Senior Design Projects

ECS193 A/B Winter & Spring 2020

AI-based color-mapping and image mode-conversion

Email levenson@ucdavis.edu
Name Richard Levenson, MD
Affiliation Dept. of Pathology and Laboratory Medicine, UC Davis Health

Project's details

Project title AI-based color-mapping and image mode-conversion
Background Our laboratory is developing a number of novel microscopy applications. One example is MUSE (Microscopy with UV Surface Excitation)--a pdf of a recent publication is attached. We have other novel imaging methods that look extremely promising but are still so new we are not even talking about them yet. All of these methods generate images directly that have to be further converted either for human viewing or for downstream quantitative analysis. There are a lot of interesting and state-of-the-art tools in AI and related fields that can be applied. We already have some very promising results using Generative Adversarial Networks (GANs), but other methods need to be explored as well.
Description The task would involve learning the underlying image challenges and goals, with some understanding of histology and microanatomy required (can be learned on the job). Then, testing a number of AI- and related approaches for image color-mapping, super-resolution, and suppression of sample-preparation artifacts. The exact task will depend on what is most intriguing and useful when this project commences. Acceleration of the converter algorithms using GPUs or other, perhaps newer hardware platforms for near-real-time performance after the learning phase is completed is a goal.
Deliverable Image-processing results, and description of the research tasks and evolution of the solution(s). Functional converters that can process images in near-real-time would constitute major success, but do not have to be achieved.

Intellectual Property Rights: As a UC Davis student working on this project as part of your coursework, you hold the intellectual property rights for the code you write and other materials you create. Because this project may be of interest to a start-up is intended to be spun out from UC Davis, you may be asked to transfer ownership of any intellectual property you create for this project to Newco with no guarantee of compensation by Newco. Please do not sign up for this project if these terms are not acceptable to you. Publications and presentations will be encouraged as long as necessary IP protections are in place in advance.
Skill set desirable Some facility with biology, and exposure to AI-based tools and approaches. Experience with digital image processing would be a plus.
Phone number 617-803-8860 (cell); 916-734-2335 (office)
Client time availability 30-60 min weekly or more
IP requirement Client wishes to keep IP of the project
Attachment Click here