PROJECT TITLE :
Classification of Volumetric Images Using Multi-Instance Learning and Extreme Value Theorem
Medical practitioners use volumetric imaging as a diagnostic tool. Popular approaches such as convolutional neural networks (CNN) can only be used for volumetric Image Processing if training data and GPU memory are readily available. During the training phase, the volumetric image classification problem is posed as a multi-instance classification problem and a unique strategy is proposed to adaptively select positive instances from positive bags of positive examples These images without a pathology are modelled using the extreme value theory and then used to detect positive occurrences of a pathology. Using three separate image classification tasks (i.e., classify retinal OCT images according to the presence or absence of fluid buildups in retinal OCT images, pulmonary 3D-CT images, and histopathology 2D images) the experimental results show that the proposed method produces classifiers with similar performance to fully supervised methods and achieves the state of the art performance in all examined test cases..
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here