Data augmentation techniques for medical image analysis
Riferimenti FABRIZIO LAMBERTI
Riferimenti esterni Lia Morra
Gruppi di ricerca GR-09 - GRAphics and INtelligent Systems - GRAINS
Descrizione Recent advances in deep learning models have been largely attributed to the quantity and diversity of data gathered in recent years. Data augmentation is a strategy that enables practitioners to significantly increase the diversity of data available for training models, without actually collecting new data. Data augmentation techniques such as cropping, padding, and horizontal flipping are commonly used to train large neural networks for image analysis, and in more recent years adaptive techniques, such as AutoAugment or GAN-based approaches, have been proposed to increase the effectiveness of data augmentation strategies. Different data augmentation strategies are likely to perform differently depending on the type of input and visual task. For this reason, it is conceivable that medical imaging may require specific augmentation strategies that produce plausible data samples and allow effective regularization of deep learning model. Data augmentation may also be used to enhance specific classes that are under-represented in the training set, e.g. to generate artificial lesion samples. The goal of this thesis is to do an extensive systematic literature review in order to answer the following research questions: (i) which are the most commonly used and most effective techniques for different medical imaging tasks, (ii) which specific data augmentation techniques are available for medical images, (iii) what are the goals of data augmentation in the medical domain? Few papers are available that directly compare the efficacy of different data augmentation schemes; instead, authors usually report their chosen solution as part of the experimental details. We envision that in order to answer these research questions, a representative sample of papers published in reputable venues will be retrieved and analyzed to highlight trends in recent literature. If feasible, a comparison with data augmentation strategies proposed in the RGB domain will also be performed.The key findings of the systematic literature review may be complemented by few practical experiments in Keras/Tensorflow. Strong analytical, data analysis and writing (English) skills are required.
Vedi anche http://grains.polito.it/work.php
Scadenza validita proposta 06/02/2021 PROPONI LA TUA CANDIDATURA