DevMesh

Skin Cancer Classification Platform

Szymon Kocot

Szymon Kocot

Bytom, Silesian Voivodeship

6 0
  • 0 Collaborators

The goal of this project is to create a prototype skin cancer classification web platform. ...learn more

Project status: Concept

Artificial Intelligence

Intel Technologies
Intel Opt ML/DL Framework, Intel Python

Code Samples [1]

Overview / Usage

Skin cancer is one of most common type of cancer. Early and precise diagnosys can help make this disease less seviere and to treat it easier.

Functionalities:

  • Melanoma malignancy prediction using Convolutional Neural Network classifier
  • Visual representation of the masked image and prediction pie chart
  • Access and filtering ISIC database

Melanoma malignancy can be classified with over 70% accuracy.
This value may seem to be low, but you need to consider, that image and expert label quality are not that good.
Masking and resizing of melanoma images helped to improve results by a few percents.

Technologies Used

  • Flask (web server)
  • Tensorflow and Keras (Convolutional Neural Networks)
  • Pillow (image manipulations)
  • jQuery
  • PostgreSQL (database) and psycopg2 (database interface framework)
  • Intel optimized Python distribution and dependencies for faster execution

Repository

https://github.com/szkocot/SCALP

Comments (0)