Program
Tentative School Curriculum
The summer school program consists of 6-day lecture/discussion and 1-day social activities. Students will be grouped to have in-class discussions with the speakers and hands on experience. Informal interactions between faculty and students will be strongly encouraged during daily coffee breaks, lunch/dinner meeting, project discussion, and social activities. We also plan to organize a summer school gala to bring students and speakers together showing off their respective talents. As follows is a list of topics to be covered.
- Shape analysis, Segmentation and Visualization
- Human (animal, cell) crowd tracking, Behavior modeling
- Machine learning for computer aided diagnosis, Large, multi-modal, multi-scale biomedical image database indexing and retrieval, Bioinformatics
- Pattern discovery, tracking and quantification
- Neuroimage: Multimodality high-dimensional deformable registration (cross modality, cross subjects). Computational and Statistical Anatomy (Digital atlas)
Schedule
Monday (7/23)
8:30-8:45 am: Opening Talks
8:45-9:30 am: Dr. Hanchuan Peng
High-throughput analysis of microscopic images using 3D digital atlases of model animals
1.1. Applications in cell biology, gene expression analysis, neuroscience, etc.
1.2. Methods: 3D segmentation, tracing, registration, visualization, data mining, modeling
1.3. Major challenges and future directions
9:30-10:00 am: Tea/Coffee Break
10:00 - 11:00 am :Dr. Hanchuan Peng
High-throughput analysis of microscopic images using 3D digital atlases of model animals (Cont’d)
1.4. Case studies:
1.4.1 High resolution C. elegans digital map and single cell analyses
1.4.2 Zebrafish image segmentation and informatics
1.4.3 Fruit fly bioimage informatics (segmentation/tracing, registration, annotation, and modeling)
1.4.4 Others & industrial applications (e.g. electron microscopy image alignment, segmentation and 3D reconstruction, cell segmentation and tracking, etc.)
11:00 - 12:00 noon: Prof. Yu, Hongbo
Multiple scale functional and structural imaging in vivo
12:00 - 2:00 pm: Lunch
2:00 - 3:00 pm: Prof. Anant Madabhushi
Digital Pathology: Role of Image Analysis (Part I)
- What is Digital Pathology?
- Why is digital pathology so different from radiologic images?
- What are the unique questions one can pose with digital pathology
- Need for unique image computing tools in digital pathology
3:00 - 3:30 pm : Tea/Coffee Break
3:30 - 4:30 pm : Prof. Anant Madabhushi
Computer aided prognosis (Part II)
- What is computer aided prognosis
- Feature extraction and characterization fro digital pathology
- Classification of high dimensional feature spaces
- Use case of Computer aided prognosis in prostate cancer
- Use case in breast cancer
4:30 - 5:30 pm: Review/Discussion
Evening event Welcome Dinner
Tuesday (7/24)
8:30-9:30 am: Prof. Anant Madabhushi
Histologic image analysis - unique challenges in digital pathology Quantitative data convergence (Part III)
- Introduction to QDC
- Convergence of radiology and histology
- Convergence of radiologic and molecular data
- Convergence of histogic and molecular data
- Cross modality correlations -- What can we learn and how?
9:30-10:00 am: Tea/Coffee Break
10:00 - 11:00 am: Dr. Hanchuan Peng
Vaa3D: Visualization-assisted analysis in 3D
http://vaa3d.org
https://code.google.com/p/vaa3d/#What_Vaa3D_means
11:00 - 12:00 noon: Dr. Tiehua Ni
Model-Based Integration and Visualization of Heterogeneous Data
12:00 - 2:00 pm: Lunch
2:00 - 3:00 pm: Prof. B. S. Manjunath
Bioimage Informatics, Part 1: Image analysis challenges
A fundamental problem in bioimage analysis--and a major bottleneck in the workflow--is that of spatial-temporal segmentation of the data. In the context of microscopy images, these data include the traditional 2D images, as well as the 3D stack images (consisting of optical section, 4D (time lapse plus 3D) and 5D (4D plus spectral).
a) Overview of the problem and some biological examples drawn from current research.
b) Segmentation and tracing in electron micrographs towards building a retinal connectome .
c) intro to probabilistic graphical models in image analysis, Markov Random Fields and interactive segmentation
3:00 - 3:30 pm: Tea/Coffee Break
3:30 - 4:30 pm: Prof. B. S. Manjunath
Bioimage Informatics, Part 1: Image analysis challenges (Cont’d)
d) 3D tracing of multiple structures in EM data and issues of scalability
e) Super-pixel based multiple segmentations and fusion;
f) Tracking examples (microtubule and melansome tracking), challenges; Simultaneous detection and tracking framework, scalability and robustness issues in tracking.
4:30 - 5:30 pm: Review/Discussion
Wednesday (7/25)
8:30-9:30 am: Prof. B.S. Manjuanth
Bioimage Informatics, Part 2: Bisque platform for high-throughput, web-based, bioimage analysis
Bisque is a web-based image database system for storing, managing, analyzing and sharing bioimages. This is an open source software infrastructure that integrates image analysis with databases, and supports most existing microscopy image formats, and including 2D, 3D, 4D and 5D images. In addition to basic functionalities such as viewing, enhancement, and basic analysis modules, users can build and integrate their own modules into Bisque. More information can be found at http://www.bioimage.ucsb.edu.
a) Introduction to BISQUE
b) loading, annotating, sharing, processing and publishing images in Bisque
9:30-10:00 am: Tea/Coffee Break
10:00 - 11:00 am: Prof. B.S. Manjuanth
Bioimage Informatics, Part 2: Bisque platform for high-throughput, web-based, bioimage analysis (Cont’d)
c) graphical annotations and tagging
d) Image analysis workflow and case studies on nuclei detection, root tip tracking.
11:00 - 12:00 noon Prof. Hong Ma
Chromosome behaviors during Arabidopsis meiosis reveal meiotic recombination mechanisms
12:00 - 2:00 pm Lunch
2:00 - 3:00 pm Prof. Serge Belongie
Visual Recognition With Humans in the Loop
- Overview of Visipedia
- Subordinate vs. Basic Level Categorization
- Human Computation Overview
- The Parts and Attributes Framework
- Decision Trees for Interactive Classification
- Dealing with Noise in User Responses
3:00 - 3:30 pm Tea/Coffee Break
3:30 - 4:30 pm Prof. Serge Belongie
Crowdsourcing and Its Applications in Computer Vision
- Introduction to Mechanical Turk
- Task Incentives
- Experimental Design
- Quality Management
- Cost Effective Strategies for Obtaining Labels
- Example Applications
4:30 - 5:30 pm Review/Discussion
Thursday (7/26): Full-day Outing: Suzhou Day Trip
Friday (7/27)
8:30-9:30 am Prof. Erik Learned-Miller
Image Alignment, Image Comparison, and Digital Atlases, Part I
- Introduction to image alignment [Ref. 1]
- Problem definition
- Choosing a representation
- Choosing an alignment criterion
- Optimizing the alignment criterion
- Some simple representations and alignment criteria
- Optimization methods
- Exhaustive search
- Keypoint methods
- Gradient descent
- Multimodal alignment [Ref. 2]
- Non-parametric density estimation
- alignment by maximization of mutual information
9:30-10:00 am Tea/Coffee Break
10:00 - 11:00 am Prof. Erik Learned-Miller
Image Alignment, Image Comparison, and Digital Atlases, Part II
- Multi-image alignment [Ref. 3]
- Problem definition
- A joint alignment criterion
- The congealing algorithm
- Congealing 3D brain volumes [Ref. 4]
- Digital brain atlases
11:00 - 12:00 noon Prof. Wu
Phenotypic Analysis of Developmental and Metabolic Defects in Mice
12:00 - 2:00 pm Lunch
2:00 - 3:00 pm Prof. Erik Learned-Miller
Image Alignment, Image Comparison, and Digital Atlases, Part III
- Using the warp as an information source
- Learning from one example [Ref. 5]
- Learning about anatomy from joint alignment
- Broadening the notion of alignment
- Removing multiplicative offsets in MRI
- Removing slowly varying bias fields in MRI [Ref. 6]
- Congealing complex images [Ref. 7]
- Deep belief nets for congealing
3:00 - 3:30 pm Tea/Coffee Break
3:30 - 4:30 pm Prof. Erik Learned-Miller
Image Alignment, Image Comparison, and Digital Atlases, Part IV
- Distribution fields for alignment
- The basin of attraction issue for gradient algorithms
- Traditional methods for increasing basin of attraction
- Distribution fields [Ref. 8]
- Applications of distribution fields
- Tracking [Ref. 9]
- Backgrounding
- Summary
References, Prof Learned-Miller:
1. http://research.microsoft.com/pubs/75695/Szeliski-FnT06.pdf
2. http://people.csail.mit.edu/sw/papers/IJCV-97.pdf
3. http://people.cs.umass.edu/~elm/papers/PAMI_congeal.pdf
4. http://people.cs.umass.edu/~elm/papers/congeal_3D.pdf
5. http://people.cs.umass.edu/~elm/papers/cvpr2000.pdf
6. http://people.cs.umass.edu/~elm/papers/nips_bias.pdf
7. http://people.cs.umass.edu/~elm/papers/iccv07alignment.pdf
8. http://people.cs.umass.edu/~elm/papers/Tech-Report-DF.pdf
9. http://people.cs.umass.edu/~elm/papers/trackingDF.pdf
4:30 - 5:30 pm Discussion
Evening event Talent Show
Saturday (7/28)
8:30-9:30 am Prof. Charless Fowlkes
Automating Biological Image and Shape Analysis
9:30-10:00 am Tea/Coffee Break
10:00 - 11:00 am Prof. Charless Fowlkes
Automating Biological Image and Shape Analysis (Cont’d)
11:00 - 12:00 noon Prof. Tao P. Zhong
Pattern formation and morphogenesis in vertebrate embryo and organ development
12:00 - 2:00 pm Lunch
2:00 - 3:00 pm Prof. Yanxi Liu
Capturing Near-regular Patterns in Digitized Life Sciences
Part I: Motivation and Theory
- A brief introduction to Pattern Theory and practice
- Symmetry group-based regularity space - a novel, computable model
- Demonstrations of Typical patterns in digitized life science data sets: universal, low-rank, deformed regular patterns
- Computational challenges: why is it hard for computer vision algorithms to discover real world, free-form symmetries?
3:00 - 3:30 pm Tea/Coffee Break
3:30 - 4:30 pm Prof. Yanxi Liu
Capturing Near-regular Patterns in Digitized Life Sciences (cont.)
Part II Tools and Applications
Tools:
- discriminative feature subset selection (off-line, on-line)
- curved glide reflection symmetry detection
- skewed rotation symmetry detection
- texture regularity discovery (translation symmetry)
Applications:
- 2d/3D human identification, expression/gender classification
- Quantified patterns (firing fields of grid cells)
- Tracking patterns (gated cardiac MRI)
- Evaluation of Scoliosis
- Computer aided diagnosis for neurodegenerative diseases(Alzheimer’s Disease, Schizophrenia)
- Brain tumor detection and segmentation
- Zebra fish (wild versus mutant)
- Human gaits/dance
4:30 - 5:30 pm Discussion
Sunday (7/29)
8:30-9:30 am Prof. Robert Collins
Video Tracking and Crowd Scene Analysis
Part 1: Tracking Foundations
Appearance-based vs tracking-by-detection
Single target tracking
Recursive filtering
Dynamic programming
Multiple target tracking
Filtering + data association
Multi-frame formulations
9:30-10:00 am Tea/Coffee Break
10:00 - 11:00 am Prof. Robert Collins
Video Tracking and Crowd Scene Analysis (Cont’d)
Part 2: Analyzing Crowd Behavior
Detection and counting
Crowd flow analysis
Social force models
Detecting small groups
11:00 - 12:00 noon Prof. Chen
Multi-view 3D tracking of particle systems and deforming surfaces, and its application to biomedical research.
12:00 - 2:00 pm Lunch
2:00 - 3:00 pm Prof. Zhuowen Tu
Discriminative Models for Medical Imaging
One direction in medical image analysis is to effectively represent knowledge and efficiently extract biomedical information (such as a deformable shape) from medical images. In particular, machine learning techniques (supervised, weakly-supervised, and unsupervised) have played increasingly important role. The large-scale data learning and analysis have also recently played a significant role in medical imaging. The goal of this lecture is to provide a comprehensive assessment of discriminative learning techniques used for medical imaging applications such as anatomical structure detection and segmentation, image categorization, etc. Learning from an annotated dataset the covers the uncertainties involved in the applications, these techniques are able to derive compact descriptions between the image and knowledge and gain improvements in performance and speed when compared with conventional algorithms without using learning. In the lecture, we will cover: (1) some basics about supervised and semi-supervised learning.
3:00 - 3:30 pm Tea/Coffee Break
3:30 - 4:30 pm Prof. Zhuowen Tu
Discriminative Models for Medical Imaging (Cont’d)
(2) applications of supervised and semi-supervised learning in recent medical imaging applications.
4:30 - 5:30 pm Closing Ceremony