CS 3710 / ISSP 3565
Monday, Wednesday
4:30PM - 5:45PM
Office hour: Thursday 1:30PM - 2:30PM, or by appointment
sjh95 (at) pitt.edu
www.pitt.edu/~sjh95
5427 Sennott Square
Summary: This course will cover introductory to advanced topics in medical imaging analysis using methods in statistics, machine learning, and computer vision. Throughout the course, students will implement various methods on real medical imaging data (e.g., MRI, fMRI), see how various problems can be addressed (e.g., finding brain regions associated with disease risk factors), and understand what it means to solve them. By the end of the course, students will be able to identify problems medical imaging analysis, apply appropriate methods from various fields, and derive scientific findings from their analyses. Recommended: Basic knowledge in Linear algebra, Algorithms, and Artificial Intelligence; Basic proficiency in Matlab and Python or Java. Pre-requisites: Artificial intelligence (CS-2710), or by consent of the instructor.
- 1/6/20 (M): Course overview
- Course logistics
- What to expect
- Research problems
- 1/8/20 (W): Basics in medical imaging analysis 1
- Imaging modalities: MRI, fMRI, PET, etc.
- Basics of image
- 1/13/20 (M): Basics in medical imaging analysis 2
- Filtering
- Smoothing
- Reading 1: Effects of spatial smoothing on fMRI group inferences (Mikl et al., 2008)
- Reading 2: An optimized blockwise nonlocal means denoising filter for 3-D magnetic resonance images. (Coupe et al., 2008)
- 1/15/20 (W): No Lecture
- 1/20/20 (M): No Lecture (MLK)
- 1/22/20 (W): Basics in medical imaging analysis 2
- Segmentation
- Reading: White Matter Hyperintensity segmentation (Muschelli et al., 2017)
- 1/27/20 (M): Statistical analysis 1
- Regression, Correlation
- 1/29/20 (W): Statistical analysis 2
- p-value (univariate/multivariate response)
- Finding associations in neuroimaging
- Reading: Amyloid burden and neural function in people at risk for Alzheimer's Disease (Johnson et al., 2013)
- Multiple testing corrections
- Reading: Independent filtering increases detection power for high-throughput experiments (Bourgon et al., 2010)
- 2/3/20 (M): Spectral methods 1
- Harmonic analysis
- Wavelet transform on graph
- Reading: Multi-resolution statistical analysis of brain connectivity graphs in preclinical Alzheimer's disease (Kim et al., 2015)
- Heat kernel on graph
- Reading: Cortical thickness analysis in autism with heat kernel smoothing (Chung et al., 2005)
- 2/5/20 (W): Spectral methods 2
- Harmonic analysis for collaborative filtering
- Reading: Adaptive Signal Recovery on Graphs via Harmonic Analysis for Experimental Design in Neuroimaging (Kim et al., 2016)
- Network diffusion of disease pathology
- Reading: A Network Diffusion Model of DiseaseProgression in Dementia (Raj et al., 2012)
- 2/10/20 (M): Computer vision in medical imaging 1
- Intro to computer vision
- Normalized-cut for image segmentation
- Reading: Normalized Cut (Shi et al., 2000)
- 2/12/20 (W): Computer vision in medical imaging 2
- Snakes: Active Contour Model:
Snakes: Active Contour Model (Kass et al., 1988) - Optical flow
- Reading: Image matching as a diffusion process: an analogy with Maxwell’s demons (Thirion et al., 1998)
- 2/17/20 (M): Computer vision in medical imaging 3
- 2/19/20 (W): Intro to machine learning and deep learning
- Basics
- 2/24/20 (M): Intro to deep learning in medical imaging (a.k.a. potential project topics)
- Project description
- Reading 1: A survey on deep learning in medical image analysis (Litjens et al., 2017)
- Reading 2: An overview of deep learning in medical imaging focusing on MRI (Lundervold et al., 2018)
- 2/26/20 (W): Deep learning in medical imaging 1
- Convolutional neural networks
- Reading 1: U-Net: Convolutional Networks for Biomedical Image Segmentation (Ronneberger et al., 2015)
- Reading 2: Discriminative Unsupervised Feature Learning with Convolutional Neural Networks (Dosovitskiy et al., 2014)
- 3/2/20 (M): Deep learning in medical imaging 2
- Spatial transformer networks and its applications
- Reading 1: Spatial Transformer Networks (Jaderberg et al., 2015)
- Reading 2: Generating Patient-like Phantoms Using Fully Unsupervised Deformable Image Registration with Convolutional Neural Networks (Chen et al., 2019)
- Reading 3: TETRIS: Template Transformer Networks for Image Segmentation With Shape Priors (Lee et al., 2019)
- 3/4/20 (W): Deep learning in medical imaging 3
- Graph convolutional neural networks
- Reading 1: Semi-Supervised Classification with Graph Convolutional Networks (Kipf et al., 2017)
- Reading 2: Disease Prediction using Graph Convolutional Networks: Application to Autism Spectrum Disorder and Alzheimer’s Disease (Parisot et al., 2018)
- Reading 3: BrainNetCNN: Convolutional Neural Networks for Brain Networks; Towards Predicting Neurodevelopment (Kawaharaa et al., 2017)
- Reading 4: Distance Metric Learning using Graph Convolutional Networks: Application to Functional Brain Networks (Ktena et al., 2017)
- Reading 5: Structured Sequence Modeling with Graph Convolutional Recurrent Networks (Seo et al., 2016)
- 3/9/20 (M): No Lecture (Spring Break)
- 3/11/20 (W): No Lecture (Spring Break)
- 3/16/20 (M): Short project presentation
- Based on the initial project proposal
- 5 min presentation + 3 min discussion
- 3/18/20 (W): Deep learning in medical imaging 4
- Recurrent neural networks
- Reading 1: Identifying Autism from Resting-State fMRI Using Long Short-Term Memory Networks (Dvornek et al., 2017)
- Reading 2: Brain decoding from functional MRI using long short-term memory recurrent neural networks (Li et al.)
- Topics: Tensor-train decomposition for a large feature space
- Reading 1: Tensor-train Decomposition (Oseledets et al., 2011)
- Reading 2: Tensor-Train Recurrent Neural Networks for Video Classification (Yang et al., 2017)
- Reading 3: Scaling Recurrent Models via Orthogonal Approximations in Tensor Trains (Mehta et al., 2019)
- 3/23/20 (M): Deep learning in medical imaging 5
- Recurrent neural networks 2
- Other types of sequential neural networks
- Reading 1: Attention Is All You Need (Vaswani et al., 2017)
- Reading 2: Autoregressive Convolutional Neural Networks for Asynchronous Time Series (Yang et al., 2017)
- 3/25/20 (W): Deep learning in medical imaging 6
- Generative models
- Reading 1: Generative Adversarial Network in Medical Imaging: A Review (Yi et al., 2019)
- 3/30/20 (M): Topics
- Flow-based models
- Reading 1: Density Estimation using Real NVP (Dinh et al., 2016)
- Reading 2: DUAL-GLOW: Conditional Flow-Based Generative Model for Modality Transfer (Sun et al., 2019)
- Reading 3: Conditional Recurrent Flow: Conditional Generation of Longitudinal Samples with Applications to Neuroimaging (Hwang et al., 2019)
- 4/1/20 (W): Topics
- Transfer learning and domain adaptation
- Reading 1: Transfusion: Understanding Transfer Learning for Medical Imaging (Raghu et al., 2019)
- Reading 2: Domain adaptation for Alzheimer's disease diagnostics (Wachinger et al., 2016)
- Reading 3: Transfer Learning for Domain Adaptation in MRI: Application in Brain Lesion Segmentation (Ghafoorian et al., 2017)
- 4/6/20 (M): One-on-one project meeting
- We will schedule a 5~10 min one-on-one meeting for each project during class.
- 4/8/20 (W): Topics
- Uncertainty in DL
- Reading 1: Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning (Gal et al., 2016)
- Reading 2: Uncertainty-driven Sanity Check: Application to Postoperative Brain Tumor Cavity Segmentation (Jungo et al., 2018)
- Reading 3: A Probabilistic U-Net for Segmentation of Ambiguous Images (Kohl et al., 2019)
- 4/13/20 (M): Final project presentation 1
- ~20 min (15 min presentation + 5 min discussion)
- The project itself does not need to be complete by this point.
- 4/15/20 (W): Final project presentation 2
- 4/20/20 (M): Final project office hour 1 (Optional attendence)
- I will be in the class if you want to get feedback or comment on the project.
- 4/22/20 (W): Final project office hour 2 (Optional attendence)
If the project meets the submission quality, these are some of the potential venues to submit:
- Late April of 2020: BMVC (this is for 2019)
- March/April of 2020 (estimated): CVPR 2020 Workshop
- May 18, 2020 (estimated): NeurIPS
- June 30, 2020: MICCAI 2020 Workshop
- August 31, 2020 (estimated): AAAI
- September of 2020 (estimated): ICRA (this is for 2019)
- Datasets: These are public datasets. Some may need to be requested which is a fairly standard and smooth procedure.
- Tools
All assignment submissions must be the sole work of each individual student. Students may not read or copy another student's solutions or share their own solutions with other students. Students may not review solutions from students who have taken the course in previous years. Submissions that are substantively similar will be considered cheating by all students involved, and as such, students must be mindful not to post their code publicly. The use of books and online resources is allowed, but must be credited in submissions, and material may not be copied verbatim. Any use of electronics or other resources during an examination will be considered cheating.
If you have any doubts about whether a particular action may be construed as cheating, ask the instructor for clarification before you do it. The instructor will make the final determination of what is considered cheating.
Cheating in this course will result in a grade of F for the course and may be subject to further disciplinary action.
Using an open-source codebase is accepted, but you must explicitly cite the source, especially following the owner's guideline if it exists. For any writing involved in the project, plagiarism is strictly prohibited. If you are unclear whether your work will be considered as plagiarism, ask the instructor before submitting or presenting the work.