A method and system for fine-grained classification of cardiomyocytes

Abstract

The present invention provides a method and system for fine-grained myocardial segmentation. The method for fine-grained myocardial segmentation includes: obtaining an image to be processed and performing image segmentation to obtain a left heart cavity segmentation map, a left ventricle segmentation map, and a right ventricle segmentation map; S2. Fitting the left ventricle segmentation map and the right ventricle segmentation map to obtain a left ventricle ellipse and two intersection points between the left ventricle and the right ventricle; S3. Obtaining a left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricle segmentation map; S4. Calculating a dividing line according to the left ventricle ellipse and the two intersection points; S5. Dividing the left myocardial segmentation map according to the dividing line to obtain a segmentation map of the fine-grained myocardial region. The method and system for fine-grained myocardial segmentation enable a machine to calculate a dividing line according to the left ventricle ellipse and the two intersection points for fine-grained myocardial segmentation, facilitating automated computer-aided diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of myocardial tissue examination, and particularly to a method and system for fine-grained myocardial division. Background Art

[0002] T1 mapping technology is a good non-invasive myocardial tissue examination method, which can play a very good role in the diagnosis, examination and prognosis of various cardiovascular diseases. In order to accurately describe the regions with abnormal values in cardiac T1 mapping images, the myocardium needs to be divided into multiple regions. The 16-segment myocardial division method of the American Heart Association divides the heart into three equal parts along the long axis of the heart, namely the apical part, the mid part, and the basal part (apical, mid, basal). Then, with the long axis as the center, the apical part is equally divided into 4 regions, and the mid part and the basal part are equally divided into 6 regions, totaling 16 regions.

[0003] The 16-segment myocardial division method of the American Heart Association can generally be used for cardiac MRI short-axis images, but since it does not give a method for determining the left and right ventricular junction points, it is more suitable for manual drawing. Summary of the Invention

[0004] In view of this, the technical problem to be solved by the present invention is to provide a method and system for fine-grained myocardial division, which can facilitate the machine to perform fine-grained myocardial division.

[0005] The technical solution of the present invention is realized as follows:

[0006] A method for fine-grained myocardial division includes the following steps:

[0007] S1. Obtain the image to be processed, and perform image segmentation to obtain a left heart cavity segmentation map, a left ventricle segmentation map, and a right ventricle segmentation map;

[0008] S2. Fit the left ventricle segmentation map and the right ventricle segmentation map to obtain a left ventricle ellipse and two junction points between the left ventricle and the right ventricle;

[0009] S3. Obtain a left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricle segmentation map;

[0010] S4. Calculate the dividing line according to the left ventricle ellipse and the two junction points;

[0011] S5. Divide the left myocardial segmentation map according to the dividing line to obtain a segmentation map of fine-grained myocardial regions.

[0012] Preferably, the image to be processed is a T1 mapping image.

[0013] Preferably, step S2 specifically includes:

[0014] S21. Fit the left ventricular segmentation map and the right ventricular segmentation map with elliptic curves respectively to obtain an ellipse e covering the left ventricle and an ellipse covering the right ventricle.

[0015] S22. Obtain the intersection points of the two ellipses, where the upper intersection point is A and the lower intersection point is B.

[0016] Preferably, the specific steps of S3 include:

[0017] Subtract the overlapping part of the left ventricular segmentation map and the left heart cavity segmentation map from the left ventricular segmentation map to obtain the left myocardial segmentation map.

[0018] Preferably, the specific steps of S4 include:

[0019] S41. Connect the intersection points A and B to obtain the line segment AB.

[0020] S42. Draw the perpendicular bisector of AB to intersect the ellipse e at points C and D.

[0021] S43. Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e.

[0022] S44. Draw a straight line l through AE and use l as the starting division line for the fine-grained division of the myocardium.

[0023] S45. Rotate the previous division line by n° around point E to obtain a new division line; n is a factor of 60.

[0024] S46. Repeat S45 to obtain 180 / n division lines.

[0025] S47. Save the 180 / n division lines.

[0026] Preferably, the specific steps of S5 include:

[0027] Divide the left myocardial segmentation map into 360 / n fine-grained regions according to the 180 / n division lines to obtain the segmentation map of the myocardial fine-grained regions.

[0028] An embodiment of the present invention also proposes a myocardial fine-grained division system, including:

[0029] A segmentation module, configured to obtain an image to be processed, perform image segmentation to obtain a left heart cavity segmentation map, a left ventricular segmentation map, and a right ventricular segmentation map, and send them to the positioning module and the division module;

[0030] The positioning module is configured to fit the left ventricular segmentation map and the right ventricular segmentation map to obtain a left ventricular ellipse and two intersection points of the left ventricle and the right ventricle;

[0031] The partitioning module is configured to obtain a left myocardial segmentation map based on the left heart chamber segmentation map and the left ventricle segmentation map; calculate a partitioning line according to the left ventricle ellipse and the two junction points; and partition the left myocardial segmentation map according to the partitioning line to obtain a segmentation map of the myocardial fine-grained regions.

[0032] Preferably, the positioning module specifically includes:

[0033] A fitting unit, configured to respectively fit the left ventricle segmentation map and the right ventricle segmentation map through elliptic curves to obtain an ellipse e covering the left ventricle and an ellipse covering the right ventricle;

[0034] A naming unit, configured to name the junction points of the two ellipses, where the upper junction point is A and the lower junction point is B.

[0035] Preferably, the partitioning module is specifically configured to:

[0036] Subtract the overlapping part of the left ventricle segmentation map and the left heart chamber segmentation map from the left ventricle segmentation map to obtain the left myocardial segmentation map.

[0037] Preferably, the partitioning module is specifically configured to:

[0038] Connect the junction points A and B to obtain a line segment AB;

[0039] Construct the perpendicular bisector of AB to intersect the ellipse e at points C and D;

[0040] Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e;

[0041] Draw a straight line l through AE, and use l as the starting partitioning line for myocardial fine-grained partitioning;

[0042] Rotate the previous partitioning line by n° around point E to obtain a new partitioning line; n is a factor of 60;

[0043] Repeat rotating the previous partitioning line by n° around point E to obtain a new partitioning line, and obtain 180 / n partitioning lines;

[0044] Save the 180 / n partitioning lines;

[0045] Partition the left myocardial segmentation map into 360 / n fine-grained regions according to the 180 / n partitioning lines to obtain a segmentation map of the myocardial fine-grained regions.

[0046] The myocardial fine-grained division method and system proposed by the present invention can obtain the left ventricular ellipse and two junction points between the left ventricle and the right ventricle by fitting the left ventricular segmentation map and the right ventricular segmentation map, and can obtain the left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricular segmentation map; thus, the machine can calculate the division line according to the left ventricular ellipse and the two junction points for myocardial fine-grained division, which is convenient for automated computer-aided diagnosis. Description of the Drawings

[0047] Figure 1 It is a flowchart of the myocardial fine-grained division method proposed by an embodiment of the present invention;

[0048] Figure 2 It is a structural block diagram of the myocardial fine-grained division system proposed by an embodiment of the present invention;

[0049] Figure 3 It is a myocardial fine-grained region map obtained by an embodiment of the present invention. Detailed Embodiments

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] As Figure 1 shown, an embodiment of the present invention proposes a myocardial fine-grained division method, including the following steps:

[0052] A myocardial fine-grained division method, including the following steps:

[0053] S1. Obtain the image to be processed, and perform image segmentation to obtain the left heart cavity segmentation map, the left ventricular segmentation map, and the right ventricular segmentation map;

[0054] S2. Fit the left ventricular segmentation map and the right ventricular segmentation map to obtain the left ventricular ellipse and two junction points between the left ventricle and the right ventricle;

[0055] S3. Obtain the left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricular segmentation map;

[0056] S4. Calculate the division line according to the left ventricular ellipse and the two junction points;

[0057] S5. Divide the left myocardial segmentation map according to the division line to obtain the segmentation map of the myocardial fine-grained region.

[0058] It can be seen that the myocardial fine-grained division method proposed in the embodiment of the present invention can obtain the left ventricular ellipse and two intersection points between the left ventricle and the right ventricle by fitting the left ventricular segmentation map and the right ventricular segmentation map, and can obtain the left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricular segmentation map; thus, the machine can calculate the division line according to the left ventricular ellipse and the two intersection points for myocardial fine-grained division, which is convenient for automated computer-aided diagnosis.

[0059] In a preferred embodiment of the present invention, the image to be processed is a T1 mapping image.

[0060] T1 mapping measures the T1 value of tissues and has been mainly used in cardiac scans recently. Currently, T1 mapping is the main technology for quantitatively evaluating myocardial lesions.

[0061] In a preferred embodiment of the present invention, S2 specifically includes:

[0062] S21: Fit the left ventricular segmentation map and the right ventricular segmentation map with elliptic curves respectively to obtain an ellipse e covering the left ventricle and an ellipse covering the right ventricle;

[0063] S22: Obtain the intersection points of the two ellipses, where the upper intersection point is A and the lower intersection point is B.

[0064] In a preferred embodiment of the present invention, S3 specifically includes:

[0065] Subtract the overlapping part of the left ventricular segmentation map from the left heart cavity segmentation map to obtain the left myocardial segmentation map.

[0066] In a preferred embodiment of the present invention, S4 specifically includes:

[0067] S41: Connect the intersection points A and B to obtain the line segment AB;

[0068] S42: Make the perpendicular bisector of AB intersect the ellipse e at points C and D;

[0069] S43: Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e;

[0070] S44: Draw a straight line l through AE and use l as the starting division line for myocardial fine-grained division;

[0071] S45: Rotate the previous division line by n° around point E to obtain a new division line; n is a factor of 60;

[0072] S46: Repeat S45 to obtain 180 / n division lines;

[0073] S47: Save 180 / n division lines.

[0074] The left myocardial segmentation map is divided into 360 / n fine-grained regions according to 180 / n dividing lines, and a segmentation map of the myocardial fine-grained regions is obtained.

[0075] Taking n = 10 as an example:

[0076] Connect the intersection points A and B to obtain the line segment AB;

[0077] Construct the perpendicular bisector of AB to intersect the ellipse e at points C and D;

[0078] Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e;

[0079] Draw a straight line l through AE and use l as the starting dividing line for the fine-grained division of the myocardium;

[0080] Rotate the previous dividing line by 10° around point E to obtain a new dividing line;

[0081] Repeat the above steps to obtain 18 dividing lines;

[0082] Taking l as the starting line, two adjacent dividing lines in the clockwise direction cut out 2 fine-grained regions with a central angle equal to 10° from the left myocardial segmentation map;

[0083] Repeat the above steps until a total of 36 fine-grained regions are obtained;

[0084] Output the segmentation map of the myocardial fine-grained regions.

[0085] The myocardial fine-grained regions obtained through this embodiment are named as Figure 3 shown. Taking l as the starting dividing line, the regions are numbered in sequence in the clockwise direction. The corresponding relationship between the myocardial fine-grained regions and the regions of the 16-segment myocardial division method is: the 1st to 6th regions correspond to segment 7 region; the 7th to 12th regions correspond to segment 12; the 13th to 18th regions correspond to segment 11; the 19th to 24th regions correspond to segment 10; the 25th to 30th regions correspond to segment 9; the 31st to 36th regions correspond to segment 8.

[0086] The present invention locates the intersection points of the left ventricle and the right ventricle in the T1 mapping image based on the image segmentation method, and realizes the fine-grained division of the myocardium based on this. Compared with the traditional myocardial division method, the present invention divides the left ventricular myocardial image into 36 fine-grained regions, which is beneficial to discovering and describing the positions where the T1 value in the myocardium is abnormal. The naming of the regions is as Figure 3As shown, the inner circle of the figure represents segments 7 - 12 of the 16 - segment myocardial division method (AHA division method), and the outer circle represents the 36 regions of the fine - grained division method in this paper, namely No.1 - No.36. For example, segment 7 corresponds to regions 1 to 6 of the fine - grained division method.

[0087] In the embodiments of the present invention, n can be a factor of 60. Currently, the AHA - recommended 16 - segment myocardial division method takes 60. However, the segment with a central angle of 60 has too large an area in numerical analysis, which makes the characteristic values deviating from the normal value easily diluted in numerical analysis. Therefore, a number that is small enough compared to 60 and divisible by 360 needs to be selected as the new central angle.

[0088] The reason for taking 10 is that it can also simplify the program. Because when n = 10, the two dividing lines symmetric with respect to point E are on the same straight line. In this way, instead of calculating the parameters of 36 dividing lines, only the parameters of 18 straight lines need to be calculated. In addition, taking 10 can also make the fine - grained division results more simply compatible with the 16 - segment myocardial division method and can be directly mapped to segment x. The above also holds for the factors of 60.

[0089] Since the left - myocardial segmentation map is obtained by subtracting the overlapping part of the left - ventricular segmentation map and the left - heart - cavity segmentation map from the left - ventricular segmentation map, after positioning and drawing the dividing lines on the left - ventricular segmentation map, it can be directly applied to the left - myocardial segmentation map.

[0090] As Figure 2 shown, the embodiments of the present invention also propose a myocardial fine - grained division system, including:

[0091] A segmentation module 1, configured to obtain an image to be processed, perform image segmentation to obtain a left - heart - cavity segmentation map, a left - ventricular segmentation map, and a right - ventricular segmentation map, and send them to a positioning module 2 and a division module 3;

[0092] A positioning module 2, configured to fit the left - ventricular segmentation map and the right - ventricular segmentation map to obtain a left - ventricular ellipse and two intersection points between the left ventricle and the right ventricle;

[0093] A division module 3, configured to obtain a left - myocardial segmentation map according to the left - heart - cavity segmentation map and the left - ventricular segmentation map; calculate dividing lines according to the left - ventricular ellipse and the two intersection points; and divide the left - myocardial segmentation map according to the dividing lines to obtain a segmentation map of myocardial fine - grained regions.

[0094] In a preferred embodiment of the present invention, the positioning module 2 specifically includes:

[0095] A fitting unit 201, configured to respectively fit the left - ventricular segmentation map and the right - ventricular segmentation map through an elliptic curve to obtain an ellipse e covering the left ventricle and an ellipse covering the right ventricle;

[0096] A naming unit 202 for naming the intersection points of two ellipses, where the upper intersection point is A and the lower intersection point is B.

[0097] In a preferred embodiment of the present invention, the partitioning module 3 is specifically configured to:

[0098] Subtract the overlapping part of the left ventricular segmentation map from the left heart cavity segmentation map to obtain the left myocardial segmentation map.

[0099] In a preferred embodiment of the present invention, the partitioning module is specifically configured to:

[0100] Connect the intersection points A and B to obtain the line segment AB;

[0101] Construct the perpendicular bisector of AB to intersect the ellipse e at points C and D;

[0102] Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e;

[0103] Draw a straight line l through AE and use l as the starting partitioning line for the fine-grained partitioning of the myocardium;

[0104] Rotate the previous partitioning line by n° around point E to obtain a new partitioning line; n is a factor of 60;

[0105] Repeat rotating the previous partitioning line by n° around point E to obtain a new partitioning line, and obtain 180 / n partitioning lines;

[0106] Save the 180 / n partitioning lines;

[0107] According to multiple partitioning lines, divide the left myocardial segmentation map into 360 / n fine-grained regions to obtain the segmentation map of the myocardial fine-grained regions.

[0108] The method and system for fine-grained partitioning of the myocardium proposed by the present invention can obtain the left ventricular ellipse and two intersection points of the left ventricle and the right ventricle by fitting the left ventricular segmentation map and the right ventricular segmentation map, and can obtain the left myocardial segmentation map according to the left heart cavity segmentation map and the left ventricular segmentation map; thus, the machine can calculate the partitioning lines based on the left ventricular ellipse and two intersection points for fine-grained partitioning of the myocardium, which is convenient for automated computer-aided diagnosis.

[0109] Finally, it should be noted that: The above are only preferred embodiments of the present invention, which are only used to illustrate the technical solutions of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention are included in the protection scope of the present invention.

Claims

1. A method for fine-grained division of myocardium, characterized in that, it includes the following steps: S1. Obtain the image to be processed, and perform image segmentation to obtain the left heart cavity segmentation map, the left ventricle segmentation map, and the right ventricle segmentation map; S2. Fit the left ventricle segmentation map and the right ventricle segmentation map to obtain the left ventricle ellipse and two junction points between the left ventricle and the right ventricle; S3. Obtain the left myocardium segmentation map according to the left heart cavity segmentation map and the left ventricle segmentation map; S4. Calculate the division line according to the left ventricle ellipse and the two junction points; S5. Divide the left myocardium segmentation map according to the division line to obtain the segmentation map of the fine-grained region of the myocardium.

2. The method for fine-grained division of myocardium according to claim 1, characterized in that, the image to be processed is a T1mapping image.

3. The method for fine-grained division of myocardium according to claim 1, characterized in that, the specific content of S2 includes: S21. Fit the left ventricle segmentation map and the right ventricle segmentation map respectively with an elliptic curve to obtain the ellipse e covering the left ventricle and the ellipse covering the right ventricle; S22. Obtain the junction points of the two ellipses, where the upper junction point is A and the lower junction point is B.

4. The method for fine-grained division of myocardium according to claim 1, characterized in that, the specific content of S3 includes: Subtract the overlapping part of the left ventricle segmentation map and the left heart cavity segmentation map to obtain the left myocardium segmentation map.

5. The method for fine-grained division of myocardium according to claim 3, characterized in that, the specific content of S4 includes: S41. Connect the junction points A and B to obtain the line segment AB; S42: Make the perpendicular bisector of AB intersect the ellipse e at points C and D; S43: Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e; S44: Draw a straight line l through AE, and use l as the starting division line for the fine-grained division of the myocardium; S45: Rotate the previous division line by n° around point E to obtain a new division line; n is a factor of 60; S46: Repeat S45 to obtain 180 / n division lines; S47: Save the 180 / n division lines.

6. The method for fine-grained division of myocardium according to claim 5, characterized in that, the specific content of S5 includes: Divide the left myocardium segmentation map into 360 / n fine-grained regions according to the 180 / n division lines to obtain the segmentation map of the fine-grained region of the myocardium.

7. A system for fine-grained division of myocardium, characterized in that, it includes: A segmentation module, which is used to obtain the image to be processed, perform image segmentation to obtain the left heart cavity segmentation map, the left ventricle segmentation map, and the right ventricle segmentation map, and send them to the positioning module and the division module; The positioning module, which is used to fit the left ventricle segmentation map and the right ventricle segmentation map to obtain the left ventricle ellipse and two junction points between the left ventricle and the right ventricle; The division module, which is used to obtain the left myocardium segmentation map according to the left heart cavity segmentation map and the left ventricle segmentation map; calculate the division line according to the left ventricle ellipse and the two junction points; divide the left myocardium segmentation map according to the division line to obtain the segmentation map of the fine-grained region of the myocardium.

8. The system for fine-grained division of myocardium according to claim 7, It is characterized in that the positioning module specifically includes: a fitting unit, configured to respectively fit the left ventricle segmentation map and the right ventricle segmentation map through an elliptic curve to obtain an ellipse e covering the left ventricle and an ellipse covering the right ventricle; a naming unit, configured to name the intersection points of the two ellipses, where the upper intersection point is A and the lower intersection point is B.

9. The myocardial fine-grained division system according to claim 7, It is characterized in that the division module is specifically configured to: Subtract the overlapping part of the left ventricle segmentation map and the left heart cavity segmentation map from the left ventricle segmentation map to obtain a left myocardial segmentation map.

10. The myocardial fine-grained division system according to claim 8, It is characterized in that the division module is specifically configured to: Connect the intersection points A and B to obtain a line segment AB; Construct the perpendicular bisector of AB to intersect the ellipse e at points C and D; Take the midpoint E of the line segment CD, and this midpoint E is the center point of the ellipse e; Draw a straight line l through AE, and use l as the starting division line for myocardial fine-grained division; Rotate the previous division line by n° around point E to obtain a new division line; n is a factor of 60; Repeat rotating the previous division line by n° around point E to obtain a new division line, and obtain 180 / n division lines; Save 180 / n division lines; According to the 180 / n division lines, divide the left myocardial segmentation map into 360 / n fine-grained regions to obtain a segmentation map of myocardial fine-grained regions.

Images