An ultrasonic image processing system and method based on artificial intelligence

Abstract

This invention discloses an artificial intelligence-based ultrasound image processing system and method, belonging to the field of ultrasound imaging and intelligent image processing technology. The system includes an ultrasound signal acquisition and preprocessing module, an intelligent image processing module, and an output module. The ultrasound signal acquisition and preprocessing module is used to acquire ultrasound signals, perform analog-to-digital conversion, and perform digital preprocessing to generate ultrasound data for imaging. The intelligent image processing module converts the ultrasound data into images and inputs them into a pre-loaded AI model for real-time inference. The output module presents the results. The method includes steps such as signal acquisition and preprocessing, image generation and adjustment, AI model inference, and result output. This invention achieves efficient and real-time intelligent image processing within an embedded platform by tightly coupling and integrating the AI ​​inference acceleration module with the ultrasound acquisition and processing module. This significantly improves the accuracy and adaptability of image classification, segmentation, and enhancement, while optimizing system power consumption, stability, and engineering feasibility.

Description

Technical Field

[0001] This invention relates to the field of ultrasound imaging and intelligent image processing technology, specifically to an ultrasound image processing system and method that combines artificial intelligence algorithms with ultrasound signal acquisition and imaging processing, applicable to image classification, image enhancement, image segmentation, etc. in ultrasound equipment. Background Technology

[0002] Currently, ultrasound equipment generally employs a computing architecture centered on traditional embedded processors for image processing and analysis. These platforms primarily utilize low-power ARM processors, relying mainly on the CPU to execute ultrasound image processing algorithms in software. Examples include image enhancement, image classification, and image segmentation algorithms based on thresholding, gradient, variance, edge detection, erosion, and dilation. These traditional image processing methods are highly dependent on manual parameter settings and have limited adaptability to noise, tissue structure differences, and individual variations. Under complex tissue morphology or low signal-to-noise ratio imaging conditions, they are prone to inaccuracies and insufficient stability.

[0003] On the other hand, artificial intelligence has significant advantages in applications such as image segmentation, image enhancement, and image classification. It can automatically learn high-dimensional feature information through model training, demonstrating higher accuracy and applicability in complex image scenarios. However, in existing ultrasound equipment, due to the engineering implementation requirements of ultrasound equipment in terms of power consumption control, system stability, and R&D cycle, the selection of processing platforms usually tends to adopt mature low-computing-power embedded solutions, which are difficult to directly support efficient real-time inference of artificial intelligence models.

[0004] Although some existing solutions attempt to improve computing power by using general-purpose computing platforms or external computing devices, these solutions still have significant shortcomings in terms of power consumption control, real-time performance assurance, and system-level collaboration with ultrasound acquisition and processing modules. They are difficult to form a tightly coupled real-time processing architecture and are not conducive to engineering integration and widespread application in ultrasound equipment.

[0005] Therefore, how to introduce a processing platform with artificial intelligence model inference acceleration capabilities while meeting the engineering implementation requirements of ultrasonic equipment, and effectively combine it with the ultrasonic signal acquisition and imaging processing process to achieve efficient, accurate and intelligent processing of ultrasonic images, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the problems that existing ultrasound equipment generally relies on low-computing-power embedded processing platforms for image processing and analysis, making it difficult to efficiently run artificial intelligence models, or that while external general-purpose computing platforms can improve computing power, they suffer from significant shortcomings in power consumption control, real-time performance assurance, and system-level collaboration, this invention proposes an artificial intelligence-based ultrasound image processing system and method.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] An artificial intelligence-based ultrasound image processing system includes:

[0009] The ultrasonic signal acquisition and preprocessing module is used to control the ultrasonic probe to complete the transmission and reception of ultrasonic signals, perform analog-to-digital conversion on the received analog ultrasonic echo signals, and perform digital signal processing on the digital ultrasonic signals through programmable logic devices to generate ultrasonic data for imaging.

[0010] The intelligent image processing module is communicatively connected to the ultrasound signal acquisition and preprocessing module. It is used to perform digital scanning transformation on the ultrasound data to generate ultrasound images, and after format conversion and size adjustment of the ultrasound images, it inputs them into a pre-loaded artificial intelligence model for inference calculation. The intelligent image processing module is implemented by a processor with neural network acceleration capability to support the efficient real-time inference of the artificial intelligence model.

[0011] The output module is used to present the results of the inference calculation or ultrasound images.

[0012] Furthermore, the ultrasonic signal acquisition and preprocessing module includes an ultrasonic transmitting and receiving circuit, an analog-to-digital conversion unit, and a programmable logic device. The programmable logic device is used to perform at least one digital signal processing operation among dynamic focusing, digital filtering, decimation, and logarithmic compression.

[0013] Furthermore, the artificial intelligence model is an image segmentation model, an image classification model, or an image enhancement model.

[0014] Furthermore, the intelligent image processing module is also used to perform at least one operation among system peripheral control, user interface management, or acquisition parameter distribution based on the inference calculation results.

[0015] Furthermore, the various functional modules within the system communicate and collaborate through a data and command communication bus, forming an integrated and tightly coupled real-time processing architecture.

[0016] An artificial intelligence-based ultrasound image processing method includes the following steps:

[0017] S1: Control the ultrasound probe to emit ultrasound signals and receive echo signals, perform analog-to-digital conversion and digital preprocessing on the echo signals, and generate ultrasound data;

[0018] S2: Perform digital scanning transformation processing on the ultrasound data to generate an ultrasound image, and perform format conversion and size adjustment on the ultrasound image;

[0019] S3: In a processor with neural network acceleration capabilities, the adjusted ultrasound image is input into a pre-loaded artificial intelligence model for inference calculation to obtain the inference result;

[0020] S4: Output the reasoning result and / or the corresponding ultrasound image.

[0021] Furthermore, the digital preprocessing in step S1 includes at least one of the following operations: dynamic focusing, digital filtering, decimation, and logarithmic compression.

[0022] Furthermore, the artificial intelligence model mentioned in step S3 is an image segmentation model, an image classification model, or an image enhancement model.

[0023] Furthermore, the inference computation described in step S3 is performed with the support of the processor's built-in hardware acceleration unit.

[0024] Furthermore, after step S4, the procedure also includes the step of performing at least one of system control, parameter adjustment, or human-computer interaction operation based on the reasoning result.

[0025] This invention integrates an intelligent processing module with neural network hardware acceleration capabilities within an ultrasound device, enabling efficient real-time inference of AI models on the module. This significantly improves the accuracy and adaptability of ultrasound images in tasks such as classification, segmentation, and enhancement, and is particularly suitable for complex tissues and low signal-to-noise ratio scenarios. Simultaneously, this integrated architecture effectively controls system power consumption and complexity while ensuring high-performance processing, reducing reliance on external computing devices and sensitivity to manual parameter settings. This enhances system stability and engineering feasibility, providing an integrable and easily scalable solution for intelligent, real-time image processing in ultrasound devices. Attached Figure Description

[0026] Figure 1 : Overall structural diagram of the image processing system described in this invention. Detailed Implementation

[0027] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0028] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0029] This invention introduces an intelligent image processing module with artificial intelligence model inference acceleration capabilities into the ultrasound equipment, and designs it in a tightly coupled and collaborative manner with the ultrasound signal acquisition and preprocessing module. Under the premise of meeting the engineering implementation requirements of the ultrasound equipment, it realizes the real-time generation and efficient intelligent analysis and processing of ultrasound images, as well as the control of system peripherals, user interface management, or the distribution of acquisition parameters.

[0030] like Figure 1 As shown, an artificial intelligence-based ultrasound image processing system includes an ultrasound signal acquisition and preprocessing module, an intelligent image processing module, and an output module.

[0031] The ultrasonic image processing system of this invention enables various functional modules to communicate and coordinate through data and command communication buses, forming an integrated and tightly coupled real-time ultrasonic image processing architecture. This architecture effectively controls overall power consumption while improving computing power, overcoming the shortcomings of external general-purpose computing platforms in terms of power consumption, heat dissipation, and integration. The overall structure of this system can realize a continuous real-time processing flow of ultrasonic data from acquisition and imaging to artificial intelligence model inference, display, and interaction.

[0032] The ultrasonic signal acquisition and preprocessing module is used to acquire ultrasonic signals from the object under test. Its functions include, but are not limited to: controlling the ultrasonic probe to complete the transmission and reception of ultrasonic signals; performing analog-to-digital conversion on the received analog ultrasonic echo signals; and receiving the digitized ultrasonic signals by a programmable logic device (FPGA) to perform digital signal processing operations such as dynamic focusing, digital filtering, decimation, and logarithmic compression to form ultrasonic data for subsequent imaging processing.

[0033] The ultrasonic signal acquisition and preprocessing module is composed of ultrasonic transmitting and receiving circuits, an analog-to-digital conversion unit, and a field-programmable logic device (FPGA). This module provides stable and repeatable ultrasonic data input to meet the real-time and accuracy requirements of subsequent intelligent image processing.

[0034] The intelligent image processing module is used to generate images and perform model inference on ultrasound data from the ultrasound signal acquisition and preprocessing module. Its functions include, but are not limited to: performing digital scanning transformation processing on ultrasound data to generate two-dimensional or multi-dimensional ultrasound images; performing format conversion and size adjustment on the generated ultrasound images; and inputting the processed ultrasound images into a pre-loaded artificial intelligence model for inference calculation to obtain inference results.

[0035] The intelligent image processing module can directly use or post-process the inference results to complete systematic tasks such as system peripheral control, UI management, and distribution of acquisition parameters.

[0036] The artificial intelligence model can be an image segmentation model, an image classification model, an image enhancement model, or other artificial intelligence-based models suitable for ultrasound image applications.

[0037] The intelligent image processing module is implemented by a processor with neural network acceleration capabilities. With the support of the hardware acceleration unit, the intelligent image processing module completes efficient inference calculations of artificial intelligence models without relying on external computing devices or cloud processing, thereby realizing real-time, low-power intelligent image processing inside the ultrasound equipment.

[0038] By introducing an intelligent image processing module with artificial intelligence model inference acceleration capabilities, the ultrasound image processing system can automatically learn and extract high-dimensional features from ultrasound images, significantly improving the accuracy, robustness, and adaptability in tasks such as image classification, enhancement, and segmentation, and performing more stably, especially under complex tissue structures or low signal-to-noise ratio conditions.

[0039] The output module is used for result presentation, and its functions include, but are not limited to:

[0040] The calculation results or ultrasound images can be displayed via a display device or output to the outside via a communication interface.

[0041] This invention also provides an artificial intelligence-based ultrasound image processing method, comprising the following steps:

[0042] S1: Control the ultrasound probe to emit ultrasound signals to the object under test and receive the corresponding echo signals, perform analog-to-digital conversion on the echo signals, and perform digital preprocessing operations, including but not limited to dynamic focusing, digital filtering, decimation and logarithmic compression processing, to generate ultrasound data for imaging processing;

[0043] S2: Perform digital scanning transformation processing on the ultrasound data preprocessed in step S1 to convert the ultrasound data into two-dimensional or multi-dimensional ultrasound images, and perform format conversion and size adjustment on the ultrasound images to meet the input requirements of the artificial intelligence model.

[0044] S3: Inside a processor with neural network acceleration capabilities, the ultrasound image obtained in step S2 is input into a pre-loaded artificial intelligence model for inference calculation, and the segmentation result, image enhancement result, or feature parameters of the target region are obtained with the support of the hardware acceleration unit.

[0045] S4: The reasoning results obtained in step S3 are used directly or post-processed, and the processing results and corresponding ultrasound images are displayed through a display device or output to an external system through a communication interface to realize human-computer interaction or data transmission.

[0046] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An ultrasound image processing system based on artificial intelligence, characterized in that, include: The ultrasonic signal acquisition and preprocessing module controls the ultrasonic probe to complete the transmission and echo reception of ultrasonic signals, performs analog-to-digital conversion on the received analog ultrasonic echo signals, and performs digital signal processing on the digital ultrasonic signals through a programmable logic device to generate ultrasonic data for imaging. The intelligent image processing module is communicatively connected to the ultrasound signal acquisition and preprocessing module. It is used to perform digital scanning transformation on the ultrasound data to generate ultrasound images, and after format conversion and size adjustment of the ultrasound images, it inputs them into a pre-loaded artificial intelligence model for inference calculation. The intelligent image processing module is implemented by a processor with neural network acceleration capability to support the efficient real-time inference of the artificial intelligence model. The output module is used to present the results of the inference calculation or ultrasound images.

2. The ultrasound image processing system according to claim 1, characterized in that, The ultrasonic signal acquisition and preprocessing module includes an ultrasonic transmitting and receiving circuit, an analog-to-digital conversion unit, and a programmable logic device. The programmable logic device is used to perform at least one digital signal processing operation among dynamic focusing, digital filtering, decimation, and logarithmic compression.

3. The ultrasound image processing system according to claim 1, characterized in that, The artificial intelligence model is an image segmentation model, an image classification model, or an image enhancement model.

4. The ultrasound image processing system according to claim 1, characterized in that, The intelligent image processing module is also used to perform at least one operation among system peripheral control, user interface management, or acquisition parameter distribution based on the inference calculation results.

5. The ultrasound image processing system according to claim 1, characterized in that, The various functional modules within the system communicate and collaborate through a data and command communication bus, forming an integrated and tightly coupled real-time processing architecture.

6. An ultrasound image processing method based on artificial intelligence, characterized in that, Includes the following steps: S1: Control the ultrasound probe to emit ultrasound signals and receive echo signals, perform analog-to-digital conversion and digital preprocessing on the echo signals, and generate ultrasound data; S2: Perform digital scanning transformation processing on the ultrasound data to generate an ultrasound image, and perform format conversion and size adjustment on the ultrasound image; S3: In a processor with neural network acceleration capabilities, the adjusted ultrasound image is input into a pre-loaded artificial intelligence model for inference calculation to obtain the inference result; S4: Output the reasoning result and / or the corresponding ultrasound image.

7. The ultrasound image processing method according to claim 6, characterized in that, The digital preprocessing described in step S1 includes at least one of the following operations: dynamic focusing, digital filtering, decimation, and logarithmic compression.

8. The ultrasound image processing method according to claim 6, characterized in that, The artificial intelligence model mentioned in step S3 is an image segmentation model, an image classification model, or an image enhancement model.

9. The ultrasound image processing method according to claim 6, characterized in that, The inference calculation described in step S3 is performed with the support of the processor's built-in hardware acceleration unit.

10. The ultrasound image processing method according to claim 6, characterized in that, Step S4 is followed by the step of performing at least one of system control, parameter adjustment, or human-computer interaction based on the reasoning result.

Images