Melo uses the latest machine learning technology to detect whether a sample of melanoma is 'benign' or 'malignant'!
Project Melo aims to accurately provide both doctors and patients with a credible software platform that not only enhances communication of certain health aspects, but also assists in affirming possible health cases. At its basis, Melo allows for patients and doctors to better communicate through a texting interface. The doctor also has the ability to upload important information, such as blood pressure, triglycerides, cholesterol, and chemotherapy results. Once doctors upload the microscopy of the Melanoma tumor, Melo will determine whether the sample is malignant or benign, which can help doctors modify their treatment procedures as per the progress the patient is making.
Melo is a convolutional neural network that has been trained with over 2,000 samples of both benign and malignant Melanoma tumors. To train itself, Melo breaks down the images into RGB values and finds "features of interest" of both benign and malignant Melanoma samples. Then, it finds a correlation between these features and whether a melanoma sample is malignant or not. After doing this, Melo can identify features of interest in future samples to determine the corresponding label class for the image.