阅读说明：本技术 一种基于头部肿瘤治疗的电极贴片定位方法 (Electrode patch positioning method based on head tumor treatment ) 是由 段红杰 刘胜军 张华� 赵希超 张轩 宋羽 张建义 于 2020-04-02 设计创作，主要内容包括：本发明涉及属于医学处理技术领域的一种基于头部肿瘤治疗的电极贴片定位方法,包括构建头部肿瘤图像训练样本集合、利用CT对患者的头部进行三维图像采集、头部肿瘤的识别、确定肿瘤的坐标和确定电极贴片在头部的具体坐标位置。采用本发明的技术方案,使得更智能化,电极贴片的定位更准确,在肿瘤部位形成最优电场方案,生成患者电极阵列佩戴方案,取得的电磁疗效更好。(The invention relates to an electrode patch positioning method based on head tumor treatment, belonging to the technical field of medical treatment. By adopting the technical scheme of the invention, the intelligent electrode patch positioning system is more intelligent, the positioning of the electrode patch is more accurate, the optimal electric field scheme is formed at the tumor part, the electrode array wearing scheme of a patient is generated, and the obtained electromagnetic curative effect is better.)

1. An electrode patch positioning method based on head tumor treatment is characterized by comprising the following steps:

(1) Constructing a head tumor image training sample set: extracting image texture features and shape features of the obtained head CT image containing the tumor; the tumor identification BP neural network identifies whether the human organ CT image contains suspected tumor or not according to the image texture characteristics; marking each CT image pixel by pixel, and packaging the CT images and the corresponding marked CT images to form a training sample; selecting M training samples from the obtained total tumor target samples, requiring the training samples to contain tumors of different types as much as possible, and manually judging and marking the attributes of the training samples; training a tumor recognition BP neural network by adopting a CT image training sample set to obtain a head tumor image training sample set;

(2) Acquiring a three-dimensional image of the head of a patient by utilizing CT (computed tomography), and extracting image texture characteristics and shape characteristics;

(3) the identification of the head tumor comprises the steps of intercepting a rectangular area containing a tumor target according to the textural features of the tumor target image, rotating the tumor target by- α degrees based on the orientation angle α of the tumor target obtained by detecting the tumor target to obtain a horizontal tumor target sample, enveloping the horizontal tumor target by using a minimum rectangle, intercepting the minimum rectangle to obtain a horizontal sample target slice of the tumor target, and marking the horizontal sample target slice as a test sample;

(4) Determining the coordinates of the tumor: establishing head characteristics of a patient for modeling to obtain a head model of the patient, and forming digital coordinates of the head model of the patient in a three-dimensional space; inputting the head tumor target identified in the step (3) into a head model of the patient, so as to obtain a coordinate position of the tumor in the head model of the patient;

(5) According to the principle that the conductivity and the resistivity of a human body and the electromagnetic action center position of an electrode patch array coincide with the center position of a head tumor, a neural network algorithm is adopted, the electrode patch array is input into a head model of a patient to carry out finite element analysis, a maximum electric field generated at the tumor part is simulated and formed, and then the specific coordinate position of each electrode patch in the electrode patch array at the head is obtained according to the distribution of the electric field; the electromagnetic direction of each electrode patch in the electrode patch array is orthogonally arranged;

(6) Converting the specific coordinate position of each electrode patch on the head into the head position of the patient;

And completing electrode patch positioning based on head tumor treatment.

2. The electrode patch positioning method for head tumor treatment according to claim 1, wherein: the tumor recognition BP neural network comprises 1 input layer, 1 hidden layer and 1 output layer, the texture feature vector and the shape feature parameter of the area where the suspected focus is located are used as neurons of the input layer, the number of nodes of the hidden layer is set to be 2 times of the number of the neurons of the input layer according to the prior experience, and the output layer comprises 1 node and three output values; the three output values respectively represent the lesion in the CT image as benign tumor, malignant tumor and non-tumor.

3. The electrode patch positioning method for head tumor treatment according to claim 2, wherein: in the step (4), the brain is subdivided into 5 large divisions according to parietal lobe, frontal lobe, occipital lobe, temporal lobe and islet lobe of a human brain structure, and then the brain is subdivided into left parietal lobe, right parietal lobe, left frontal lobe, left occipital lobe, right occipital lobe, left temporal lobe, right temporal lobe, left islet lobe and right islet lobe according to tumor positions; A3D head model is generated from the patient's brain structure.

4. The electrode patch positioning method for head tumor treatment according to claim 3, wherein: the brain model of the patient forms a stereoscopic image in software, a highlighted display area is formed in the model according to the size of the tumor part of the patient, the tumor of the patient is generated in a color with obvious contrast, and the position of the tumor on the head is displayed stereoscopically.

5. The electrode patch positioning method for head tumor treatment according to claim 4, wherein: the pulse electrodes of the electrode patch output pulse voltages in a cyclic single or multiple manner.

Technical Field

The invention belongs to the technical field of medical treatment, and particularly relates to an electrode patch positioning method based on head tumor treatment.

Background

Brain gliomas are malignant tumors derived from neuroepithelial tissues, commonly known as "brain cancers". Brain glioma is the most common intracranial primary tumor, and foreign clinical statistics show that the incidence rate of intracranial primary tumors is 21/10 ten thousand, and glioma accounts for about 60%. Domestic literature reports that brain glioma accounts for about 35.26% -60.96% of intracranial tumors. The most common treatment method at present is surgery and radiotherapy and chemotherapy, and the brain glioma grows infiltratively, so the surgery is often difficult to completely cut. And because the tumor is a radiation-resistant tumor and is resistant to most chemotherapeutics, the overall curative effect is poor, especially high-grade glioma has the growth characteristics of high degree anaplasia, the postoperative recurrence is fast, the prognosis is poor, and the health of human beings is seriously threatened.

The use of electric fields and currents to treat neurological disorders and brain diseases is becoming widespread. Examples of such treatments include, but are not limited to: transcranial Direct Current Stimulation (TDCS), Transcranial Magnetic Stimulation (TMS), and tumor therapy field (TTField). These treatments rely on the delivery of low frequency electromagnetic fields to targeted areas within the brain. For example, Woods et al, clinical neurophysiology, 1271031-1048 (2016), review technical aspects of TDCS; and Thielscher et al, "proceedings of conference," Institute of Electrical and Electronics Engineers (IEEE), institute of medical and biological engineers, 222-225 (2015), which teaches methods for simulating TMS. As another example, Miranda et al, Physics in medicine and biology, 594137-4147 (2014), teaches the creation of a computational head model of a healthy individual that simulates the delivery of TTField using a Magnetic Resonance Imaging (MRI) dataset, where the model creation is performed in a semi-automated manner. Furthermore, Wenger et al, Physics in medicine and biology, 607339-7357 (2015), teaches a method for creating a computational head model of a healthy individual to simulate the delivery of TTField, where the model is created from an MRI dataset of a healthy individual.

However, the position of the head electrode patch array is inaccurate, so that the curative effect of treating head tumors by using electric fields and electric currents is influenced.

Disclosure of Invention

The invention aims to provide an electrode patch positioning method based on head tumor treatment, which is determined by utilizing tumor coordinates and a mode of simulating the maximum electric field generated at a tumor part.

The technical scheme adopted by the invention is as follows:

An electrode patch positioning method based on head tumor treatment is characterized by comprising the following steps:

(1) Constructing a head tumor image training sample set: extracting image texture features and shape features of the obtained head CT image containing the tumor; the tumor identification BP neural network identifies whether the human organ CT image contains suspected tumor or not according to the image texture characteristics; marking each CT image pixel by pixel, and packaging the CT images and the corresponding marked CT images to form a training sample; selecting M training samples from the obtained total tumor target samples, requiring the training samples to contain tumors of different types as much as possible, and manually judging and marking the attributes of the training samples; training a tumor recognition BP neural network by adopting a CT image training sample set to obtain a head tumor image training sample set;

(2) Acquiring a three-dimensional image of the head of a patient by utilizing CT (computed tomography), and extracting image texture characteristics and shape characteristics;

(3) the identification of the head tumor comprises the steps of intercepting a rectangular area containing a tumor target according to the textural features of the tumor target image, rotating the tumor target by- α degrees based on the orientation angle α of the tumor target obtained by detecting the tumor target to obtain a horizontal tumor target sample, enveloping the horizontal tumor target by using a minimum rectangle, intercepting the minimum rectangle to obtain a horizontal sample target slice of the tumor target, and marking the horizontal sample target slice as a test sample;

(4) Determining the coordinates of the tumor: establishing head characteristics of a patient for modeling to obtain a head model of the patient, and forming digital coordinates of the head model of the patient in a three-dimensional space; inputting the head tumor target identified in the step (3) into a head model of the patient, so as to obtain a coordinate position of the tumor in the head model of the patient;

(5) According to the principle that the conductivity and the resistivity of a human body and the electromagnetic action center position of an electrode patch array coincide with the center position of a head tumor, a neural network algorithm is adopted, the electrode patch array is input into a head model of a patient to carry out finite element analysis, a maximum electric field generated at the tumor part is simulated and formed, and then the specific coordinate position of each electrode patch in the electrode patch array at the head is obtained according to the distribution of the electric field; the electromagnetic direction of each electrode patch in the electrode patch array is orthogonally arranged;

(6) Converting the specific coordinate position of each electrode patch on the head into the head position of the patient;

And completing electrode patch positioning based on head tumor treatment.

Further, the tumor identification BP neural network comprises 1 input layer, 1 hidden layer and 1 output layer, the texture feature vector and the shape feature parameter of the area where the suspected focus is located are used as neurons of the input layer, the number of nodes of the hidden layer is set to be 2 times of the number of the neurons of the input layer according to the prior experience, and the output layer comprises 1 node and three output values; the three output values respectively represent the lesion in the CT image as benign tumor, malignant tumor and non-tumor.

Further, in the step (4), the brain is subdivided into 5 large divisions according to parietal lobe, frontal lobe, occipital lobe, temporal lobe and islet lobe of a human brain structure, and then the brain is subdivided into left parietal lobe, right parietal lobe, left frontal lobe, right frontal lobe, left occipital lobe, right occipital lobe, left temporal lobe, right temporal lobe, left islet lobe and right islet lobe according to tumor positions; A3D head model is generated from the patient's brain structure.

Furthermore, a three-dimensional image is formed in the software by the brain model of the patient, a highlight display area is formed in the model according to the size of the tumor part of the patient, the tumor of the patient is generated in a color with obvious contrast, and the position of the tumor on the head is displayed in a three-dimensional mode.

Further, a plurality of pulse electrodes of the electrode patch output pulse voltages in a cyclic single or multiple manner.

Compared with the prior art, the invention has the following beneficial effects:

1. According to the invention, the head tumor image training sample set and the tumor recognition BP neural network are adopted, so that the collected samples are more intelligent and the accuracy is higher;

2. The invention adopts a neural network algorithm, inputs the electrode patch array into a head model of a patient for finite element analysis, simulates and forms a maximum electric field generated at a tumor part, and then obtains the specific coordinate position of each electrode patch in the electrode patch array at the head according to the distribution of the electric field, which is more in line with the human body and the electromagnetic principle, so that the positioning of the electrode patches is more accurate and the electromagnetic curative effect is better.

Drawings

FIG. 1 is a schematic diagram of an embodiment employing the present invention.

Detailed Description

The present invention will be further described with reference to fig. 1 and the following detailed description.

An electrode patch positioning method based on head tumor treatment is characterized by comprising the following steps:

(1) Constructing a head tumor image training sample set: extracting image texture features and shape features of the obtained head CT image containing the tumor; the tumor identification BP neural network identifies whether the human organ CT image contains suspected tumor or not according to the image texture characteristics; marking each CT image pixel by pixel, and packaging the CT images and the corresponding marked CT images to form a training sample; selecting M training samples from the obtained total tumor target samples, requiring the training samples to contain tumors of different types as much as possible, and manually judging and marking the attributes of the training samples; training a tumor recognition BP neural network by adopting a CT image training sample set to obtain a head tumor image training sample set;

The tumor recognition BP neural network comprises 1 input layer, 1 hidden layer and 1 output layer, the texture feature vector and the shape feature parameter of the area where the suspected focus is located are used as neurons of the input layer, the number of nodes of the hidden layer is set to be 2 times of the number of the neurons of the input layer according to the prior experience, and the output layer comprises 1 node and three output values; the three output values respectively represent the lesion in the CT image as benign tumor, malignant tumor and non-tumor.

The texture feature is a visual feature reflecting homogeneity phenomenon in an image, and embodies the surface structure organization arrangement attribute with slow change or periodic change of the surface of an object; texture has three major landmarks: some local sequence is repeated continuously, non-random arrangement and uniform unity in the texture region, and can be extracted by adopting a statistical method, a model method and the like.