Convolutional Neural Network for Categorization of Lung Tissue Patterns in Interstitial Lung Diseases

In computer aided diagnosis (CAD), automatic tissue categorization is very crucial part. Deep learning methods provide excellent results in the area where medicinal image analysis is required. Here, we have propose and develop a framework in which Convolutional Neural Network (CNN) will be used for tissue categorization of Interstitial Lung Diseases. The planned framework made up of 5 convolutional layers and 2 fully connected layers. The last layer provide different outputs of the tissue patterns such as healthy, Honeycombing, Ground Glass Opacity (GGO), Reticulation etc. We used a dataset of 100 HRCT scan which are collected from different radiology centers for training and evaluation of the system. In future we can use a three- dimensional images of the CT scans and also we can integrate this system into a computer aided diagnosis (CAD) system which will assist radiologist for better diagnosis.