Ultrasound image enhancement processing system and method thereof

An ultrasound image enhancement processing system, comprising:

• at least one first ultrasound device, used to generate a first ultrasound original image file;
• at least one second ultrasound device, used to generate a second ultrasound original image file;
• at least one server apparatus, used to receive the first ultrasound original image file and the second ultrasound original image file, wherein the server apparatus at least comprises:
• a speckle reduction processing module, used to process the first ultrasound original image file through a speckle reduction algorithm to generate a first processed image file;
• a first neural network learning module, connected to the speckle reduction processing module, performing a deep learning training on the first ultrasound original image file and the first processed image file to generate a first neural network model;
• a first speckle reduction enhancement output module, connected to the first neural network learning module, used to input the first ultrasound original image file into the first neural network model to output a first speckle reduction enhancement image file, and input the second ultrasound original image file into the first neural network model to output a second processed image file;
• an image enhancement processing module, connected to the first speckle reduction enhancement output module, used to process the second processed image file through an image enhancement algorithm to generate a third processed image file;
• a second neural network learning module, connected to the first speckle reduction enhancement output module, used to perform a deep learning training on the second ultrasound original image file and the third processed image file to generate a second neural network model;
• a second speckle reduction enhancement output module, connected to the second neural network learning module, used to input the second ultrasound original image file into the second neural network model to output a second speckle reduction enhancement image file;
• and
• a training correction module connected to the speckle reduction processing module, the first neural network learning module, the first speckle reduction enhancement output module, the image enhancement processing module, the second neural network learning module and the second speckle reduction enhancement output module, wherein the training correction module is used to calculate a loss function respectively for the first processed image file, the first speckle reduction enhancement image file, the third processed image file and the second speckle reduction enhancement image file after normalizing the first processed image file, the first speckle reduction enhancement image file, the third processed image file and the second speckle reduction enhancement image file, wherein the loss function is calculated by dividing the image into small blocks and then comparing the structural similarity to evaluate the first neural network model and a learning error value of the first neural network model.