Apparatus for collagen evaluation and prognostic prediction of colorectal cancer and storage medium

1. A non-transitory computer readable storage medium storing a computer program, wherein the computer program comprises program instructions that, when executed by a processor, cause the processor to perform following steps:

acquiring a multi-photon imaging image of a target imaging region in a histopathological section, and performing image processing on the multi-photon imaging image to partition the multi-photon imaging image into a center of tumor (CT) region, an invasive margin (IM) region and a normal tissue (N) region;

extracting a value of each collagen feature parameter in each region, and calculating scoring parameters according to the value of each collagen feature parameter, wherein the scoring parameters comprise a mean of each collagen feature parameter in each region, an inter-region difference value and a variation value, wherein the inter-region difference value is a difference between means of each collagen feature parameter for any two regions, and the variation value is a ratio of two inter-region difference values; and the mean of each collagen feature parameter is an average value of a plurality of values of the collagen feature parameter corresponding to a plurality of subregions in a same region of three partitioned regions from the multi-photon imaging image; and

selecting the scoring parameters corresponding to target collagen features for collagen feature score calculation, thereby obtaining a collagen feature score of the histopathological section.

2. The non-transitory computer readable storage medium according to claim 1 , wherein

the performing image processing on the multi-photon imaging image to partition the multi-photon imaging image into a CT region, an IM region and an N region comprises:

extracting a boundary line between a tumor and a normal tissue, and determining a region containing the tumor as an initial CT region and a region containing the normal tissue as an initial N region;

translating the boundary line toward the initial CT region by a first distance to form a first boundary line, and translating the boundary line toward the initial N region by a second distance to form a second boundary line;

determining a region between the first boundary line and the second boundary line as the IM region; and

determining a region in the initial CT region other than the IM region as the CT region and a region in the initial N region other than the IM region as the N region.

3. The non-transitory computer readable storage medium according to claim 1 , wherein

the extracting a value of each collagen feature parameter in each region, and calculating scoring parameters according to the value of each collagen feature parameter comprises:

randomly selecting three CT subregions ROI1, ROI2 and ROI3 in the CT region, and calculating an average value of values of each collagen feature parameter for the three CT subregions as a mean of the corresponding collagen feature parameter of the CT region;

randomly selecting three IM subregions ROI4, ROI5 and ROI6 in the IM region, and calculating an average value of values of each collagen feature parameters for the three IM subregions as a mean of the corresponding collagen feature parameter of the IM region; and

randomly selecting three N subregions ROI7, ROI8 and ROI9 in the N region, and calculating an average value of values of each collagen feature parameter for the three N subregions as a mean of the corresponding collagen feature parameter of the N region.

4. The non-transitory computer readable storage medium according to claim 3 , wherein

the extracting a value of each collagen feature parameter in each region, and calculating scoring parameters according to the value of each collagen feature parameter further comprise:

calculating a difference between means of each collagen feature parameter for the CT region and the IM region as a first inter-region difference value;

calculating a difference between means of each collagen feature parameter for the IM region and the N region as a second inter-region difference value;

calculating a difference between means of each collagen feature parameter for the CT region and the N region as a third inter-region difference value; and

deeming a ratio of the first inter-region difference value to the third inter-region difference value as an inter-region variation value.

5. The non-transitory computer readable storage medium according to claim 4 , wherein

the selecting the scoring parameters corresponding to target collagen features for collagen feature score calculation comprises:

establishing a collagen feature scoring model through a least absolute shrinkage and selection operator (LASSO) regression algorithm as follows:

Cf

score

=

∑

i

=

1

k

coef

i

·

collagen_feature

i

;

wherein Cf score represents the collagen feature score; collagen_feature i represents a scoring parameter corresponding to an ith target collagen feature; coef i represents a calculation coefficient corresponding to the parameter collagen_feature i ; and k represents a number of the target collagen features.

6. The non-transitory computer readable storage medium according to claim 5 , wherein

k=16;

collagen_feature 1 represents a fiber area of the IM region, and coef 1 is 0.181;

collagen_feature 2 represents a fiber number of the IM region, and coef 2 is 0.081;

collagen_feature 3 represents a fiber length of the IM region, and coef 3 is 0.303;

collagen_feature 4 represents a fiber orientation of the N region, and coef 4 is −0.515;

collagen_feature 5 represents the first inter-region difference value of a fiber crosslink density, and coef 5 is −0.126;

collagen_feature 6 represents a correlation for a pixel displacement of 3 and in an angle of 0° in a gray-level co-occurrence matrix of the CT region, and coef 6 is 0.018;

collagen_feature 7 represents a mean of image convolution of a Gabor filter with a scale of 3 and an angle of 60° for the CT region, and coef 7 is 0.477;

collagen_feature 8 represents a mean of image convolution of a Gabor filter with a scale of 1 and an angle of 60° for the IM region, and coef 8 is −0.119;

collagen_feature 9 represents a mean of image convolution of a Gabor filter with a scale of 3 and an angle of 60° for the IM region, and coef 9 is 0.232;

collagen_feature 10 represents a third inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 3 and an angle of 60°, and coef 10 is 0.016;

collagen_feature 11 represents a first inter-region difference value of a contrast for a pixel displacement of 3 and in an angle of 45° in a gray-level co-occurrence matrix, and coef 11 is 0.141;

collagen_feature 12 represents a first inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 2 and an angle of 90°, and coef 12 is 0.048;

collagen_feature 13 represents the first inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 4 and an angle of 90°, and coef 13 is 0.12;

collagen_feature 14 represents an inter-region variation value of a correlation for a pixel displacement of 1 and in an angle of 90° in a gray-level co-occurrence matrix, and coef 14 is 0.24;

collagen_feature 15 represents an inter-region variation value of a correlation for a pixel displacement of 2 and in an angle of 135° in a gray-level co-occurrence matrix, and coef 15 is 0.145; and

collagen_feature 16 represents an inter-region variation value of a correlation for a pixel displacement of 4 and in an angle of 0° in a gray-level co-occurrence matrix, and coef 16 is 0.303.

7. An apparatus for collagen evaluation and prognostic prediction of colorectal cancer, comprising a processor, an input device, an output device and a memory that are connected to one another, wherein the memory comprises the non-transitory computer readable storage medium according to claim 1 , and the processor is configured to call the program instructions.

8. The apparatus according to claim 7 , wherein

the performing image processing on the multi-photon imaging image to partition the multi-photon imaging image into a CT region, an IM region and an N region comprises:

extracting a boundary line between a tumor and a normal tissue, and determining a region containing the tumor as an initial CT region and a region containing the normal tissue as an initial N region;

translating the boundary line toward the initial CT region by a first distance to form a first boundary line, and translating the boundary line toward the initial N region by a second distance to form a second boundary line;

determining a region between the first boundary line and the second boundary line as the IM region; and

determining a region in the initial CT region other than the IM region as the CT region and a region in the initial N region other than the IM region as the N region.

9. The apparatus according to claim 7 , wherein

the extracting a value of each collagen feature parameter in each region, and calculating scoring parameters according to the value of each collagen feature parameter comprises:

randomly selecting three CT subregions ROI1, ROI2 and ROI3 in the CT region, and calculating an average value of values of each collagen feature parameter for the three CT subregions as a mean of the corresponding collagen feature parameter of the CT region;

randomly selecting three IM subregions ROI4, ROI5 and ROI6 in the IM region, and calculating an average value of values of each collagen feature parameters for the three IM subregions as a mean of the corresponding collagen feature parameter of the IM region; and

randomly selecting three N subregions ROI7, ROI8 and ROI9 in the N region, and calculating an average value of values of each collagen feature parameter for the three N subregions as a mean of the corresponding collagen feature parameter of the N region.

10. The apparatus according to claim 9 , wherein

the extracting a value of each collagen feature parameter in each region, and calculating scoring parameters according to the value of each collagen feature parameter further comprise:

calculating a difference between means of each collagen feature parameter for the CT region and the IM region as a first inter-region difference value;

calculating a difference between means of each collagen feature parameter for the IM region and the N region as a second inter-region difference value;

calculating a difference between means of each collagen feature parameter for the CT region and the N region as a third inter-region difference value; and

deeming a ratio of the first inter-region difference value to the third inter-region difference value as an inter-region variation value.

11. The apparatus according to claim 10 , wherein

the selecting the scoring parameters corresponding to target collagen features for collagen feature score calculation comprises:

establishing a collagen feature scoring model through a least absolute shrinkage and selection operator (LASSO) regression algorithm as follows:

Cf

score

=

∑

i

=

1

k

coef

i

·

collagen_feature

i

;

wherein Cf score represents the collagen feature score; collagen_feature i represents a scoring parameter corresponding to an ith target collagen feature; coef i represents a calculation coefficient corresponding to the parameter collagen_feature i ; and k represents a number of the target collagen features.

12. The apparatus according to claim 11 , wherein

k=16;

collagen_feature 1 represents a fiber area of the IM region, and coef 1 is 0.181;

collagen_feature 2 represents a fiber number of the IM region, and coef 2 is 0.081;

collagen_feature 3 represents a fiber length of the IM region, and coef 3 is 0.303;

collagen_feature 4 represents a fiber orientation of the N region, and coef 4 is −0.515;

collagen_feature 5 represents the first inter-region difference value of a fiber crosslink density, and coef 5 is −0.126;

collagen_feature 6 represents a correlation for a pixel displacement of 3 and in an angle of 0° in a gray-level co-occurrence matrix of the CT region, and coef 6 is 0.018;

collagen_feature 7 represents a mean of image convolution of a Gabor filter with a scale of 3 and an angle of 60° for the CT region, and coef 7 is 0.477;

collagen_feature 8 represents a mean of image convolution of a Gabor filter with a scale of 1 and an angle of 60° for the IM region, and coef 8 is −0.119;

collagen_feature 9 represents a mean of image convolution of a Gabor filter with a scale of 3 and an angle of 60° for the IM region, and coef 9 is 0.232;

collagen_feature 10 represents a third inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 3 and an angle of 60°, and coef 10 is 0.016;

collagen_feature 11 represents a first inter-region difference value of a contrast for a pixel displacement of 3 and in an angle of 45° in a gray-level co-occurrence matrix, and coef 11 is 0.141;

collagen_feature 12 represents a first inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 2 and an angle of 90°, and coef 12 is 0.048;

collagen_feature 13 represents the first inter-region difference value of a variance of image convolution of a Gabor filter with a scale of 4 and an angle of 90°, and coef 13 is 0.12;

collagen_feature 14 represents an inter-region variation value of a correlation for a pixel displacement of 1 and in an angle of 90° in a gray-level co-occurrence matrix, and coef 14 is 0.24;

collagen_feature 15 represents an inter-region variation value of a correlation for a pixel displacement of 2 and in an angle of 135° in a gray-level co-occurrence matrix, and coef 15 is 0.145; and

collagen_feature 16 represents an inter-region variation value of a correlation for a pixel displacement of 4 and in an angle of 0° in a gray-level co-occurrence matrix, and coef 16 is 0.303.