| CPC G16H 50/20 (2018.01) [G06T 7/11 (2017.01); G06T 2207/10081 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/30028 (2013.01)] | 5 Claims |
|
1. A tumor diagnosis system, comprising:
a plurality of modules, the modules comprising a computer executable code stored on a non-transitory computer-readable storage medium, the modules including:
a radiomics signature-generation module configured to perform acquiring target radiomics information and clinical data information, and obtaining a radiomics signature according to the target radiomics information and the clinical data information, based on a pre-trained radiomics signature model, wherein the radiomics signature comprises a first radiomics signature and a second radiomics signature;
a benign and malignant comprehensive classification module configured to perform acquiring an ambient situation of a lesion outside an intestinal wall, the clinical data information and the first radiomics signature, and generating a first classification comprehensive diagnosis result according to the ambient situation of the lesion outside the intestinal wall, the clinical data information and the first radiomics signature, based on a pre-trained first classification comprehensive diagnosis model;
a malignant focus sub-classification module configured to perform acquiring the first classification comprehensive diagnosis result from the benign and malignant comprehensive classification module, and judging whether the first classification comprehensive diagnosis result comprises high-risk malignant information, wherein if the first classification comprehensive diagnosis result comprises the high-risk malignant information, the ambient situation of the lesion outside the intestinal wall and the clinical data information are obtained from the benign and malignant comprehensive classification module, the second radiomics signature is obtained from the radiomics signature-generation module, and a second classification comprehensive diagnosis result is generated according to the ambient situation of the lesion outside the intestinal wall, the clinical data information and the second radiomics signature, based on a pre-trained second classification comprehensive diagnosis model;
a result displaying module configured to perform acquiring the first classification comprehensive diagnosis result and/or the second classification comprehensive diagnosis result, obtaining a final diagnosis result according to the first classification comprehensive diagnosis result and/or the second classification comprehensive diagnosis result, and displaying the final diagnosis result for a user to check;
a CT colonoscopy (CTC) image preprocessing and reconstructing module configured to perform acquiring an enhanced CTC image of a patient to be diagnosed, and standardizing signal strength and layer thickness of the CTC image through a filter to obtain a preprocessed CTC image, and reconstructing the preprocessed CTC image based on a virtual endoscopy post-processing technology to obtain CT virtual endoscopic imaging for the user to conduct focus localization on the virtual endoscopic imaging and further obtain a focus localization image; and acquiring the focus localization image, and generating a CTC tomographic image according to the focus localization image;
a lesion marking module configured to perform acquiring the CTC tomographic image, recording the ambient situation of the lesion outside the intestinal wall according to the CTC tomographic image, and sending the ambient situation of the lesion outside the intestinal wall to the benign and malignant comprehensive classification module; providing the CTC tomographic image to the user for the user to delineate the CTC tomographic image layer by layer to obtain a region of interest; and acquiring the region of interest and sending the region of interest to a radiomics feature extraction module;
the radiomics feature extraction module configured to perform feature extraction on the region of interest to obtain the target radiomics information, and send the target radiomics information to the radiomics signature-generation module;
a clinical data acquisition module configured to perform acquiring the clinical data information of the patient to be diagnosed input by the user, and sending the clinical data information to the radiomics signature-generation module, wherein the clinical data information comprises one or more of a gender, age, body mass index, family history of cancer, smoking history, constipation history, fecal occult blood and serological test result;
wherein the clinical data acquisition module further comprises a training sample data acquisition unit, the training sample data acquisition unit is used for acquiring training sample data information input by the user and sending the training sample data information to a radiomics signature model training module, wherein the training sample data information comprises clinical data information, colonoscopy and pathological information of a training set;
wherein in the tumor diagnosis system, the radiomics signature model training module comprises a data cleaning unit, a logistic regression unit, and a signature vector calculation unit, wherein the data cleaning unit is used for acquiring an omics feature value and the training sample data information, and performing data cleaning on the omics feature value and the training sample data information to obtain cleaned multi-omics feature data;
the logistic regression unit is used for performing dimensionality reduction screening on the cleaned multi-omics feature data to obtain a first relevant radiomics feature and a second relevant radiomics feature; and the signature vector calculation unit is used for incorporating the first relevant radiomics feature and the second relevant radiomics feature into a vector formula to obtain the radiomics signature model;
a first classification comprehensive diagnosis model training module configured to perform acquiring the training sample data information and the ambient situation of the lesion outside the intestinal wall of the training set, performing screening and analysis according to the training sample data information and the ambient situation of the lesion outside the intestinal wall of the training set to obtain a clinical risk factor; and performing training according to the clinical risk factor through a model training method to obtain the first classification comprehensive diagnosis model; and
a second classification comprehensive diagnosis model training module configured to perform judging whether the first classification comprehensive diagnosis result of the training set generated by the first classification comprehensive diagnosis model comprises the high-risk malignant information, wherein if the first classification comprehensive diagnosis result of the training set comprises the high-risk malignant information, the training sample data information and the ambient situation of the lesion outside the intestinal wall of the training set are acquired from the first classification comprehensive diagnosis model training module, and training is conducted in connection with a classification criteria according to the training sample data information and the ambient situation of the lesion outside the intestinal wall of the training set to obtain the second classification comprehensive diagnosis model.
|