Pink Code

1 out of 8 Women are diagnosed with breast cancer in their lifetime, and 49% of them report delays in wait time, this project aims to reduce long waitings during diagnoses of high risk cases, using AI

Comment 1

Inspiration

  • The main inspiration of this project is to provide a service to the community and healthcare providers that reduces the amount of time to get diagnosed with breast cancer and gives a second opinion about the possible diagnosis.

What it does

  • The project consists in coding a convolutional neural network(CNN) written in python that reads a dataset of mammography scans from breast cancer patients.
    • Then the CNN, after being trained, is able to quickly tell what density of masses you have and even tell if the type of breast cancer is benign or Malignant. As of right now, the accuracy of the CNN is approaching to 93%.

How we built it

  • We built it utilizing open source code gathered from many websites regarding CNNs.

Challenges we ran into

  • One of the biggest challenges was, aside from the short time, uncertainty when programming, not knowing if we were coding the right thunk. When we began we did not know how many steps breast cancer detection has before a valid diagnosis as well as the process here in mexico, so we proceed to search on the internet but found nothing concreate, so we proceeded to ask hospital staff for pointers in the process.
  • Also, we could not manage to run the program with so much data, but once amazon gave us access to their servers we managed to run our program achieving a certainty of 93% and if we make our data set bigger, this certainty will be higher.

Accomplishments that we're proud of

  • We managed to get 93% accuaracy with a neutal network capable of classifying a mommagraphy scan as benign or malignant, with a possibility of a higher percent.

What we learned

  • We mainly learned how to get along with a team we are not familiar with in aspects as organization and compromise.

What's next for Pink Code

  • To enlarge the data pool to ensure higher accuracy rating overall.

  • Develop a web page that works with the program so the users only have to upload their mammograms to get the results.

  • Take this page to all health sectors so that with its use breast cancer is detected quicker.

Built With

+ 51 more
Share this project:

Updates