Method and system for modeling features based on multimodal sono-omics

Abstract

The present invention discloses a method and system for modeling based on multimodal ultrasonic omics features, including: acquiring several ultrasonic data of lesion parts to obtain multimodal ultrasonic omics data (including "time" mode); Multiple characteristic data are screened out from the modal sono-omics data; two-dimensional convolution and three-dimensional convolution are performed on the multiple characteristic data through the feature engineering network, and according to the obtained three-dimensional convolution and the change rate difference of the multiple characteristic data , analyze and obtain the best feature combination; according to the best feature combination, a feature combination model is established. Compared with the traditional feature modeling method, the present invention performs two-dimensional convolution and three-dimensional convolution on multiple feature data of multi-modal ultrasound data through feature engineering network, and analyzes the three-dimensional convolution and change rate of multiple feature data After the difference, a number of feature data that are complementary in space and time are obtained as the best feature combination for modeling, so as to solve the problem of model overfitting and more truly reflect the overall picture of the disease.

Description

technical field

[0001] The present invention relates to the technical field of ultrasound medicine, in particular to a method and system for modeling based on multimodal ultrasound omics features. Background technique

[0002] With the advent of precision medicine and big data, it has become one of the core links of clinical medicine to analyze the lesion by establishing a characteristic model of ultrasound data. In the prior art, the single-mode ultrasound data of the lesion is usually extracted, and a feature model is established by training the single-mode ultrasound data through deep learning. However, the actual disease outcome is affected by the interaction and influence of different pathophysiology, hemodynamics, space and time states, so the characteristic model established by using single modality cannot be accurately used for disease analysis, and when the single modality medical image When the number is limited, using a deep learning network for feature modeling ...

Examples

Embodiment Construction

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

[0028] see figure 1 .

[0029] see figure 1 , is a schematic flow chart of an embodiment of the method for modeling based on multimodal sono-omics features provided by the present invention, such as figure 1 As shown, the analysis method includes steps S11 to S14. Each step is as follows:

[0030] Step S11: Obtain a number of ultrasonic data of the lesion to form multimodal ultrasonic omics data.

[0031] Step S12: image data processing is performed on th...

technical field

[0001] The present invention relates to the technical field of ultrasound medicine, in particular to a method and system for modeling based on multimodal ultrasound omics features. Background technique

[0002] With the advent of precision medicine and big data, it has become one of the core links of clinical medicine to analyze the lesion by establishing a characteristic model of ultrasound data. In the prior art, the single-mode ultrasound data of the lesion is usually extracted, and a feature model is established by training the single-mode ultrasound data through deep learning. However, the actual disease outcome is affected by the interaction and influence of different pathophysiology, hemodynamics, space and time states, so the characteristic model established by using single modality cannot be accurately used for disease analysis, and when the single modality medical image When the number is limited, using a deep learning network for feature modeling ...

Examples

Embodiment Construction

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

[0028] see figure 1 .

[0029] see figure 1 , is a schematic flow chart of an embodiment of the method for modeling based on multimodal sono-omics features provided by the present invention, such as figure 1 As shown, the analysis method includes steps S11 to S14. Each step is as follows:

[0030] Step S11: Obtain a number of ultrasonic data of the lesion to form multimodal ultrasonic omics data.

[0031] Step S12: image data processing is performed on th...

Images