An Efficient Multilevel Thresholding Scheme for Heart Image Segmentation Using a Hybrid Generalized Adversarial Network

Pseudocode 1

Pseudocode for segmentation.

Input: Load the Heart CT_Scan Images, HCTSCAN

Output: Segmentation of the images based on the threshold values

Begin:

1. Define the hyper parameters like learning rate, batch size, and number of epochs

2. Define callbacks with the following estimators

i. model checkpoint to store the best model after compilation

 ii. set the CSV logger to save the best scores acquired by the model

  iii. define the Tensor Flow board with an early stopping mechanism to reduce the validation loss

3. Fit and compile the training dataset by shuffling the records

4. i. if threshold value<= length(masked_image) then

  Create a predicted mask to have a segmented image

 ii. else expand the image to concatenate with previous ROI bounds

5. store the images in the necessary directory

End