AloNet Supplementary Repository
datasetposted on 31.10.2019 by Solam Lee
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
AloNet Author: Solam Lee (email@example.com) AloNet is a convolutional neural network based on U-Net that can identify the hair loss and the scalp area by analying clinical photograph. This model was developed for the automated calculation of the Severity of Alopecia Tools (SALT) score in assessment of patients with alopecia areata. This repository posts the Mendeley Supplementary Materials, the program code, and the relevant data used in the paper titled "Clinically Applicable Deep Learning Framework for Measurement of the Severity of Alopecia Tool Score in Patients with Alopecia Areata". Along with the programs in the "/Program/" directory, a total of 2716 pixelwise annotations used for train the hair loss identifier (mask) and the hair loss identifier (target) could be find in the "/Data/" directory. However, please note that the clinical photograph of the patients could not be made publicly available because of strict privacy regulation. Before using AloNet program with your dataset, you should convert your dataset into numpy files. One clinical photograph (saved in .jpg with RGB format) need each annotation for the scalp area (saved in .gif with black&white color) and the hair loss (saved in .gif with black&white color), respectively. Please make sure that they have same image size each other, or the conversion will fail. We are now currently working on several postprocessing algorithms for AloNet to be available for general use. The Flask web application and its code will be made available publicly when the program is ready to use.