Inspiration

We are democratizing healthcare with cost-free self-diagnosis.

Melanoma is the deadliest form of skin cancer. Though this disease is highly treatable with an early diagnosis, the cost of getting an appointment with a dermatologist acts as an obstacle, deterring many people from getting early treatment. We wanted to build a solution that would allow people with suspicious moles on their body to know instantly whether that doctor's visit may be worth it or not, and get the medical attention they need earlier.

What it does

Our app uses computer vision to analyze photos of skin and machine learning to compare results with a pre-exisiting database of benign and malignant melanoma skin lesions.

Behind The Scenes

On the frontend, we used Ionic and Cordova to build a native iOS application. On the backend, we had an Express webserver running on an AWS EC2 instance to handle images sent from the user, and a Python script using the OpenCV library to classify images of moles as "harmful" or "not harmful." The computer vision algorithm preprocesses the image to remove noise, segments out the contour of the mole, and extracts features based off the standard ABCDE rule used by dermatologists to diagnose melanoma.

Sources:

  1. American Cancer Society
  2. Skin Cancer Foundation
  3. Healthcare BlueBook
  4. A developed system for melanoma diagnosis http://cennser.org/IJCVSP/finalPaper/030102.pdf

Disclaimer: The information provided in this site, or through linkages to other sites, is not a substitute for medical or professional care, and you should not use the information in place of a visit, call consultation or the advice of your physician or other healthcare provider. SkinSense is not liable or responsible for any advice, course of treatment, diagnosis or any other information, services or product you obtain through this site.

+ 6 more
Share this project:
×

Updates