2021 · 10 citations · 6 references
ComputingBiomedical EngineeringComputer ScienceComputer VisionDementiaMachine LearningAlzheimer DiseaseDeep LearningGenerative ModelingNeural NetworksNeurodegenerative DiseasesRepresentation LearningConvolutional Neural Network DetectionConvolutional Neural NetworksDeep Learning In RadiologyDeep Learning-based ClassificationDiagnostic StagesGenerative Adversarial NetworksGenerative Image SynthesisMachine Learning In Neuroimaging
In this paper, an unsupervised generative modeling method that generates synthetic images with the help of Deep Convolutional Generative Adversarial Networks (DCGANs). A method has been put forward that uses labelled Magnetic Resonance Imaging (MRI) dataset and then applies DCGAN on the limited amount of data, hence enlarging the dataset size as well as its diversity through the use of GAN. Further the synthetic images and training dataset are merged together and trained into a Convolutional Neural Network (CNN) and its different architectures for the classification of four different stages of Alzheimer's disease. The classification performance using CNN yielded an accuracy of 69%, ResNet50 yielded respectively. On increasing the number of epochs as well as the number of dataset images more accuracy can be obtained.
6
Combining Noise-to-Image and Image-to-Image GANs: Brain MR Image Augmentation for Tumor Detection
Changhee Han, Leonardo Rundo, Ryosuke Araki et al. · IEEE Access · 2019 · 249 citations · Full text