SNIC
SUPR
SNIC SUPR
Developing deep learning algorithms for uveal naevi classification
Dnr:

SNIC 2018/5-96

Type:

SNAC Small

Principal Investigator:

Charlotta All-Eriksson

Affiliation:

Karolinska Institutet

Start Date:

2018-07-23

End Date:

2019-02-01

Primary Classification:

10207: Computer Vision and Robotics (Autonomous Systems)

Webpage:

Allocation

Abstract

Birthmarks in the eye (uveal naevi), is a small benign tumor that in rare cases develop into malignant melanoma. Uveal melanoma has a tendency to metastasize to the liver and when the liver metastases are clinically manifested the survival rate is usually limited to a few months. Approximately 50% of patients succumb to metastasis within 10 years of diagnosis. The risk of developing uveal melanoma from a naevus (birthmark) is determined by an examination at a doctor specialized in eye diseases. The assessment is done using a biomicroscope and takes approximately 10-15 minutes. The aim of this project is to create a convolutional neural network (CNN) that can replace the examination by training the network on retinal images of benign nevi and malignant melanomas. The network is trained on images of patients from S:t Eriks eye hospital in Stockholm.