Convolutional Neural Network Models for Content Based X-Ray Image Classification
P Arti, Abhishek Agrawal, A Adishesh, V M Lahari, Krupa B Niranjana · 2019
In this paper, transfer learning using pre-trained Convolutional Neural Network models such as GoogleNet and AlexNet is employed for X-Ray image classification and retrieval. This is carried out on IRMA database comprising of 55 categories of 6406 X-Ray images. These images are preprocessed using adaptive histogram equalization. The performance of the networks have been evaluated over five trials considering an 80:20 training to validation set split resulting in an average accuracy of 0.87 and 0.93 for GoogleNet and AlexNet, respectively. Confusion matrices are generated which is used to estimate the precision and recall of the networks. The proposed work resulted in a robust X-Ray image classification mechanism.