Fall 2024 Teaching

CS 684 – Advanced Computational Methods for Biomedical Imaging

https://aaz-imran.github.io/teaching/2024-fall-cs684

Syllabus

Time: MW 9:00 AM – 10:15 AM
Location: Register course to know

Course Instructor: Dr. Abdullah-Al-Zubaer Imran
Office: 319 Marksbury
Office Hours: Mondays 11:00 AM – 1:00 PM, and by appointment.

Course Description:
The field of imaging science is undergoing rapid growth. Biomedical imaging and its analysis play a fundamental role in comprehending, visualizing, and quantifying medical images for various clinical applications. Computational image analysis techniques can significantly improve disease diagnosis by making it faster, more effective, and fairer. This seminar course will focus on understanding different imaging tasks associated with medicine and applications of advanced computational methods (such as computer vision and deep learning) to solve important biomedical imaging problems. The course will also examine some key topics and advanced techniques in computer vision and medical imaging, reading, reviewing, presenting, and discussing papers published in major computer vision and medical imaging venues (e.g., CVPR, IEEE TMI, MedIA, MICCAI, ICCV, etc.). The course will be taught with a good mix of theories and applications with different case studies.

Topics include:

Introduction to digital image processing and computer vision
Major imaging modalities (e.g., X-ray, CT, MRI, Ultrasound)
Fundamental image processing techniques
Machine learning and its applications to biomedical imaging
Deep learning and its applications to biomedical imaging
Important image databases, ML/DL, and Medical Imaging tools
Challenges in medical imaging and recent computational methods to tackle them
Reading and writing medical imaging research papers

Course Outcomes:
Upon completion of this course, students will:

Understand different medical imaging systems and their usages
Comprehend basic and advanced image processing techniques
Analyze imaging data related to problems in medicine
Use ML/DL and medical imaging tools
Identify and solve medical image analysis tasks by applying appropriate computational methods
Define and implement their own project that explores an important problem in medical imaging
Gain experience in reading technical papers, and presenting research outcomes in a professional setting

Prereqs:
Course prerequisite: Programming ability at an intermediate level, familiarity with probability and statistics, knowledge of Python (highly recommended), machine learning, or approval of the instructor.

Required books:
There is no required textbook for CS 684.

Recommended books:

Digital Image Processing 4th Ed. by Gonzalez and Woods
Insight into images: principles and practice for segmentation, registration, and image analysis by Terry S. Yoo
Deep Network Design for Medical Image Computing by Haofu Liao; Kevin Zhou; Jiebo Luo
Handbook of Medical Image Computing and Computer Assisted Intervention by Kevin Zhou, Daniel Rueckert, Gabor Fichtinger

Course Schedule/Outline (Tentative):

Week	Topic	Note
1.	Aug 26: Course introduction Aug 28: Medical imaging basics, Project ideas
2.	Sep 2: No lecture – Labor day Sep 4: Image processing fundamentals	release of hw1
3.	Sep 9: Machine learning Sep 11: Machine learning in medical imaging	paper bidding due
4.	Sep 16: Deep learning Sep 18: Deep learning for medical imaging	hw1 due
5.	Sep 23: Deep generative models Sep 25: Vision Transformers	release of hw2 project proposal due
6.	Sep 30: Self-supervised learning Oct 2: Fairness & Explainability
7.	Oct 7: Project discussion Oct 9: Project discussion	hw2 due
8.	Oct 14: [Paper 1] Devil is in the Queries: Advancing Mask Transformers for Real-world Medical Image Segmentation and Out-of-Distribution Localization (Presenter: N. Munia) Oct 16: [Paper 2] Geometric Visual Similarity Learning in 3D Medical Image Self-Supervised Pre-training(Presenter: R. Sheridan)
9.	Oct 21: [Paper 3] Diffusion Transformer U-Net for Medical Image Segmentation(M. Qasim) Oct 23: Project midterm presentation
10.	Oct 28: No Lecture – Fall Break Oct 30: [Paper 4] MA-SAM: Modality-agnostic SAM adaptation for 3D medical image segmentation(Presenter: T. Ward)
11.	Nov 4: [Paper 5] PerceptionGPT: Effectively Fusing Visual Perception into LLM(Presenter: R. Rifa) Nov 6: [Paper 6] A case for reframing automated medical image classification as segmentation(Presenter: T. Ward)
12.	Nov 11: [Debate] LLMs and FMs are universal medical solvers! Nov 13: [Paper 7] Masked Vision and Language Pre-training with Unimodal and Multimodal Contrastive Losses for Medical Visual Question Answering(Presenter: R. Rifa)	release of hw3
13.	Nov 18: [Paper 8] Unsupervised Medical Image Translation With Adversarial Diffusion Models(M. Qasim) Nov 20: [Paper 9] BerDiff: Conditional Bernoulli Diffusion Model for Medical Image Segmentation(Presenter: R. Sheridan)
14.	Nov 25: [Paper 10] I2MVFormer: Large Language Model Generated Multi-View Document Supervision for Zero-Shot Image Classification(Presenter: N. Munia) Nov 27: No Lecture – Thanksgiving	hw3 due
15.	Dec 2: Final Project Presentation Dec 4: Final Project Presentation	final report due
16.	Dec 9: Prep Days – Concluding Remarks and Final Discussion Dec 11: No lecture – Reading day

Course Activities:

Class participation (15%) – lecture quizzes, class performance, and debate
Paper discussion (15%) – paper presentations and online discussions
Assignments (30%) – two coding and a paper reviewing assignments
Project (40%) – proposal submission, midterm presentation, abstract submission, final presentation, and code and report submission.

Grading Scale:
After the final percentage grade is calculated, the following scale will be used to determine the final letter grade.

For graduate students:

90–100% (A)
80–89% (B)
70–79% (C)
0–69% (E)

READ THIS:

Attendance Policy: Students are required to attend every lecture. Students only present in the class can take the associated quiz and get the available points. Missing a maximum of 2 lectures will be excused. Any student missing more than 2 lectures without any reasonable excuses will start losing 10 points for every miss. So, you will get 0 out of 80 points available in Attendance & Quizzes if you miss 10 lectures in total.

Academic Integrity: Please strictly follow the Academic Offenses Rules (plagiarism, cheating, and falsification or misuse of academic records). Also, keep in mind that any copyrighted materials (e.g., images and other media), and published contents (e.g., academic papers, books, web sources, online tools) used in your submissions and project should be properly cited. Ideas from people other than your own (for the project—ideas from outside your group) should be acknowledged.

Late Policy: Late submissions (assignment, project proposal, code, project final report) will be subject to a 50% score penalty if you submit within 1 day after the deadline. A score of 0 will be given for any submissions after that. Late submissions will be accepted only for certain circumstances (e.g., medical) with proper evidence.
Exceptions to this rule may be requested by providing appropriate documentation (e.g., medical) which will be considered at the discretion of the instructor.

Generative AI Policy: GenAI tools such as ChatGPT may be used in this course for the purposes of pre-submission activities. However, students are responsible for submitting work that meets the assignment standards for quality and factual accuracy. Check the Canvas page for more detailed instructions on this. If you have any questions or concerns about the policy, contact the instructor before submitting any assignments.

Disability and Special Accommodation: Please let the instructor know of any needed accommodation in the first two weeks. Please also see Academic Accommodation for further assistance.

Academic Policy Statements, DEI, Resources Available to Students

Useful Resources: