ANN for Lung Diseases Recognition

Greetings, friends! Here is an update from our programmers!

In our previous blog post we have told you about the creation of a neural network, which can be used to diagnose lung diseases. Here is some additional information regarding this ANN.

Recently we have decided to take a deep look into the computer recognition of lung diseases, when using chest X-rays. This task is quite challenging even for radiologists. Our team has analyzed the most recent developments in this field and we used the neural network, created in Stanford University as a basis for our own ANN. We have changed the design of the neural network by adding additional layers and by using higher resolution images. To make training process faster we have pre-processed the original data. In the end, after several days of training, we got results comparable to the State of the Art.

Dmitry Musinov — Skychain team member, neural network developer

Today this neural network can diagnose the following diseases: Atelectasis, Cardiomegaly, Effusion, Infiltration, Mass, Nodule, Pneumonia, Pneumothorax, Consolidation, Edema, Emphysema, Fibrosis, Pleural Thickening, Hernia.

This graph shows the amount of errors the neural network has, when examining the special test x-rays (these x-rays were never used for ANN trainig)

The accuracy of neural network varies for each disease, but the overall accuracy is ~81–82%.