An algorithm for calculating the optical parameter distribution of skin tissue

Abstract

The present invention relates to the field of medical technology, and discloses an algorithm for calculating the distribution of optical parameters of skin tissue, comprising the following steps: S1, using biophotonics Monte Carlo simulation to calculate the diffuse reflection intensity distribution of the optical tissue surface of a single property, and using this result to train the corresponding convolution kernel for imaging calculation; S2, using a digital phantom with different optical parameters at different positions and a previously obtained convolution kernel for calculating the light propagation of a single property tissue, to construct and train a deep convolution network for imaging calculation; S3, collecting the diffuse reflection light intensity distribution of the patient's skin surface and performing preprocessing; using the previously trained deep convolution network and the patient's diffuse reflection light intensity distribution obtained by the detection system to perform imaging calculation, and obtain a three-dimensional distribution imaging diagram of the detected optical parameters.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to an algorithm for calculating the distribution of optical parameters of skin tissue. Background Art

[0002] The core problem of diffuse optical tomography is that the method directly calculating according to the diffuse optical tomography method has the problem of low resolution or too long calculation time.

[0003] Therefore, it is necessary to provide a new calculation method based on artificial neural network, which uses the experience of Monte Carlo simulation results to learn the relationship between diffuse reflection distribution and internal optical parameters of skin tissue, such as absorption coefficient and reflection coefficient distribution, to complete the rapid computational imaging of its three-dimensional distribution. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the deficiencies in the prior art, the present invention provides an algorithm for calculating the distribution of optical parameters of skin tissue, which solves the problem of providing a fast three-dimensional imaging algorithm that can non-invasively measure the distribution of optical parameters of tissue within a certain depth under the skin tissue, and uses a specific light source to set the corresponding diffuse reflection light intensity distribution to calculate the location of heterogeneities inside the tissue and their optical properties, so as to assist in distinguishing the properties of the heterogeneities.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solution: an algorithm for calculating the distribution of optical parameters of skin tissue, comprising the following steps:

[0008] S1. Use biophotonics Monte Carlo simulation to calculate the diffuse reflection intensity distribution of the optical tissue surface of a single property, and use this result to train the corresponding convolution kernel for imaging calculation;

[0009] S2. Use the digital phantom simulation calculation containing different optical parameters of tissues at different positions or the diffuse reflection two-dimensional light intensity distribution obtained by the real phantom and measurement system, as well as the convolution kernel of the light propagation calculation of single-property tissue obtained previously, to build and train a deep convolutional network for imaging calculation;

[0010] S3. Collect the diffuse reflection light intensity distribution of the patient's skin surface and perform preprocessing; use the previously trained deep convolutional network and the patient's diffuse reflection light intensity for imaging calculation.

[0011] Preferably, the step S1 obtains the diffuse reflection intensity distribution of a uniform tissue with known optical parameters under specific light intensity input conditions through a detection system, or uses an equivalent setting of the detection system to perform biophotonics Monte Carlo simulation, obtain the diffuse reflection intensity distribution, establish a convolutional network, and use the obtained diffuse reflection light intensity distribution to train a convolution kernel that reflects the propagation law of light in the tissue with corresponding optical parameters according to the convolutional network error back propagation rule.

[0012] Preferably, the step S2 uses the convolution principle and formula 1 to calculate the diffuse reflection intensity to establish a deep convolution network:

[0013]

[0014] Where K is the convolution kernel that can describe the transmission characteristics corresponding to the optical properties of the medium. Assuming isotropic diffusion, that is, the diffusion function for propagation to the shallow layer and the deep layer is the same, the measurement operator at the boundary of the medium surface is M. According to the analysis of the propagation path of light between layers, the diffuse reflection light intensity distribution of the measurement surface is calculated and expressed as:

[0015] The pixel network output is obtained according to the deep network calculation.

[0016] Preferably, the step S3 establishes a deep convolutional network according to the network for obtaining uniform tissue convolution kernel parameter calculation and formula three, wherein the light transmission convolution kernel part indicated by the label is a set of light transmission convolution kernels of different tissues obtained in step S2, and the rest of the convolution network is a modifiable part, and the initial value is randomly generated. In the formula, Mki,n(x,y) represents the mask corresponding to the nth tissue characteristic at the i-th layer, and its value is 0 or 1, indicating whether the tissue at the coordinate (x,y) position is the nth tissue;

[0017] Formula 3

[0018] The diffuse reflection light intensity distribution of the patient's skin surface is obtained, and an error detection is performed on the result output by the network in step S2.

[0019] (III) Beneficial effects

[0020] The present invention provides an algorithm for calculating the distribution of optical parameters of skin tissue, which has the following beneficial effects:

[0021] The present invention obtains the diffuse reflection intensity distribution of a uniform tissue with known optical parameters under a specific light intensity input condition through a detection system, or uses an equivalent setting of the detection system to perform biophotonic Monte Carlo simulation, obtain the diffuse reflection intensity distribution, establish a convolution network, use the obtained diffuse reflection intensity distribution, train a convolution kernel that reflects the propagation law of light in the tissue with corresponding optical parameters according to the convolution network error back propagation rule, use the convolution principle and formula 1 to calculate the diffuse reflection intensity, establish a deep convolution network, then calculate the network output according to the deep network, use the obtained two-dimensional diffuse reflection intensity distribution as the target intensity, use the root mean square of the difference between the network output of all pixels and the supervision image intensity as the error, and then establish a deep convolution network to calculate and obtain the diffuse reflection intensity distribution of the patient's skin surface according to the network and formula 3 for obtaining the uniform tissue convolution kernel parameter calculation, detect the error, and return to the deep convolution network for recalculation if the condition is not met. Through the final corrected three-dimensional tissue optical parameter distribution imaging, it can be judged whether there is a heterogeneous body in the tissue or its potential possibility or degree of harm can be inferred through the optical properties of the heterogeneous body. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of a detection system for obtaining diffuse reflection light intensity distribution according to the present invention;

[0023] Figure 2 This is a diagram explaining the principle of using convolution to calculate diffuse reflection light intensity in the present invention;

[0024] Figure 3 A network structure diagram for obtaining uniform tissue convolution kernel parameter calculation in the present invention;

[0025] Figure 4 A network structure diagram for performing optical tomography calculations in the present invention;

[0026] Figure 5 This is a complete algorithm flow chart of the present invention;

[0027] Figure 6 This is a pseudo-color schematic diagram of the tissue diffuse reflection intensity distribution of the present invention;

[0028] Figure 7 It is a schematic diagram of the tested tissue and imaging results of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] like Figure 1-7 As shown, the present invention provides a technical solution: an algorithm for calculating the distribution of optical parameters of skin tissue, comprising the following steps:

[0031] S1. Calculate the diffuse reflection intensity distribution of the optical tissue surface of a single property by using biophotonics Monte Carlo simulation, and use this result to train the corresponding convolution kernel for imaging calculation. Step S1 obtains the diffuse reflection intensity distribution of a uniform tissue body with known optical parameters under specific light intensity input conditions through a detection system, or uses an equivalent setting of the detection system to perform biophotonics Monte Carlo simulation, obtain the diffuse reflection intensity distribution, establish a convolution network, use the obtained diffuse reflection light intensity distribution, and train a convolution kernel that reflects the propagation law of light in the tissue with corresponding optical parameters according to the convolution network error back propagation rule, such as Figure 1 As shown;

[0032] S2. Using the digital phantom simulation calculation containing different tissues with different optical parameters at different positions or the diffuse reflection two-dimensional light intensity distribution obtained by the real phantom and measurement system and the previously obtained single-property tissue light propagation calculation convolution kernel, a deep convolution network for imaging calculation is constructed and trained. Step S2 uses the convolution principle of diffuse reflection light intensity calculation and formula 1 to establish a deep convolution network:

[0033]

[0034] Where K is the convolution kernel that can describe the transmission characteristics corresponding to the optical properties of the medium. Assuming isotropic diffusion, that is, the diffusion function for propagation to the shallow layer and the deep layer is the same, the measurement operator at the boundary of the medium surface is M. According to the analysis of the propagation path of light between layers, the diffuse reflection light intensity distribution of the measurement surface is calculated and expressed as:

[0035] The pixel network output is obtained according to the deep network calculation, such as Figure 2 As shown;

[0036] S3, collect the diffuse reflection light intensity distribution of the patient's skin surface and perform preprocessing; use the previously trained deep convolutional network and the patient's diffuse reflection light intensity to perform imaging calculations. Step S3 establishes a deep convolutional network based on the network for obtaining uniform tissue convolution kernel parameter calculations and formula three, wherein the light transmission convolution kernel part indicated by the label is a set of light transmission convolution kernels of different tissues obtained in step S2, and the rest of the convolution network is a modifiable part, which randomly generates initial values. In the formula, Mki,n(x,y) represents the mask corresponding to the nth tissue characteristic at the i-th layer, and its value is 0 or 1, indicating whether the tissue at the coordinate (x,y) is the nth tissue;

[0037] Formula 3

[0038] Obtain the diffuse reflection light intensity distribution on the patient's skin surface, and perform error detection on the network output result of step S2, such as Figure 3 shown.

[0039] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0040] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calculating the distribution of optical parameters of skin tissue, Features: The following steps are involved: S1. Use biophotonics Monte Carlo simulation to calculate the diffuse reflection intensity distribution of the optical tissue surface of a single property, and use this result to train the corresponding convolution kernel for imaging calculation; S2. Use the digital phantom simulation calculation containing different optical parameters of tissues at different positions or the diffuse reflection two-dimensional light intensity distribution obtained by the real phantom and measurement system, as well as the convolution kernel of the light propagation calculation of single-property tissue obtained previously, to build and train a deep convolutional network for imaging calculation; S3, collect the diffuse reflection light intensity distribution of the patient's skin surface and perform preprocessing; use the previously trained deep convolutional network and the patient's diffuse reflection light intensity for imaging calculation; The step S1 obtains the diffuse reflection intensity distribution of a uniform tissue with known optical parameters under specific light intensity input conditions through a detection system, or uses an equivalent setting of the detection system to perform biophotonics Monte Carlo simulation, obtain the diffuse reflection intensity distribution, establish a convolutional network, and use the obtained diffuse reflection light intensity distribution to train a convolution kernel that reflects the propagation law of light in the tissue with corresponding optical parameters according to the convolutional network error back propagation rule.

Images