Fig. 1
From: A multi-label classification model for full slice brain computerised tomography image
Example of the proposed model. We propose an architecture that utilizes a convolutional neural network (CNN; VGG16) to learn image features from a variable-length series of brain CT scans and that uses a recurrent neural network (RNN; GRU) to learn slice dependencies for predicting nine categories of brain CT images simultaneously. The nine categories of brain CT images are denoted as intracranial haemorrhage, intraparenchymal, intraventricular, subdural, extradural, subarachnoid, calvarial fracture, mass effect, and midline shift