image at an appropriate angle based on this result, and put it into the model for testing again.

Images