A New Approach for the Design of Medical Image ETL Using CNN.

Nowadays, the combination of digital images and machine learning techniques to solve COVID-19 problems has been one of the most explored elements. Most efforts have focused on the detection and classification of lung diseases, which requires a large amount of images to process. Extracted images from different sources need to be loaded into big data base after required transformation to reduce error and minimize data loss. This process is also known as Extraction-Transformation-Loading (ETL). It is responsible for extracting, transforming, conciliating, and loading data for supporting decision-making requirements. This paper provides the innovative approach of using an images extract, transform, load (MI-ETL) solution, to provide a large number of images of interest from heterogeneous data sources into a specialized database. The main objective of the paper is to present the three stages of the MI-ETL process starting with the collection of medical images from several sources using different techniques. Then, applying deep learning techniques (CNN filter) to extract only images of the lungs, and finally loading the features of the images in a big database.