Assessment of deep convolutional neural networks for medical image segmentation

Max Kivits
Presentation BSc presentation
Date 2018-11-29
Time 16:00
Location Carré 3446

Early detection of melanoma is difficult for the human eye but a crucial step towards reducing its death rate. Computerized detection of these melanoma and other skin lesions is necessary. The central research question of this presentation is "How to segment skin lesion images using a neural network with low available data?” This question is divided into three sub questions regarding best performing network structure, training data and training method.

In an attempt to answer these questions relevant theory will be discussed. Two experiments have been conducted. The first experiment trains a network on data sets of different size. The second experiment proposes twelve network structures and trains them on the same data set. From the theory and experimental results an optimal network structure for skin lesion segmentation will be proposed.

Posted on Monday, November 5, 2018