The Information Fusion Lab at UMass Amherst CICS focuses on ML for multimodal data, including methods that support a wide range of biomedical applications. Our research includes a wide variety of topics including deep learning for the fusion of multi-resolution time series, images and structured information, the incorporation of domain knowledge or saliency in imaging and integration of multiple views for MRI analysis. Most recently, we introduced new methods for normalizing flows and transfer of causal models. Please see our research page for a full list of projects and our GitHub page for code releases.
Lab home page: https://groups.cs.umass.edu/infofusion/home/
The objective of this project is to automatically determine congenital heart conditions from phase-contrast heart MRIs and their correlation with long-term clinical outcomes. We use data from the UK biobank. Currently, we are working on using the long-axis view from cardiac MRI sequences to detect mitral valve regurgitation. The framework we introduced includes a sequence classification model, an image segmentation model and an ensemble model. Given that the severity of mitral valve regurgitation varies and the pathology is visible across 3 different views of cardiac MRI sequences, we are working towards a multi-view sequence classification model for detecting mitral valve regurgitation. We also use weak supervision for heart chamber segmentation. To better use the information contained in the images, we are developing weakly-supervised models to segment the unlabeled datasets from the UK Biobank. We are working toward a weakly-supervised or unsupervised solution to extract salient information from cardiac MRI sequences, which will be useful in other medical imaging applications.
The prerequisites are:
intermediate programing skills
logical reasoning ability
mathematics skills (basic linear algebra, calculus, statistics and probability)
Here are some examples of skills you will gain:
how to design and implement models for medical image segmentation and disease diagnosis/prediction
how to appropriately evaluate machine learning models (how to design experiments, select test sets, perform cross-validation)
how to optimize models, how to efficiently train and test them