A cardiac dose prediction system and method based on heart-chest Euclidean distance

Abstract

The present invention relates to the field of cardiac dose prediction technology, and in particular to a cardiac dose prediction system and method based on the Euclidean distance between the heart and chest. The system comprises: an information identification module, an information reconstruction module, and a prediction module; the information identification module is used to identify the CT image to be tested and obtain the cardiac contour and the chest wall target area contour; the information reconstruction module is used to construct the cardiac contour chest wall surface and the chest wall contour heart surface based on the cardiac contour and the chest wall target area contour based on the SciPy software package, and calculate the cardiac Euclidean distance based on the cardiac contour chest wall surface and the chest wall contour heart surface; the prediction module is used to input the cardiac Euclidean distance into a preset prediction model and output the cardiac dose, wherein the prediction model is constructed based on the correlation between the cardiac Euclidean distance and the average cardiac dose. The present invention can realize the automatic, rapid and accurate measurement of cardiac dose based on the cardiac Euclidean distance, and minimize the measurement error.

Description

Technical Field

[0001] The present invention relates to the technical field of cardiac dose prediction, and in particular to a cardiac dose prediction system and method based on the heart-chest Euclidean distance. Background Art

[0002] While postoperative radiotherapy for breast cancer brings survival benefits to patients, it also causes radiation-induced cardiac damage, leading to an increased risk of cardiac-related adverse events such as ischemic heart disease, valvular disease, arrhythmia, pericardial effusion, and congestive heart failure, especially in patients with left-sided breast cancer. Cardiac-related adverse events have become the main reason threatening the long-term survival of patients with early breast cancer.

[0003] Previous studies have shown that the mean heart dose (MHD) is linearly correlated with the risk of major coronary events (myocardial infarction, coronary artery revascularization, and death from ischemic heart disease), that is, for every 1 Gy increase in MHD, the risk of major coronary events increases by 7.4%. Understanding the patient's heart radiation dose can help clinicians determine the risk of radiation-induced heart damage in patients, and then conduct heart protection interventions and close monitoring of heart damage. However, in clinical practice, to determine the specific heart radiation dose of the patient, it is necessary to complete the steps of position fixation, positioning CT scanning, target area and organ at risk delineation, and plan design, which consumes a lot of time and human and material resources. Therefore, it is urgent to predict the heart as soon as possible through prediction tools. Previous studies have shown that anatomical indicators based on positioning CT images can be used to predict the patient's heart radiation dose, such as heart-chest distance, heart contact distance, and maximum heart distance. However, these indicators are all based on the technical background of three-dimensional conformal radiotherapy based on tangent fields, and they are all two-dimensional indicators measured on a certain CT level, which is difficult to well reflect the three-dimensional spatial position between the heart and the target area. In addition, these indicators are measured manually, and their accuracy is easily affected by the tangent field angle, CT layer thickness, and measurement error of the measurer. There is currently no consistent anatomical indicator recommendation. Intensity modulation radiated therapy (IMRT) has been widely used in postoperative radiotherapy for breast cancer patients because it can improve the dose uniformity of the target area while reducing the dose to organs at risk. However, there are currently few screening indicators for the use of IMRT technology.

[0004] Therefore, there is an urgent need for a cardiac dose prediction system and method based on the heart-chest Euclidean distance. Summary of the invention

[0005] The purpose of the present invention is to provide a cardiac dose prediction system and method based on the heart-chest Euclidean distance, so as to solve the problem that there is currently no cardiac dose prediction index for left breast cancer patients using IMRT, and to achieve rapid cardiac dose prediction.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A cardiac dose prediction system based on the Euclidean distance between the heart and chest, comprising:

[0008] Information recognition module, information reconstruction module, prediction module;

[0009] The information recognition module is used to recognize the CT image to be tested and obtain the heart contour and chest wall target area contour;

[0010] The information reconstruction module is used to construct a heart contour chest wall surface and a chest wall contour heart surface from the heart contour and the chest wall target area contour based on the SciPy software package, and calculate the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface;

[0011] The prediction module is used to input the heart-chest Euclidean distance into a preset prediction model to output a heart dose, wherein the prediction model is constructed based on the correlation between the heart-chest Euclidean distance and the average heart dose.

[0012] Optionally, the information reconstruction module constructs a chest wall surface of the heart contour and a heart surface of the chest wall contour from the heart contour and the chest wall target area contour based on the SciPy software package, including:

[0013] Determining a starting CT layer and an ending CT layer of the heart contour;

[0014] On the same CT layer within the range from the starting CT layer to the ending CT layer, respectively obtaining a first connecting line between the outer boundary of the chest wall target area contour and the dorsal edge of the heart contour, and a second connecting line between the inner boundary of the chest wall target area contour and the ventral edge of the heart contour;

[0015] The heart contour close to the chest wall target area between the intersection points of the first connecting line, the second connecting line and the heart contour is the heart contour chest wall surface, and the chest wall target area contour close to the heart between the intersection points of the first connecting line, the second connecting line and the chest wall target area contour is the chest wall contour heart surface.

[0016] Optionally, the information reconstruction module calculates the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface, including:

[0017] Acquire voxel points on the chest wall surface of the heart contour and construct a set Ω;

[0018] Acquire voxel points on the cardiac surface of the chest wall contour and construct a set Ω′;

[0019] The shortest Euclidean distance between the set Ω and the set Ω′ is calculated, which is the heart-chest Euclidean distance.

[0020] Optionally, the calculation formula of the heart-chest Euclidean distance is:

[0021]

[0022]

[0023] Among them, point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j ).

[0024] Optionally, the prediction model in the prediction module is:

[0025] y=-19.57x+476.60

[0026] Where x represents the Euclidean distance between the heart and chest, and y represents the average heart dose.

[0027] To further achieve the above object, the present invention also provides a method for predicting cardiac dose based on the heart-chest Euclidean distance, comprising:

[0028] Importing a CT image to be tested, and identifying a heart contour and a chest wall target area contour in the CT image to be tested;

[0029] Based on the SciPy software package, a heart contour chest wall surface and a chest wall contour heart surface are constructed from the heart contour and the chest wall target area contour, and the heart-chest Euclidean distance is calculated based on the heart contour chest wall surface and the chest wall contour heart surface;

[0030] The heart-chest Euclidean distance is input into a preset prediction model to output a heart dose, wherein the prediction model is constructed based on the correlation between the heart-chest Euclidean distance and the average heart dose.

[0031] Optionally, the heart-chest Euclidean distance is calculated as follows:

[0032] Acquire voxel points on the chest wall surface of the heart contour and construct a set Ω;

[0033] Acquire voxel points on the cardiac surface of the chest wall contour and construct a set Ω′;

[0034] The shortest Euclidean distance between the set Ω and the set Ω′ is calculated using the following formula:

[0035]

[0036]

[0037] Among them, point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j ).

[0038] Optionally, the prediction model is:

[0039] y=-19.57x+476.60

[0040] Where x represents the Euclidean distance between the heart and chest, and y represents the average heart dose.

[0041] The beneficial effects of the present invention are:

[0042] The present invention automatically identifies and locates the chest wall target area contour and the heart target area contour in the CT image, determines the upper and lower limits of the heart contour to clarify the CT plane for measurement and calculation; defines the range of the chest wall surface of the heart contour and the heart surface of the chest wall contour by connecting the outer limit of the chest wall target area with the dorsal edge of the heart contour, and the inner limit of the chest wall target area with the ventral edge of the heart contour; measures the Euclidean distance between many voxel points on the chest wall surface of the heart contour and the heart surface of the chest wall contour based on the theoretical basis of the Euclidean distance, and defines the shortest Euclidean distance as the heart-chest Euclidean distance. The anatomical index proposed by the present invention is obtained based on the measurement of the three-dimensional spatial position relationship between the heart and the chest wall target area. The measurement method realizes automated, rapid and accurate measurement, minimizes the error of the measurer, and reduces the burden on medical staff. Constructing a heart dose prediction model based on the heart-chest Euclidean distance can help clinicians quickly and accurately understand the heart exposure dose of patients with left breast cancer who are to receive adjuvant radiotherapy, so as to take intervention measures as early as possible and closely monitor and follow up. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 Schematic diagram of the structure of a cardiac dose prediction system based on the heart-chest Euclidean distance according to an embodiment of the present invention;

[0045] Figure 2 A flow chart of a cardiac dose prediction method based on the heart-chest Euclidean distance according to an embodiment of the present invention;

[0046] Figure 3A schematic diagram of a heart contour, a chest wall target contour, and a line connecting the two contours according to an embodiment of the present invention;

[0047] Figure 4 A linear correlation analysis diagram of the heart-chest Euclidean distance and the average heart dose according to an embodiment of the present invention;

[0048] Figure 5 This is a receiver operating curve of the cardiothoracic Euclidean distance for predicting the average cardiac dose according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] 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.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] This embodiment provides a cardiac dose prediction system based on the heart-chest Euclidean distance. Figure 1 As shown, it includes: information identification module, information reconstruction module and prediction module;

[0052] An information recognition module is used to recognize the CT image to be tested and obtain the heart contour and chest wall target area contour;

[0053] An information reconstruction module is used to construct a heart contour chest wall surface and a chest wall contour heart surface based on the heart contour and the chest wall target area contour based on the SciPy software package, and calculate the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface;

[0054] The prediction module is used to input the Euclidean distance between the heart and chest into a preset prediction model and output the heart dose, wherein the prediction model is constructed based on the correlation between the Euclidean distance between the heart and chest and the average heart dose.

[0055] Specifically, the information reconstruction module constructs the heart contour chest wall surface and the chest wall contour heart surface based on the heart contour and the chest wall target area contour based on the SciPy software package, including:

[0056] Determine the starting CT layer and the ending CT layer of the heart contour;

[0057] On the same CT layer within the range from the starting CT layer to the ending CT layer, respectively obtaining a first connecting line between the outer boundary of the chest wall target area contour and the dorsal edge of the heart contour, and a second connecting line between the inner boundary of the chest wall target area contour and the ventral edge of the heart contour;

[0058] The heart contour close to the chest wall target area between the intersection points of the first connecting line, the second connecting line and the heart contour is the heart contour chest wall surface, and the chest wall target area contour close to the heart between the intersection points of the first connecting line, the second connecting line and the chest wall target area contour is the chest wall contour heart surface.

[0059] Specifically, the information reconstruction module calculates the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface, including:

[0060] Obtain the voxel points on the chest wall of the heart contour and construct the set Ω;

[0061] Get the voxel points on the heart surface of the chest wall contour and construct the set Ω′;

[0062] Calculate the shortest Euclidean distance between set Ω and set Ω′, which is the heart-chest Euclidean distance.

[0063] The calculation formula of the heart-chest Euclidean distance is:

[0064]

[0065]

[0066] Among them, point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j ).

[0067] Specifically, the prediction model in the prediction module is:

[0068] y=-19.57x+476.60

[0069] Where x represents the Euclidean distance between the heart and chest, and y represents the average heart dose.

[0070] To further optimize the technical solution, this embodiment also provides a method for predicting cardiac dose based on the Euclidean distance between the heart and chest, such as Figure 2 As shown, including:

[0071] Step 1. Obtain a CT image to be tested, and identify the heart contour and chest wall target area contour in the CT image to be tested;

[0072] Import the CT image to be tested and perform intelligent detection to identify the heart contour and chest wall target contour, and determine the starting CT layer (upper boundary) and ending CT layer (lower boundary) of the heart contour.

[0073] Step 2. construct a heart contour chest wall surface and a chest wall contour heart surface based on the heart contour and the chest wall target area contour based on the SciPy software package;

[0074] like Figure 3 As shown, within the CT scanning range from the upper boundary to the lower boundary of the heart contour, the outer boundary of the chest wall target area contour and the dorsal edge of the heart contour, as well as the inner boundary of the chest wall target area and the ventral edge of the heart contour are connected on the same CT plane;

[0075] The heart contour close to the chest wall target area between the intersection points of the two connecting lines and the heart contour is the heart contour chest wall surface, and the chest wall target area contour close to the heart between the intersection points of the two connecting lines and the chest wall target area contour is the chest wall contour heart surface.

[0076] Step 3. Calculate the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface;

[0077] Calculate the vertical distance between the voxel points on the chest wall surface of the heart contour and the voxel points on the heart surface of the chest wall contour:

[0078]

[0079] Where point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j );

[0080] The voxel points on the chest wall surface of the heart contour are recorded as a set Ω, and the voxel points on the heart surface of the chest wall contour are recorded as a set Ω′. The shortest Euclidean distance between the two sets is defined as the heart-chest Euclidean distance, which is calculated as follows:

[0081]

[0082] Step 4. Input the heart-chest Euclidean distance into the preset prediction model and output the heart dose;

[0083] The prediction model is constructed based on the correlation between the heart-chest Euclidean distance and the average cardiac dose. The construction process of the prediction model specifically includes:

[0084] A total of 53 patients with left breast cancer after modified radical mastectomy who received adjuvant IMRT were included, and all patients were on free breathing.

[0085] The Euclidean distance between the heart and chest measured based on the patient's positioning CT images was 2.33±2.46mm (mean±standard deviation), with a maximum value of 12.33mm and a minimum value of 0.00mm.

[0086] The mean cardiac dose (MHD) of 53 patients was 431.00±90.50 cGy, with a maximum of 714.60 cGy and a minimum of 229.30 cGy.

[0087] The correlation between mean heart dose (MHD) and cardiothoracic Euclidean distance was analyzed, such as Figure 4 As shown, there was a moderate negative correlation between the Euclidean distance between the heart and chest and the mean cardiac dose (MHD), with a correlation coefficient of r = -0.604 (P < 0.001), that is, the larger the Euclidean distance, the smaller the mean cardiac dose (MHD). The constructed linear prediction model was: y = -19.57x + 476.60, where x represents the Euclidean distance between the heart and chest, and y represents the mean cardiac dose (MHD).

[0088] In addition, studies have shown that patients with MHD ≥ 4Gy have a greater risk of early radiation-induced cardiac injury. The above research results show that the Euclidean distance between the heart and chest is negatively correlated with MHD, so the predictive efficacy of the Euclidean distance between the heart and chest for MHD < 4Gy was analyzed. Figure 5 As shown in the figure, the area under the curve (AUC) of the cardiothoracic Euclidean distance for predicting MHD < 4 Gy was 0.8542. When the cardiothoracic Euclidean distance was greater than 2.5 mm, the sensitivity and specificity of predicting MHD < 4 Gy were 94% and 76%.

[0089] Combining the above results, it can be found that by measuring the anatomical landmark of the heart-chest Euclidean distance based on the patient positioning CT image, the heart radiation dose of patients with left breast cancer receiving adjuvant radiotherapy can be quickly and accurately predicted.

[0090] The present invention automatically identifies and locates the chest wall target area contour and the heart target area contour in the CT image, determines the upper and lower limits of the heart contour to clarify the CT plane for measurement and calculation; defines the range of the chest wall surface of the heart contour and the heart surface of the chest wall contour by connecting the outer limit of the chest wall target area with the dorsal edge of the heart contour, and the inner limit of the chest wall target area with the ventral edge of the heart contour; measures the Euclidean distance between many voxel points on the chest wall surface of the heart contour and the heart surface of the chest wall contour based on the theoretical basis of the Euclidean distance, and defines the shortest Euclidean distance as the heart-chest Euclidean distance. The anatomical index proposed by the present invention is obtained based on the measurement of the three-dimensional spatial position relationship between the heart and the chest wall target area. The measurement method realizes automated, rapid and accurate measurement, minimizes the error of the measurer, and reduces the burden on medical staff. Constructing a heart dose prediction model based on the heart-chest Euclidean distance can help clinicians quickly and accurately understand the heart exposure dose of patients with left breast cancer who are to receive adjuvant radiotherapy, so as to take intervention measures as early as possible and closely monitor and follow up.

[0091] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A cardiac dose prediction system based on the Euclidean distance between the heart and chest, characterized in that: include: Information recognition module, information reconstruction module, prediction module; The information recognition module is used to recognize the CT image to be tested and obtain the heart contour and chest wall target area contour; The information reconstruction module is used to construct a heart contour chest wall surface and a chest wall contour heart surface from the heart contour and the chest wall target area contour based on the SciPy software package, and calculate the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface; The prediction module is used to input the heart-chest Euclidean distance into a preset prediction model to output a cardiac dose, wherein the prediction model is constructed based on the correlation between the heart-chest Euclidean distance and the average cardiac dose; The information reconstruction module constructs the heart contour chest wall surface and the chest wall contour heart surface from the heart contour and the chest wall target area contour based on the SciPy software package, including: Determining a starting CT layer and an ending CT layer of the heart contour; On the same CT layer within the range from the starting CT layer to the ending CT layer, respectively obtaining a first connecting line between the outer boundary of the chest wall target area contour and the dorsal edge of the heart contour, and a second connecting line between the inner boundary of the chest wall target area contour and the ventral edge of the heart contour; The heart contour close to the chest wall target area between the intersection points of the first connecting line, the second connecting line and the heart contour is the heart contour chest wall surface, and the chest wall target area contour close to the heart between the intersection points of the first connecting line, the second connecting line and the chest wall target area contour is the chest wall contour heart surface; The information reconstruction module calculates the heart-chest Euclidean distance based on the heart contour chest wall surface and the chest wall contour heart surface, including: Acquire voxel points on the chest wall surface of the heart contour and construct a set Ω; Acquire voxel points on the cardiac surface of the chest wall contour and construct a set Ω′; Calculate the shortest Euclidean distance between the set Ω and the set Ω′, which is the heart-chest Euclidean distance; The calculation formula of the heart-chest Euclidean distance is: Among them, point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j ); The prediction model in the prediction module is: y=-19.57x+476.60 Where x represents the Euclidean distance between the heart and chest, and y represents the average heart dose.

2. A cardiac dose prediction method based on the heart-chest Euclidean distance, characterized in that: The method for implementing the cardiac dose prediction system based on the heart-chest Euclidean distance according to any one of claims 1 comprises: Importing a CT image to be tested, and identifying a heart contour and a chest wall target area contour in the CT image to be tested; Based on the SciPy software package, a heart contour chest wall surface and a chest wall contour heart surface are constructed from the heart contour and the chest wall target area contour, and the heart-chest Euclidean distance is calculated based on the heart contour chest wall surface and the chest wall contour heart surface; The heart-chest Euclidean distance is input into a preset prediction model to output a heart dose, wherein the prediction model is constructed based on the correlation between the heart-chest Euclidean distance and the average heart dose.

3. The cardiac dose prediction method based on the heart-chest Euclidean distance according to claim 2, characterized in that: The calculation method of the heart-chest Euclidean distance is: Acquire voxel points on the chest wall surface of the heart contour and construct a set Ω; Acquire voxel points on the cardiac surface of the chest wall contour and construct a set Ω′; The shortest Euclidean distance between the set Ω and the set Ω′ is calculated using the following formula: Among them, point A is the i-th voxel point on the chest wall of the heart contour (x i ,y i , z i ), point B is the jth voxel point (x′) on the cardiac surface of the chest wall contour j , y′ j , z′ j ).

4. The cardiac dose prediction method based on the heart-chest Euclidean distance according to claim 2, characterized in that: The prediction model is: y=-19.57x+476.60 Where x represents the Euclidean distance between the heart and chest, and y represents the average heart dose.

Images