From Jan. 1, 2019 to Jan. 1, 2020
A detailed description with images illustrating the clinical context can be found in the PDF attached in supplementary materials.
The goal of this challenge is to build an image classifier to assist physicians in the screening and diagnosis of esophageal cancer.
Such a tool would have a massive impact on patient management and patient lives.
There are 11161 images acquired from 61 patients to be classified as:
- Squamous Epithelium
- Intestinal Metaplasia
- Gastric Metaplasia
A detailed description of the dataset can be found in the PDF attached in supplementary materials.
The order in which the predictions must be submitted can be found in the file test_order.csv in the supplementary material.
The submission file should be a .cvs file with (N+1) lines (N = number of images in the test set, +1 = header) and 2 columns (1st column = im_XXX, 2nd column = class digit).
A detailed description of the clinical criteria for each class in Cellvizio images is provided in the ANNEXE document.
Data is provided under Creative Commons BY-NC-SA license.
Results were obtained with a standard convolutional neural network (CNN, 3 convolutional layers, each convolutional layer followed by dropout and pooling).
The non-weighted accuracy was 75% on the test set.
Note that this result is without performing any:
- Images pre-processing
- Images augmentation (noise, linear transforms or local deformations)
- Features computation
Files are accessible when logged in and registered to the challenge
The challenge provider