ISIC 2018: Skin Lesion Analysis Towards Melanoma Detection
This repository presents results of our work to detect Melanoma by Skin Lesion Analysis.
The International Skin Imaging Collaboration (ISIC) is an international effort to improve melanoma diagnosis, sponsored by the International Society for Digital Imaging of the Skin (ISDIS). The ISIC Archive contains the largest publicly available collection of quality controlled dermoscopic images of skin lesions.
The goal of this challenge is to help participants develop image analysis tools to enable the automated diagnosis of melanoma from dermoscopic images.
This challenge is broken into three separate tasks:
Task 1: Lesion Segmentation
Task 2: Lesion Attribute Detection Task 3: Disease Classification
Ours results are for Task 3: Disease Classification
Possible disease categories are:
|3||Basal cell carcinoma|
|4||Actinic keratosis / Bowen’s disease (intraepithelial carcinoma)|
|5||Benign keratosis (solar lentigo / seborrheic keratosis / lichen planus-like keratosis)|
Following Graphs show Model Accuracy for Training and Testing Phase; Model Loss for Training and Testing Phase; and Computation Time for Training the model and Training plus Tresting the model.
Model Accuracy, Model Loss, and Training Time is available at model
Class wise probabilities for each valid image is available at Valid Results
Class wise probabilities for each test image is available at Test Results
Our position at ISIC Live Challenge Leaderboards at the time of uploading the results for Task 3 is 23 and we are working to improve the results.
Following are five random images that were picked and detected one of the disease categories by our model
Below is the Training Dataset consisting of 10,015 images for different disease categories
|3||Basal cell carcinoma||514|
|4||Actinic keratosis / Bowen’s disease (intraepithelial carcinoma)||327|
|5||Benign keratosis (solar lentigo / seborrheic keratosis / lichen planus-like keratosis)||1099|
Please submit your feedback to Dr. Nagender Aneja. Please write an email (email@example.com) if you are interested to impement the model in a mobile app or web app. We welcome people and organization who can provide more data on plants from different countries to join this project.