CN113643805B - Meningioma gamma knife post-treatment edema prediction system based on image histology - Google Patents
Meningioma gamma knife post-treatment edema prediction system based on image histology Download PDFInfo
- Publication number
- CN113643805B CN113643805B CN202110913967.9A CN202110913967A CN113643805B CN 113643805 B CN113643805 B CN 113643805B CN 202110913967 A CN202110913967 A CN 202110913967A CN 113643805 B CN113643805 B CN 113643805B
- Authority
- CN
- China
- Prior art keywords
- image
- histology
- edema
- meningioma
- imaging
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 206010030113 Oedema Diseases 0.000 title claims abstract description 70
- 206010027191 meningioma Diseases 0.000 title claims abstract description 65
- 206010028980 Neoplasm Diseases 0.000 claims abstract description 41
- 238000003384 imaging method Methods 0.000 claims abstract description 40
- 230000002980 postoperative effect Effects 0.000 claims abstract description 28
- 210000003625 skull Anatomy 0.000 claims abstract description 20
- 230000004083 survival effect Effects 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 17
- 238000000034 method Methods 0.000 claims description 15
- 238000012216 screening Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 5
- 210000003128 head Anatomy 0.000 claims description 5
- 230000003902 lesion Effects 0.000 claims description 5
- 230000002093 peripheral effect Effects 0.000 claims description 4
- 210000001154 skull base Anatomy 0.000 claims description 4
- 210000004204 blood vessel Anatomy 0.000 claims description 3
- 208000031513 cyst Diseases 0.000 claims description 3
- 230000001338 necrotic effect Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000009825 accumulation Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 5
- 238000004393 prognosis Methods 0.000 abstract description 5
- 238000004590 computer program Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000002719 stereotactic radiosurgery Methods 0.000 description 5
- 238000001356 surgical procedure Methods 0.000 description 5
- 206010072360 Peritumoural oedema Diseases 0.000 description 3
- 230000002962 histologic effect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000007917 intracranial administration Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000002203 pretreatment Methods 0.000 description 2
- 208000024891 symptom Diseases 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 206010067484 Adverse reaction Diseases 0.000 description 1
- 206010003591 Ataxia Diseases 0.000 description 1
- 108700028146 Genetic Enhancer Elements Proteins 0.000 description 1
- 206010019233 Headaches Diseases 0.000 description 1
- 206010028813 Nausea Diseases 0.000 description 1
- 206010060860 Neurological symptom Diseases 0.000 description 1
- 208000037114 Symptom Flare Up Diseases 0.000 description 1
- 230000006838 adverse reaction Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000007475 c-index Methods 0.000 description 1
- 238000011088 calibration curve Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000009198 gamma knife radiosurgery Methods 0.000 description 1
- 231100000869 headache Toxicity 0.000 description 1
- 230000003054 hormonal effect Effects 0.000 description 1
- 230000002601 intratumoral effect Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 230000008693 nausea Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
- 230000004614 tumor growth Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Abstract
The invention provides a meningioma gamma knife posttumor peri-edema prediction system based on image histology; the device comprises an image histology feature extraction unit, a data processing unit and a data processing unit, wherein the image histology feature extraction unit is used for extracting image histology features of tumor areas of preoperative skull magnetic resonance conventional sequences of each patient; removing redundant features in the image histology features of each patient to obtain the screened important image histology features; combining the screened image histology characteristics, the clinical characteristics of the clinical data acquisition unit, the image characteristics acquired by the image semantic characteristics acquisition unit and the like, obtaining the relation between the clinical image characteristics and the postoperative edema incidence and occurrence period acquired by the outcome judgment unit according to the occurrence or non-occurrence and occurrence time of edema, and establishing a random survival forest model for prediction. The postoperative edema prediction system has the advantages of noninvasive property, repeatability and easiness in operation by extracting the imaging characteristics of the tumor region and combining the clinical characteristics, and can provide powerful support for assessing prognosis of meningioma patients with a gamma knife and improving clinical decisions.
Description
Technical Field
The invention belongs to the technical field of medical image processing, and particularly relates to a meningioma gamma knife postedema prediction system based on image histology.
Background
Meningiomas are the most common benign intracranial tumors, accounting for 13-26% of the primary intracranial tumors. Although most meningiomas have a good prognosis, they often recur because of the difficulty of complete excision. In addition, since meningioma boundaries are clear, it is an ideal tumor type for Stereotactic Radiosurgery (SRS) because of its accurate localization in magnetic resonance imaging. Therefore, SRS has become an important therapeutic strategy for meningioma patients to suppress tumor growth for long periods and prevent symptom exacerbations. The control rate of the local tumor in 5-10 years of SRS treatment can reach 87-98%. Gamma knife radiosurgery (hereinafter gamma knife) is the most widely used SRS method. Although gamma knife is a recommended treatment, there is a risk of peri-neoplastic radiation edema after surgery, and literature reports have a wide range of incidence rates ranging from 2% to 50% with average/median onset times between 3 and 9 months. Some GKS post-surgery oedema may be asymptomatic, but some oedema can lead to headache, nausea, ataxia, other neurological symptoms, and even death. In more severe cases, these symptoms may require hormonal treatment, even surgical removal of the lesion. The incidence of adverse reactions associated with these oedemas and the necessity of gamma knife therapy are measured by the clinician. Predicting the probability of occurrence of postoperative edema is therefore of great importance for clinical decision making. Several documents have reported factors associated with an increased risk of edema following gamma knife in meningioma patients. Potential factors include higher doses, larger tumor volumes, skull base (especially near sagittal plane) location, presence of pre-treatment edema, etc.
In recent years, imaging histology has shown tremendous potential in the prognosis of some diseases. Image histology is achieved by extracting from an image a number of quantitative features that describe aspects of the image. Such high-dimensional data is a challenge to conventional statistical techniques, such as linear regression, cox regression, and the like. At the same time, the performance of Cox regression will be unreliable when there is a high erasure rate. In contrast, machine learning methods that have rapidly progressed in recent years can effectively cope with high-dimensional problems. Among the various machine learning models, the random living forest (RSF) is a non-parametric model that achieves very high predictive performance when the data comprising the time of occurrence of an event comprises a large amount of covariates. Several studies have demonstrated good performance of RSF in a number of fields, but the use of RSF in meningioma patients in post-gamma-knife edema prediction has not been reported.
Disclosure of Invention
The invention aims to provide a meningioma gamma knife postedema prediction system based on image histology. Based on the basic clinical information, imaging features and imaging histology features of meningioma patients, tools can be established that can rapidly predict the incidence and time of edema following gamma knife.
To achieve the above object, in one aspect, the present invention provides a meningioma gamma knife posttumor peri-edema prediction system based on image histology, the system comprising:
clinical data acquisition unit:
the clinical data of meningioma patients treated by the gamma knife, the information of the head magnetic resonance images visited before and after treatment are obtained, and a clinical data set is formed;
the ending judgment unit:
-for obtaining the occurrence and time of postoperative peri-neoplastic edema from a patient's postoperative follow-up skull magnetic resonance and forming a outcome dataset;
an image histology feature extraction unit:
-for image histology feature extraction and screening from a pre-operative skull magnetic resonance conventional sequence meningioma region of a patient and for forming an image histology feature dataset;
an imaging semantic feature acquisition unit:
the method comprises the steps of carrying out interpretation analysis on images of a meningioma region of a preoperative magnetic resonance routine sequence of a patient to obtain imaging semantic features, and forming an imaging semantic feature data set;
a prediction model building unit:
the method comprises the steps of firstly, randomly dividing a patient into a training set and a testing set, establishing a series of random survival forest models containing different types of features according to the ending data of the training set and combining clinical data, image group science feature data and image science semantic feature data of the training set, verifying in the testing set, evaluating the prediction effect of the training set on the occurrence rate and the occurrence time of the meningioma gamma knife postoperative edema, and generating an optimal model;
an output unit:
and outputting the estimated predicted value of the incidence rate and the occurrence time of the peri-tumor edema after the meningioma gamma knife obtained by the optimal model.
Further, in the clinical data acquisition unit, clinical data of the meningioma patient includes sex, age, lesion range, peripheral dose, center dose, target number, isodose line, surgical history and whether or not to treat in batches.
Furthermore, the clinical data acquisition unit is also used for counting clinical data and converting variables; the conversion includes converting the surgical history and whether the treatment was fractionated in the clinical data into a dichotomous variable.
Further, in the image histology feature extraction unit, the skull magnetic resonance conventional sequence comprises a T1 enhancement sequence, a T2 sequence and an ADC sequence.
Furthermore, in the image histology feature extraction unit, the extracted features include gray histogram gray matrix (GLSZM), form factor (form factor), haralick, gray co-occurrence matrix (GLCM) and Run Length Matrix (RLM), and finally several image histology features are extracted in each sequence.
Furthermore, in the imaging semantic feature acquisition unit, the imaging semantic features include tumor position, whether boundaries are regular, whether tumors are uniformly reinforced on the enhanced T1 magnetic resonance, whether blood vessels exist in the tumors, whether cysts or necrotic components exist in the tumors, whether peridural tail symptoms exist before gamma knife treatment, and whether peritumor edema exists.
Furthermore, the imaging semantic feature acquisition unit is also used for counting imaging semantic features and converting variables; the transformation includes dividing the tumor location into two classification variables, whether it is located by the vector, whether it is located at the skull base.
Further, in the prediction model building unit, the patient is set to 7:3 are randomly divided into training sets and test sets.
Furthermore, the prediction model building unit evaluates the prediction effect of different random survival forest models on the occurrence of the meningioma gamma knife postoperative edema according to the integration area under the curve (iAUC) of the ROC curve in the cumulative/dynamic state, and generates an optimal model.
Further, the prediction model building unit is further used for visually displaying the occurrence rate of edema at different times by using the alignment chart drawn by the prediction risk score of the optimal model. Used to assist in guiding clinical decisions.
In another aspect, the present invention also provides a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
s1, acquiring clinical data of a meningioma patient treated by a gamma knife, and head magnetic resonance visited before and after treatment, and forming a clinical data set for establishing a prediction system;
s2, acquiring the occurrence condition and the occurrence time of postoperative peri-tumor edema from postoperative follow-up skull magnetic resonance of a patient, and forming a final data set for the establishment of a prediction system;
s3, extracting and screening image histology characteristics from a meningioma region of a preoperative skull magnetic resonance conventional sequence (T1 enhancement, T2 and ADC) of the patient, and forming an image histology characteristic data set for establishing a prediction system;
s4, analyzing a preoperative magnetic resonance routine sequence of the patient to obtain imaging characteristics, and forming an imaging semantic characteristic data set for establishing a prediction system;
s5, the patient is pressed by 7: and 3, randomly dividing the model into a training set test set, establishing a series of random survival forest models containing different types of characteristics according to training set data, verifying in the test set, evaluating the prediction effect of the model on occurrence of the meningioma gamma postoperative edema, and generating an optimal model.
In another aspect, the present invention also provides a computer readable storage medium storing a computer program comprising program instructions which, when executed by a processor, cause the processor to perform the steps S1-S5 described above.
Compared with the prior art, the invention has the following beneficial effects:
1) The invention provides a meningioma gamma knife posttumor peri-edema prediction system based on image histology, which comprises the steps of extracting image histology characteristics of tumor areas of preoperative skull magnetic resonance conventional sequences (T1 enhancement, T2, ADC and the like) of each patient; removing redundant features in the image histology features of each patient to obtain the screened important image histology features; combining the screened image histology characteristics, clinical characteristics, visual characteristics obtained by naked eyes and the like, obtaining the relation between the clinical image characteristics, postoperative edema incidence and occurrence period according to the occurrence or non-occurrence and occurrence time of edema, and establishing a random survival forest model for prediction;
2) The postoperative edema prediction system has the advantages of noninvasive property, repeatability and easiness in operation by extracting the imaging characteristics of a tumor region and combining clinical characteristics, achieves the incidence rate of gamma knife postoperative edema of a meningioma patient by the aid of machine learning technical characteristics, provides powerful support for prognosis and improvement of clinical decisions, and achieves the beneficial effect that a doctor can decide whether to treat gamma knife according to the situation judged by a model.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:
FIG. 1 is a schematic flow diagram of an application of an imaging-based meningioma gamma post-knife edema prediction system according to one embodiment of the invention;
FIG. 2 is a graph of time dependence AUC curve for the model with highest iAUC for one embodiment of the invention;
a in fig. 3 is an alignment chart of an embodiment of the present invention, and B is a correction curve;
FIG. 4 is a system architecture diagram of an imaging-based meningioma gamma post-knife edema prediction system according to one embodiment of the invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with one another without conflict.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the present application.
An embodiment of the invention provides an imaging-based meningioma gamma post-knife edema prediction system, comprising:
clinical data acquisition unit: the clinical data of meningioma patients treated by the gamma knife, the information of the head magnetic resonance images visited before and after treatment are obtained, and a clinical data set is formed;
the ending judgment unit: -for obtaining the occurrence and time of postoperative peri-neoplastic edema from a patient's postoperative follow-up skull magnetic resonance and forming a outcome dataset;
an image histology feature extraction unit: -for image histology feature extraction and screening from a pre-operative skull magnetic resonance conventional sequence meningioma region of a patient and for forming an image histology feature dataset;
an imaging semantic feature acquisition unit: the method comprises the steps of carrying out interpretation analysis on images of a meningioma region of a preoperative magnetic resonance routine sequence of a patient to obtain imaging semantic features, and forming an imaging semantic feature data set;
a prediction model building unit: the method comprises the steps of firstly, randomly dividing a patient into a training set and a testing set, establishing a series of random survival forest models containing different types of features according to the ending data of the training set and combining clinical data, image group science feature data and image science semantic feature data of the training set, verifying in the testing set, evaluating the prediction effect of the training set on the occurrence rate and the occurrence time of the meningioma gamma knife postoperative edema, and generating an optimal model;
an output unit: and outputting the estimated predicted value of the incidence rate and the occurrence time of the peri-tumor edema after the meningioma gamma knife obtained by the optimal model.
FIG. 1 is a flow chart showing the application of the imaging-based meningioma gamma post-knife edema prediction system according to the embodiment of the invention; as shown in fig. 1, the prediction method of the meningioma gamma post-edema prediction system applying the present invention includes the steps of:
s1: acquiring clinical data of a meningioma patient treated by a gamma knife, and performing follow-up skull magnetic resonance before treatment and after treatment;
s2: acquiring the occurrence condition and the occurrence time of postoperative peri-tumor edema from postoperative follow-up skull magnetic resonance of a patient;
s3: extracting and screening image histology characteristics from a meningioma region of a preoperative skull magnetic resonance routine sequence (T1 enhancement, T2 and ADC) of a patient;
s4: analyzing the preoperative magnetic resonance routine sequence of the patient by naked eyes to obtain imaging characteristics;
s5: the patient was treated as 7: and 3, randomly dividing the model into a training set test set, establishing a series of random survival forest models containing different types of characteristics according to training set data, verifying in the test set, evaluating the prediction effect of the model on occurrence of the meningioma gamma postoperative edema, and generating an optimal model.
In step S1, the clinical data of the meningioma patient includes the patient' S sex, age, lesion extent, peripheral dose, center dose, target number, isodose line, surgical history and whether to treat in fractions.
In a specific example, there were 9 clinical features of gender, age, lesion extent, peripheral dose, center dose, target number, isodose line, surgical history and whether to treat in fractions for 445 patients. The average age of 445 patients was 54.7±11.4 years, of which 343 women (77.1%), 102 men (22.9%). The 445 patients were concurrently collected for pre-treatment and post-treatment follow-up skull magnetic resonance image information.
In step S2, the occurrence and time of postoperative peri-neoplastic edema are acquired from the patient' S postoperative follow-up skull magnetic resonance. Peri-tumor edema is defined as the high density shadow around meningiomas on T2 sequence magnetic resonance. The occurrence of edema is defined as the progression of new edema after surgery in a patient without preoperative intratumoral edema, or edema after surgery in a patient with preoperative edema. The time of occurrence is from the date of the gamma knife to the date when the magnetic resonance finds new edema or edema progression.
In one specific example, 43 meningioma patients developed new edema, 33 had preoperative edema progression, and a total of 76 (17.1%). The remaining 369 cases did not develop new edema or no progress in preoperative edema.
In step S3, image histology feature extraction and screening is performed from the meningioma region of the patient' S preoperative skull magnetic resonance conventional sequence (T1 enhancement, T2, ADC).
In a specific embodiment, the extracted features include gray histogram gray matrix (GLSZM), form factor (form factor), haralick, gray co-occurrence matrix (GLCM), and Run Length Matrix (RLM), and finally 396 image histology features (1188 total) are extracted in each sequence.
Screening the characteristics according to the importance of the characteristics by a variable husting method, and finally screening 16 characteristics by a T1 enhancement sequence; screening 16 characteristics by the T2 sequence; the ADC map screens out 17 features. In addition, all the features of the T1 strive for, T2 and ADC map are combined, and 22 features are screened. In particular table 1 below.
Table 1.variable hunting method screens out features based on their importance
In step S4, the imaging physician obtains imaging features (i.e. semantic features, semantic features) by macroscopic analysis of the patient' S preoperative magnetic resonance routine sequence, including tumor location, whether the boundaries are regular, whether the tumor is uniformly reinforced on the enhanced T1 magnetic resonance, whether there are blood vessels in the tumor, whether there are cysts or necrotic components in the tumor, dural tail signs, and whether there is peri-tumor edema before gamma knife treatment. Wherein, the tumor position is divided into two classified variables of whether the tumor is located beside a vector and at the skull base. Thus, there are 8 imaging features.
In step S5, the patient is pressed 7: and 3, randomly dividing the training set into a training set test set, establishing a series of random survival forest models containing different types of features according to training set data, and verifying in the test set.
In a specific embodiment, the training set includes 312 meningioma patients and the test set includes 133 meningioma patients. There were no significant differences between the baseline clinical data for the training set and the test set.
The present example trained a total of 19 random living forest (RSF) models. These models include one layer (model 1-T1, 1-T2, 1-ADC, 1-Rad, 1-C, 1-S), two layers (model 2-CS, 2-CT1, 2-CT2, 2-CADC, 2-CRad, 2-ST1, 2-ST2, 2-SADC, 2-SRad) and three layers (model 3-CST1, 3-CST2, 3-CSADC, 3-CSRad) models, each containing different combinations of imaging histology, clinical features, and imaging semantic features. Each model is trained based on the following parameters: model 1-T1: the histology of 16T 1 enhancer sequences selected from 396 features; model 1-T2: a set of 16T 2 sequences selected from 396 features using the same method; model 1-ADC: a histology feature of 17 ADC maps selected from 396 extracted features; model 1-Rad: 22 histologic features selected from 1188 features extracted from the three sequences; model 1-C:9 clinical features; model 1-S:8 imaging semantic features; model 2-CS: 17 clinical and imaging semantic features are combined; model 2-CT1: clinical features and selected T1-enhanced histology features (25 total); model 2-CT2: clinical features and selected T2 sequence histology features (25 total); model 2-CADC: clinical features and selected ADC sequence histology features (26 total); model 2-CRad: clinical features and 22 histologic features of 1-Rad; model 2-ST1: semantic features and selected T1-enhanced histology features (24 total); model 2-ST2: semantic features and selected T2 sets of features (24 total); model 2-SADC: semantic features and selected ADC group features (25 total); model 2-SRad: semantic features and 22 histology features (30) of 1-Rad; model 3-CST1: clinical, semantic, and selected T1-enhanced histology features (33); model 3-CST2: clinical, semantic, and selected T2 sets of features (33); model 3-CSADC: clinical, semantic, and selected ADC histology features (34); model 3-CSRad: clinical, semantic, and 22 histologic features of 1-Rad (39).
In the embodiment, the prediction effect of different random survival forest models on the occurrence of meningioma gamma postoperative edema is estimated according to the integral area under the curve (iAUC) of the ROC curve in the cumulative/dynamic process, so as to generate an optimal model.
Fig. 2 is the AUC curves for the 7 models with the highest iaaucs. The highest model of the iAUC is model 3-CST1, the iAUC of the model reaches 0.942 (95% credible interval: 0.939-0.944), and the c-index is 0.964+/-0.002,Brier score:0.131.
The present embodiment also provides for a nomogram drawn based on the predicted risk score of the optimal model for visually displaying the occurrence of oedema at different times for guiding clinical decisions. See fig. 3. Fig. 3A is an alignment chart showing the occurrence of oedema at 0.5, 1, 1.5 and 2 years, and fig. 3B is a calibration curve at different times to obtain bias correction estimates.
According to the invention, the random survival forest model is predicted by constructing the peritumoral edema after the meningioma gamma knife based on image histology, the optimal model is selected, and the method has the advantages of noninvasive property, repeatability and easiness in operation, and can provide powerful support for assessing the prognosis of meningioma patients of the gamma knife and improving clinical decisions.
Figure 4 is a schematic diagram of a meningioma gamma postknife peri-tumoral edema prediction system according to one embodiment of the invention. As shown in fig. 4, a preliminary screening diagnosis system for predicting peri-tumor edema after a meningioma gamma knife based on image histology, which implements the steps of predicting peri-tumor edema after a meningioma gamma knife based on image histology, as shown in the above method, includes:
the data acquisition unit is used for acquiring clinical data of a meningioma patient treated by the gamma knife and head magnetic resonance visited before and after treatment;
the outcome judging unit is used for acquiring the occurrence condition and the occurrence time of postoperative peri-tumor edema from the postoperative follow-up skull magnetic resonance of the patient;
the (image histology) feature extraction unit is used for extracting and screening image histology features from a meningioma region of a patient preoperative skull magnetic resonance conventional sequence (T1 enhancement, T2 and ADC);
a (imaging) semantic feature acquisition unit for acquiring imaging features by visual analysis of a preoperative magnetic resonance routine sequence of the patient;
prediction model building unit, patient is according to 7: and 3, randomly dividing the model into a training set test set, establishing a series of random survival forest models containing different types of characteristics according to training set data, verifying in the test set, evaluating the prediction effect of the model on occurrence of the meningioma gamma postoperative edema, and generating an optimal model.
Fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. As shown in fig. 5, an electronic device of one embodiment of the present invention includes one or more input devices, one or more output devices, one or more processors, and memory.
In one embodiment of the invention, the processor, input device, output device, and memory may be connected by a bus or other means. The input device, the output device may be a standard wired or wireless communication interface.
The processor may be a central processing module (Central Processing Unit, CPU), which may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may be a high-speed RAM memory or a non-volatile memory such as a disk memory. The memory is used to store a set of computer programs and the input device, the output device and the processor may invoke program code stored in the memory.
The memory-stored computer program comprises program instructions which, when executed by a processor, cause the processor to perform the steps of imaging-group-based post-meningioma gamma knife peri-tumoral edema prediction as described in the above embodiments.
An embodiment of the present invention also provides a computer-readable storage medium. The computer readable storage medium may be a high speed RAM memory or may be a non-volatile memory such as a disk memory. The computer readable storage medium may be connected through an external computing device or network to read a set of computer programs stored by the computer readable storage medium. The computer readable storage medium stores a computer program comprising program instructions which, when executed by a processor, cause the processor to perform the steps of the method for predicting peri-tumor edema of a meningioma gamma knife based on image histology as described in the above embodiments.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (4)
1. A meningioma gamma knife post-tumor peri-edema prediction system based on image set, the system comprising:
clinical data acquisition unit:
the clinical data of meningioma patients treated by the gamma knife, the information of the head magnetic resonance images visited before and after treatment are obtained, and a clinical data set is formed;
the ending judgment unit:
-for obtaining the occurrence and time of postoperative peri-neoplastic edema from a patient's postoperative follow-up skull magnetic resonance and forming a outcome dataset;
an image histology feature extraction unit:
-for image histology feature extraction and screening from a pre-operative skull magnetic resonance conventional sequence meningioma region of a patient and for forming an image histology feature dataset; the skull magnetic resonance routine sequence comprises T1 enhancement, T2 and ADC sequences; the extracted features comprise a gray level histogram gray level matrix, a form factor, a Haralick, a gray level co-occurrence matrix and a run-length matrix, and finally, a plurality of image histology features are extracted in each sequence;
an imaging semantic feature acquisition unit:
the method comprises the steps of carrying out interpretation analysis on images of a meningioma region of a preoperative magnetic resonance routine sequence of a patient to obtain imaging semantic features, and forming an imaging semantic feature data set; the imaging semantic features comprise tumor positions, whether boundaries are regular, whether tumors are uniformly reinforced on enhanced T1 magnetic resonance, whether blood vessels exist in the tumors, whether cysts or necrotic components exist in the tumors, and whether peri-tumor edema exists before gamma knife treatment;
a prediction model building unit:
-to press 7:3, randomly dividing the model into a training set and a testing set, establishing a series of random survival forest models containing different types of characteristics according to the ending data of the training set and combining clinical data, image histology characteristic data and image semantic characteristic data of the training set, verifying the random survival forest models in the testing set, and evaluating the prediction effect of the different random survival forest models on the occurrence rate and the occurrence time of the meningioma gamma knife postoperative edema according to the integration area under the ROC curve when in accumulation/dynamic state to generate an optimal model;
the prediction model building unit is also used for drawing a nomogram according to the prediction risk score of the optimal model and visually displaying the occurrence rate of edema at different times;
an output unit:
and outputting the estimated predicted value of the incidence rate and the occurrence time of the peri-tumor edema after the meningioma gamma knife obtained by the optimal model.
2. The imaging-based postmeningioma gamma knife peri-tumor edema prediction system of claim 1, wherein the clinical data of the meningioma patient in the clinical data acquisition unit includes patient gender, age, lesion size, peripheral dose, central dose, target number, isodose line, surgical history, and whether to treat in fractions.
3. The imaging-based meningioma gamma post-tumor peri-edema prediction system according to claim 2, wherein the clinical data acquisition unit is further configured to calculate clinical data and convert variables; the conversion includes converting the surgical history and whether the treatment was fractionated in the clinical data into a dichotomous variable.
4. The imaging-based meningioma gamma post-tumor peri-edema prediction system according to claim 1, wherein the imaging semantic feature acquisition unit is further configured to count imaging semantic features and convert variables; the transformation includes dividing the tumor location into two classification variables, whether it is located by the vector, whether it is located at the skull base.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110913967.9A CN113643805B (en) | 2021-08-10 | 2021-08-10 | Meningioma gamma knife post-treatment edema prediction system based on image histology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110913967.9A CN113643805B (en) | 2021-08-10 | 2021-08-10 | Meningioma gamma knife post-treatment edema prediction system based on image histology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113643805A CN113643805A (en) | 2021-11-12 |
| CN113643805B true CN113643805B (en) | 2024-04-02 |
Family
ID=78420499
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110913967.9A Active CN113643805B (en) | 2021-08-10 | 2021-08-10 | Meningioma gamma knife post-treatment edema prediction system based on image histology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113643805B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815481A (en) * | 2017-01-19 | 2017-06-09 | 中国科学院深圳先进技术研究院 | A kind of life cycle Forecasting Methodology and device based on image group |
| CN111657945A (en) * | 2020-06-16 | 2020-09-15 | 中南大学湘雅医院 | Nasopharyngeal carcinoma prognosis auxiliary evaluation method based on enhanced MRI (magnetic resonance imaging) imaging omics |
| CN112185569A (en) * | 2020-09-11 | 2021-01-05 | 中山大学孙逸仙纪念医院 | Breast cancer patient disease-free survival period prediction model and construction method thereof |
| CN112768072A (en) * | 2021-01-12 | 2021-05-07 | 哈尔滨医科大学 | Cancer clinical index evaluation system constructed based on imaging omics qualitative algorithm |
| CN112951406A (en) * | 2021-01-27 | 2021-06-11 | 安徽理工大学 | Lung cancer prognosis auxiliary evaluation method and system based on CT (computed tomography) image omics |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200058125A1 (en) * | 2018-08-14 | 2020-02-20 | Mikhail Teverovskiy | Comparative cancer survival models to assist physicians to choose optimal treatment |
| WO2020056372A1 (en) * | 2018-09-14 | 2020-03-19 | Krishnan Ramanathan | Multimodal learning framework for analysis of clinical trials |
-
2021
- 2021-08-10 CN CN202110913967.9A patent/CN113643805B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815481A (en) * | 2017-01-19 | 2017-06-09 | 中国科学院深圳先进技术研究院 | A kind of life cycle Forecasting Methodology and device based on image group |
| CN111657945A (en) * | 2020-06-16 | 2020-09-15 | 中南大学湘雅医院 | Nasopharyngeal carcinoma prognosis auxiliary evaluation method based on enhanced MRI (magnetic resonance imaging) imaging omics |
| CN112185569A (en) * | 2020-09-11 | 2021-01-05 | 中山大学孙逸仙纪念医院 | Breast cancer patient disease-free survival period prediction model and construction method thereof |
| CN112768072A (en) * | 2021-01-12 | 2021-05-07 | 哈尔滨医科大学 | Cancer clinical index evaluation system constructed based on imaging omics qualitative algorithm |
| CN112951406A (en) * | 2021-01-27 | 2021-06-11 | 安徽理工大学 | Lung cancer prognosis auxiliary evaluation method and system based on CT (computed tomography) image omics |
Non-Patent Citations (2)
| Title |
|---|
| 低级别胶质瘤的机器学习和影像组学表型:提高生存预测;Y.S.Choi;S.S.Ahn;J.H.Chang;S.G.Kang;E.H.Kim;S.H.Kim;赵海清;;国际医学放射学杂志(第05期);全文 * |
| 基于深度学习的时间飞跃法磁共振血管成像对脑动脉瘤自动检测方法研究;耿辰;上海医学(第5期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113643805A (en) | 2021-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10939874B2 (en) | Deep-learning-based cancer classification using a hierarchical classification framework | |
| US11263793B2 (en) | Method and system for assessing images using biomarkers | |
| Chen et al. | Computer-aided grading of gliomas combining automatic segmentation and radiomics | |
| Tandel et al. | Performance optimisation of deep learning models using majority voting algorithm for brain tumour classification | |
| Smith et al. | Radiomics and radiogenomics of prostate cancer | |
| He et al. | Convolutional neural network to predict the local recurrence of giant cell tumor of bone after curettage based on pre-surgery magnetic resonance images | |
| US11610304B2 (en) | Predicting tumor prognoses based on a combination of radiomic and clinico-pathological features | |
| Gawade et al. | Application of the convolutional neural networks and supervised deep-learning methods for osteosarcoma bone cancer detection | |
| Zhu et al. | Multimodal medical image fusion using adaptive co-occurrence filter-based decomposition optimization model | |
| Abou Ghayda et al. | Artificial intelligence in andrology: from semen analysis to image diagnostics | |
| Pandey et al. | Tumorous kidney segmentation in abdominal CT images using active contour and 3D-UNet | |
| Qiu et al. | Automatic segmentation of hippocampus in hippocampal sparing whole brain radiotherapy: A multitask edge‐aware learning | |
| Artzi et al. | Classification of pediatric posterior fossa tumors using convolutional neural network and tabular data | |
| Nizamani et al. | Advance Brain Tumor segmentation using feature fusion methods with deep U-Net model with CNN for MRI data | |
| Khan et al. | MKIS-Net: A light-weight multi-kernel network for medical image segmentation | |
| CN113643805B (en) | Meningioma gamma knife post-treatment edema prediction system based on image histology | |
| CN111951962B (en) | Bevacizumab curative effect prediction model for treating radioactive brain injury and construction method thereof | |
| Kumar et al. | Improved UNet deep learning model for automatic detection of lung cancer nodules | |
| SobhaniNia et al. | Brain tumor classification using medial residual encoder layers | |
| Mantha et al. | A transfer learning method for brain tumor classification using efficientnet-b3 model | |
| Wang et al. | Automatic segmentation of vestibular schwannomas from T1-weighted MRI with a deep neural network | |
| Anastasiadis et al. | Transforming urinary stone disease management by artificial intelligence-based methods: A comprehensive review | |
| Chang et al. | Image segmentation in 3D brachytherapy using convolutional LSTM | |
| Sobhaninia et al. | Medial residual encoder layers for classification of brain tumors in magnetic resonance images | |
| Miao et al. | Application of deep learning and XGBoost in predicting pathological staging of breast cancer MR images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |