Welcome to the AI Microfluidics Integration Laboratory

Department of Biomedical Engineering, Chang Gung University, TAIWAN

Deep learning for PhC data automation

While many groups are embracing the powerfulness of convolutional neural network and other machine learning techniques for microscopy image processing especially in fluorescence domain, we are focusing on application using brightfield images which can be a more difficult dataset.

While, there are many free tools also available, like deepimageJ, ilastik, zerocostDL4mic, our ultimate goal and Ryudyn's ultimate goal is knowing what you want and automate the entire data processing for you from images to statistically inferred data that you can trust.

We has explored binary, semantic, and instance segmentation using U-net, Deeplabv3, fasterCNN, yolov4, and Mask RCNN.

Try it out on our cloud test suite.

(subject to intermittent disallocation, if interested please contact)

T98G PhCCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel image

Integrated microfluidic platform for organ-on-chip applications

We use combinatorial fabrication techniques (CO2 laser, soft lithography, microengineering) to fabricate heterogeneous microfluidic platform for organ-on-chip studies (plastic, PDMS, glass).

We design, simulate, and fabricate new chips to study various topics surrounding biochemistry, microorganisms, and cells.

We also develop the electromechanical systems for environmental and experimental control, such as incubator, penumatic control as well as our own microscope.

Carousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel image

Entrepreneur thinking for scientific exploration and lean startup

"A startup is an organization formed to search for a repeatable and scalable business model" - Steve Blank

and Lean startup methods extends three principles: hypothesis-driven development, get-out-of-the-building customer development, agile development to deliver minimum viable product with limited resources (usually time) but maximum value for your customer.

In essence, academic scientific discovery is also hypothesis driven, and aimed to deliver value in terms of scientific output in a resource-limited environment.

The only difference is a startup looks for new business model in return on generating profit and in general scientific exploration just look for research results and publication and not return in money.

But several key things are share between the two:

  1. Both are looking for new unmet need (customer) and (new scientific discovery)

  2. In both spaces, although searching in hypothesis-driven, often times the most fruitful outcome came out of serendipity

  3. Both are looking to generate maximum output with limited resources

I have an honor to learn some knowledge in this space aside from traditional scientific training and I look forward to continue learning more. If prospective students who are also entrepreneurial-spirited, you're welcome to brainstorm together, finding real world applications with laboratory technologies or identifying unmet needs and lets work it in the laboratory from scratch.

Welcome prospective data scientists, AI engineers, electromechanical integration engineers, biomedical scientists to join us.


(2022.09.01) Welcome Li-Minn & Mao-Chan joining our lab!

(2022.04.08) Per CGU guidelines for COVID prevention starting 4/7, any personal level meeting is inadvisable. Office hour is now restricted to graduate student interview when needed. All meetings are switched to online.

(2022.04.08) Our long term bacterial imaging work stemming from OIST time is published on Micromachines.

(2022.01.25) For perspective students (對加入有興趣的同學),請善用Google translate,並請參考此頁

(2021.12.10) This laboratory will officially kick off on 2022.2.1 at Department of Biomedical Engineering, Chang Gung University, Taiwan

(2021.09.01) Our spinoff company Ryudyn is registered. To deliver time saving image analysis solution to our fellow researchers.