IPAB workshop - 12/01/2017

Febrian Rachmadi  will be speaking.

Title:

Spatial Information in Deep Neural Network for MR Brain Image Analysis: Dementia's White Matter Hyperintensities Segmentation

Abstract:

In this talk, we will see how deep neural network is used in medical image analysis, especially in MR brain image analysis. Brain image analysis can be very interesting and challenging because of the nature of MR image, i.e.

multi-channel high-resolution 3D data with grayscale intensity. Furthermore, biomarker of pathology shown in MR image (i.e. hyperintensities) also correlates highly with spatial information (i.e. location in the brain) which makes further modification of deep neural network is needed. We will see some related works on how spatial information can be incorporated into deep neural network in brain image analysis.

Dylan Ross will be speaking:

Title

Mechanics of exploration in larval Drosophila

Abstract:

The larval form of the fruitfly, Drosophila melanogaster, offers a simple model system with which to study the biological basis of intelligent behaviour. When placed on a planar substrate, the animal executes a stereotypical exploratory routine which appears to consist of straight runs punctuated by turning events. The current view is that exploration results from a decision making process in the brain which determines transitions between discrete running and turning states.

In this talk I will present a new model of larval exploration which is grounded in the animal's body mechanics. The model "brain" is constructed to satisfy some basic physical requirements on locomotion. It neither senses nor directly controls turning motions, and does not explicitly perform any decision making function. Nevertheless, the behaviour of the simulated animal appears similar to that of the real larva. I will use the model to argue that the exploratory routine may represent a deterministic diffusion process arising primarily from the chaotic dynamics of the larval body.