Please join us for a UW Data Science Seminar on Tuesday, November 28th from 4:30 to 5:20 p.m. PST. The seminar will feature Dr. Eleftherios Garyfallidis, an Associate Professor of Intelligent Systems Engineering (ISE) at Indiana University.
This event will take place in the Physics/Astronomy Auditorium 102 (PAA A102) on the University of Washington campus.
“AI breakthroughs for pre-processing structural and diffusion MRI data”
Abstract: During my talk I will focus on two novel approaches for segmentation and denoising.
Segmentation: Brain extraction is a computational necessity for researchers using brain imaging data. However, the complex structure of the interfaces between the brain, meninges and human skull have not allowed a highly robust solution to emerge. While previous methods have used machine learning with structural and geometric priors in mind, with the development of Deep Learning (DL), there has been an increase in Neural Network based methods. Most proposed DL models focus on improving the training data despite the clear gap between groups in the amount and quality of accessible training data between. We propose an architecture we call Efficient V-net with Additional Conditional Random Field Layers (EVAC+). EVAC+ has 3 major characteristics: (1) a smart augmentation strategy that improves training efficiency, (2) a unique way of using a Conditional Random Fields Recurrent Layer that improves accuracy and (3) an additional loss function that fine-tunes the segmentation output. We compare our model to state-of-the-art non-DL and DL methods.
Denoising: Patch2Self (P2S), which performs self-supervised denoising of dMRI data using the statistical independence of noise, has previously shown state-of-the-art results by performing a series of regression analyses on the so-called Casorati matrix. P2S however is resource intensive, both in terms of running time and memory usage. This work exploits the redundancy imposed by P2S to alleviate its performance issues and inspect regions that influence the noise disproportionately. Specifically, this study makes a two-fold contribution: (1) We present Patch2Self2 (P2S2), a method that uses matrix sketching to perform self-supervised denoising. By solving a sub-problem on a smaller sub-space, so called, coreset, we show how P2S2 can yield a significant speedup in training time while using less memory. (2) We show how the so-called statistical leverage scores can be used to interpret the denoising of dMRI data, a process that was traditionally treated as a black-box. Our experiments are conducted on simulated and real data and clearly demonstrate that P2S2 does not lead to any loss in denoising quality, while providing significant speedup and improved memory usage by training on only a small fraction of the data.
Biography: Dr. Garyfallidis holds the position of Associate Professor of Intelligent Systems Engineering (ISE) at Indiana University (IU). Prof. Garyfallidis works on the interface between machine learning, medical imaging and engineering visualization. He is the inventor of multiple ground breaking algorithms including QuickBundles. QuickBundles was the first fast and unsupervised algorithm in neuroimaging for grouping tractographies using streamlines. Prof. Garyfallidis is the inventor of SLR. SLR is the most accurate method for affinely registering bundles or tractograms. His research has been fundamental in understanding the challenges of brain tractography. Due to a method called RecoBundles, in 2015 Garyfallidis enabled the evaluation of tractographies in data with distortions. Prof. Garyfallidis is the founder and lead engineer of DIPY. The pioneering work that Dr. Garyfallidis started and is today championed by his graduate students. See for example his labs work on Patch2Self denoising, EVAC+ and Bundle Analytics (BUAN). Prof. Garyfallidis is organizing yearly workshops (see DIPY workshops) to train faculty and students to use the latest methods in neuroimaging.
The UW Data Science Seminar is an annual lecture series at the University of Washington that hosts scholars working across applied areas of data science, such as the sciences, engineering, humanities and arts along with methodological areas in data science, such as computer science, applied math and statistics. Our presenters come from all domain fields and include occasional external speakers from regional partners, governmental agencies and industry.
The 2023-2024 seminars will be held in person, and are free and open to the public.