Creatic Cancer Chip

Our medical device incorporates a microfluidic lab on a chip that takes a urine sample and runs a lab analysis. We use ML models to determine a diagnosis order guide clinicians to treatment options.

Comment

Inspiration

Pancreatic cancer, specifically Pancreatic ductal adenocarcinoma (PDAC) is an extremely harmful disease that has a survival rate of ~10% after 5 years of being diagnosed [1]. It is important that clinicians are able to diagnose the disease as early as possible for effective treatment. There are no useful biomarkers for detection in its the cancer's early stages; however, a recent study has found a promising panel of biomarkers for earlier detection using urine analysis [1].

Link to data set

What it does

Enter Creatic Cancer Chip (CCC), our medical device incorporates an end-to-end microfluidic lab on a chip that takes a urine sample and runs a full laboratory analysis. We use standard ELISA assays with fluorophore conjugated antigens in a spinning chamber for the reaction. We then use machine learning algorithms to determine a predictive diagnosis in order to guide clinicians to early measure treatment options. All of the inferences/patient data are then sent to our webapp and stored on our CockroachDB database for secure and safe storage.

How we built it

CAD Model/ Microfluidic Chip

  • Solidworks to design the actual mechanical chassis of the device
  • We sketched the microfluidic chip design using knowledge from courses (purely a design)

ML Model

  • Tensorflow for the model development of the 2 layer neural network
  • SciKit-Learn for development of the classification network
  • Matplotlib for the data visualization

Web interface

  • Node.js for the web application development
  • Bootstrap for nice looking and interactive front-end design
  • Cockroach DB for scalable and hospital compatible distributed SQL data storage

Challenges we ran into

We are a group of biomedical engineering students, so we are used to academic work; getting used to the web development process and learning how to build a web was a very challenging process. Moreover, finding a way to get clinical insight from a limited sample of data was very difficult.

What we learned

We learned how to build a web application using Node.js, Bootstrap, and CockroachDB. Moreover, we got a chance to learn how to apply a lot of what we learnt in school to build something worthwhile. This was an excellent opportunity to learn web development

References

  1. S. Debernardi, H. O’Brien, A. S. Algahmdi, N. Malats, G. D. Stewart, M. Plješa-Ercegovac, E. Costello, W. Greenhalf, A. Saad, R. Roberts, A. Ney, S. P. Pereira, H. M. Kocher, S. Duffy, O. Blyuss, and T. Crnogorac-Jurcevic, “A combination of urinary biomarker panel and PancRISK score for earlier detection of pancreatic cancer: A case–control study,” PLOS Medicine, vol. 17, no. 12, 2020.

Built With

Share this project:

Updates