MA/IDP: Diverse Anomaly Detection Projects

Contact

Rami Eisawy

Introduction

In recent years, we have seen a rapid increase in the number of deep learning algorithms in the medical domain that for a given task surpass the performance of trained professionals. Such deep learning algorithms are supervised using large carefully annotated datasets. In the medical domain it is not really possible to make such large datasets for all kinds of diseases. The problem of diagnosing a wide range of diseases in these resource constrained environments is to use some form of unsupervised anomaly detection.

Unsupervised Anomaly Detection for disease requires estimating healthy anatomy normative distribution from a large set of healthy examples and to detect outliers at inference time from either a learned feature space or from high reconstruction loss.

Project 1 Objective (MR Brain)

A single imaging sequence will be selected (i.e. FLAIR). Current state of the art methods rely on autoencoder reconstruction based pipelines, however clinical success is limited here and this project would focus on supplementing and/or improving this approach through novelty.

Your Challenge:

• Literature review of various papers in the field of unsupervised anomaly detection from image reconstruction error methods
• Implementation of the SOTA for deep learning based MR image reconstruction
• Incorporation of self-supervised methods to supplement the reconstruction based approach
• Improve performance of the current SOTA

Project Pool Please reach out for information on a specific topic

• anomaly detection in various domains are available for interested and motivated students. Localisation and/or segmentation tasks on various forms of medical data - CT/MRI (head) brain data is generally available, however access to other anatomical data is also possible
• (Spatio)temporal divergence detection modeling (i.e. anomaly detection on evolving diseases (for ex. ADNI), (EEG data).

Requirements

• Good programming skills: Python, Unix, Dockerisation
• Knowledge of image processing and data analysis
• Sound knowledge about deep learning, linear algebra and probability theory

Literature

• MOOD-MICCAI 2020 challenge top performer https://arxiv.org/abs/2011.04197
• Incorporating transformers with reconstruction based approaches https://arxiv.org/abs/2102.11650
• A comparative study: https://arxiv.org/abs/2004.03271
• ADNI http://adni.loni.usc.edu
• EEG Data https://www.fieldtriptoolbox.org/faq/open_data