Previous Page  10 / 26 Next Page
Information
Show Menu
Previous Page 10 / 26 Next Page
Page Background

Florian Dubost

is a PhD student at the

Erasmus Medical

Center, Rotterdam

. This Friday afternoon he will present

his work at the

LABELS

workshop.

Hands-Free Segmentation Of

Medical Volumes Via Binary Inputs

Florian Dubost

Presentation

10

MICCAI Daily: Thursday

Florian proposes a novel hands-free method to interactively segment 3D

medical volumes. In his scenario, a human user progressively segments an

organ by answering a series of questions of the form ``

Is this voxel inside the

object to segment?

''. At each iteration, the chosen question is defined as the

one halving a set of candidate segmentations given the answered questions.

For a quick and efficient exploration, these segmentations are sampled

according to the Metropolis-Hastings algorithm. His sampling technique

relies on a combination of relaxed shape prior, learnt probability map and

consistency with previous answers. He demonstrates the potential of his

strategy on a prostate segmentation MRI dataset. Through the study of

failure cases with synthetic examples, he demonstrate the adaptation

potential of his method. He also shows that his method outperforms two

intuitive baselines: one based on random questions, the other one being

the thresholded probability map. Visit his poster on Friday at 17:15 at

LABELS.

For a quick and efficient exploration, these

segmentations are sampled according to

the Metropolis-Hastings algorithm