Melatect: A Machine Learning Model Approach For Identifying Malignant Melanoma in Skin Growths

Melatect: A Machine Learning Model Approach For Identifying Malignant Melanoma in Skin Growths

Publication date: Sep 07, 2021

Malignant melanoma is a common skin cancer that is mostly curable before metastasis, where melanoma growths spawn in organs away from the original site. Melanoma is the most dangerous type of skin cancer if left untreated due to the high chance of metastasis. This paper presents Melatect, a machine learning model that identifies potential malignant melanoma. A recursive computer image analysis algorithm was used to create a machine learning model which is capable of detecting likely melanoma. The comparison is performed using 20,000 raw images of benign and malignant lesions from the International Skin Imaging Collaboration (ISIC) archive that were augmented to 60,000 images. Tests of the algorithm using subsets of the ISIC images suggest it accurately classifies lesions as malignant or benign over 95% of the time with no apparent bias or overfitting. The Melatect iOS app was later created (unpublished), in which the machine learning model was embedded. With the app, users have the ability to take pictures of skin lesions (moles) using the app, which are then processed through the machine learning model, and users are notified whether their lesion could be abnormal or not. Melatect provides a convenient way to get free advice on lesions and track these lesions over time.

PDF

Concepts Keywords
Cancerous Artificial intelligence
Forest Neural network
Hematology Elegant streamlined algorithm
June Object detection
Maxpool2d Convolutional neural network
Skin cancer
Dermatoscopy
Melanocytic nevus
Cancer
Melanoma
Benign neoplasms
Artificial neural networks
Clinical medicine
Cutaneous conditions
Medicine
Organs primary site
Bias anatomical site
Healthcare
Detection algorithm
Elegant streamlined algorithm
Sequential algorithm
Neural networks
Classification systems
Shapes sizes network
Artificial intelligence systems
Neural network
App iOS devices
5 sequential algorithm
API
Smartphone
Machine learning
Android
Html
Image Processing
Detection algorithm
Pdf
CAD

Semantics

Type Source Name
disease MESH Malignant Melanoma
disease MESH skin cancer
disease MESH metastasis
pathway KEGG Melanoma
disease MESH diagnosis
disease MESH biopsy
disease MESH malignancy
drug DRUGBANK Methionine
drug DRUGBANK Water
drug DRUGBANK Alpha-1-proteinase inhibitor
drug DRUGBANK Methyltestosterone
disease MESH development
drug DRUGBANK Aspartame
disease MESH uncertainty
disease MESH skin mole
disease MESH carcinoma
disease MESH squamous carcinoma
drug DRUGBANK Coenzyme M
disease MESH Freckles
disease MESH Xeroderma pigmentosum

Download Document

Leave a Comment

Your email address will not be published. Required fields are marked *