Offline Results

This model gives results under 4 title: Normal, Pneumonia, COVID-19, Other Diseases.

Other Diseases may include 11 different diseases other than pneumonia and COVID-19, observed in the lung. (Nodule, Infiltration, Effusion, Atelectasis, Pneumothorax, Mass, Pleural_Thickening, Cardiomegaly, Fibrosis, Emphysema, Consolidation)

Anonymized Samples from datasets used:

Confusion Matrix

Tests made with 1693 different patient X-Ray Images and got accuracy of 97% for classifying 4 class.

Confusion Matrix between only Normal & Covid-19 Cases

Tests made with 600 different patient X-Ray Images and got accuracy of 99% for classifying 2 class. It classified all of the Covid-19 cases correctly and misclassified only 6 of normal images among 500.

Datasets for training model are retrieved from:

How it works?

This software determines whether the X-Ray image you have uploaded is from a normal image, pneumonia, Covid-19, or another lung disease. It makes a classification by working together of different Deep Learning Models trained with various X-Ray images.

ATTENTION: Although the models have high accuracy rates, they are never 100% accurate and have not been medically proven. For this reason, its use for diagnosis is not yet suitable.

Home