**4.1 ECNN algorithm**

Two trials of one or the other CC or MLO seen should be adjusted utilizing the picture enlistment method. At that argument, a dissimilar picture was received by removing the former trial out of the existing trial and subsequently scaled to the fullrange force. The territorial pictures from the refined district proposition are trimmed from the three pictures and scaled to 224 224 3 for each picture, which are utilized for ECNN is floodlight extraction. The three channels are rehashed from one-channel grayscale pictures (e.g., the current sweep of 224 224 1) since the pertained ECNN and ECNN models expect 3-channel pictures. Multi-measurements of three-state in floodlights (from earlier sweep, current output, and contrast pictures) are made to prepare a CNN model. For instance, The ECNN is floodlights utilizing ResNet-60 V3 of 2048 3 measurements for each view (CC or MLO) of a subject's side (left or right bosom). Remember that earlier sweep consistently relates to the ordinary (sound) status in any event, for a destructive subject. Assume we code sound and carcinogenic as 0 and 1 individually, at that point the ground realities (yields) compared to the three states (earlier, current, distinction) of the destructive view are [0 1 1]. This coding instrument can be handily stretched out to at least two earlier sweeps.

### **4.2 Algorithm**

The following is the ECNN algorithm steps:

The Omicron disease infection data index, i.e., the absolute 522 pictures, our experiment involved the related following steps:

1. Introduces mandatory collection.

2. Introduces training dataset.

3.Executes in the floodlight ordering of change data.

4.Alignment with 70-time segments and 2 yield.

5. Introduces Keras (Keras is a Deep Learning library).

6.Resets ECNN.

7.Enhances ERNN part & about regulation of loss calculation function.

8. Improvement of yield part.

9.Adds the ECNN.

*COVID-19 Data Analytics Using Extended Convolutional Technique DOI: http://dx.doi.org/10.5772/intechopen.106999*


**Figure 1.** *Input dataset of the projected prototype for COVID-19 disease detection.*

10.Adjusts ECNN in the assessment dataset.

11.Loads the Omicron disease infection test image data of the year 2020.

12.Become a predicted Omicron disease infection in Dec 2019.

13. Imagine aftereffects with anticipated or genuine Omicron disease infection.

**INPUT DATSET:** Here the input dataset is having 16 columns with target class, i.e., severity level of the COVID-19 disease consisting the database of 282 sample x-ray images (**Figure 1**).
