This repository contains the codes and scripts used in ImageCLEF Tuberculosis - Severity Scoring 2019 challenge. The challenge was dedicated to the analysis of 3D Computed Tomography (CT) image data of tuberculosis (TB) patients.
This task is aimed at assessing TB severity score. The Severity score is a cumulative score of severity of TB case assigned by a medical doctor. Originally, the score varied from 1 (“critical/very bad”) to 5 (“very good”). The goal of this subtask is to assess the severity based on the CT image. According to competition rules for final evaluation, the task was reduced to a binary classification problem.
This work is implemented in Python 3.6 and Keras using Tensorflow as backend.
Tested code using:
- Ubuntu 14.04
- Windows 8
- Python 3.6
main: Contains codes to final submissionutils: Contains helper codes to preprocess and visualize samples in dataset.
In this edition, both subtasks (SVR and CTR) use the same dataset containing 335 chest CT scans of TB patients along with a set of clinically relevant metadata. 218 patients are used for training and 117 for test. For all patient, provided are 3D CT images with slice size of 512*512 pixels and number of slices varying from about 50 to 400. All the CT images are stored in NIFTI file format with .nii.gz file extension (g-zipped .nii files). The dataset is provided by ImageCLEF and not shared here due to competition rules. More information at URL.
Here is a single CT image displayed slice by slice.
| Experiment | Results | Others |
|---|---|---|
| 16-layer 3D Convolutional Net | 61.1% AUC, 61.5% ACC | 128x128x32, 300 epochs,10th place solution |
If this was helpful in your work, please consider citing.
@inproceedings{zunair2019estimating,
title={Estimating Severity from CT Scans of Tuberculosis Patients using 3D Convolutional Nets and Slice Selection.},
author={Zunair, Hasib and Rahman, Aimon and Mohammed, Nabeel},
booktitle={CLEF (Working Notes)},
year={2019}
}
Paper published in CEUR-WS. Paper can be found at CLEF Working Notes 2019 under the section ImageCLEF - Multimedia Retrieval in CLEF.
Paper and code of an extension version can be found here.