Skin Cancer Diagnosis System

Description

This project is an AI system that can recognize rare skin cancer through images, allows you to authenticate as a doctor, and manage diagnoses (CRUD) uploading images to verify if it is a disease or not. The system consumes a previously trained model using AWS with a Jupyter Notebook, the backend is built with Flask and uses the tf.Keras.Model module to consume this model. The front-end is built with VanillaJS and HTML.

CNN

This project uses 260 images of skin lesions to train a neural network. This code implements the classification of the following diseases:

1. Benign tumours
2. Basal cell carcinoma
3. Dermatofibrosarcoma protuberans
4. Melanoma

Technologies used:

Training:

• Python
• Jupyter Notebooks
• Tensorflow
• Keras
• pandas
• numpy
• boto3

Model Hosting

• AWS S3
• AWS Sagemaker

Web Development

• Flask
• Javascript
• HTML
• CSS