Daniel Moyer
Postdoctoral Associate, MIT Computer Science & Artificial Intelligence Lab

SENSE.nano 2021
Tuesday, October 26
Session 3: Imaging
2:25 PM - 2:40PM EDT

Abstract
This talk will cover our recent MICCAI 2021 paper. We demonstrate an object tracking method for 3D volumetric images with fixed computational cost and state-of-the-art performance. Previous methods predicted transformation parameters from convolutional layers. We instead propose an architecture that neither flattens convolutional features nor uses fully connected layers, but instead relies on equivariant filters to preserve transformations between inputs and outputs (e.g., rotations/translations of inputs rotate/translate outputs). The transformation is then derived in closed form from the outputs of the filters. This method is useful for applications requiring low latency, such as real-time tracking. We demonstrate our model on synthetically augmented adult brain MRI, as well as fetal brain MRI, which is the intended use-case.

Biography
Daniel Moyer is a posdoctoral associate at MIT CSAIL working with Polina Golland. He completed his doctorate in computer science at University of Southern California under Greg Ver Steeg and Paul Thompson. His research is focused on machine learning and its applications to medical imaging.