Skip to content

Feature engineering for melanoma prediction using simple classifiers.

Notifications You must be signed in to change notification settings

sf-pear/melanoma-prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Melanoma Prediction from Images

Project by Sabrina Fonseca Pereira, Ida Maria Zachariassen, Magnus Sverdrup, Rasmus Bondo Hansen and Ruben Oliver Jonsman

The project presented in this report was developed with the purpose of researching and analysing the visual identification and classification of potential melanoma cancerous lesions through imaging. This project also aims to investigate how a model based on the characteristics of melanoma lesions performs when classifying keratosis lesions.

This was done to gain insight and knowledge about extracting features from medical imaging and using these features in simple classifiers. Lastly the report is an evaluation on whether visual classification of skin lesions is a reliable source for cancer detection.

Based on our analysis of various models for classifying features of skin lesions, the following research question was formulated: "How do models trained on melanoma lesions perform when classifying seborrhoeic keratosis lesions?"

Example data was included for illustration only, the training set is avaible from ISIC.

The repo

data

  • Includes data from computed features, so instead of recalculating, we load the csv files.
  • example subset of data used for this project as an example.

notebooks

  • main.ipynb full project notebook with text explanations.
  • personal-notebook.ipynb personal notebook for experimentation, feature engineering and analysis.

About

Feature engineering for melanoma prediction using simple classifiers.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published