Diagnosing Gastrointestinal Diseases from Endoscopy Images through a Multi-Fused CNN with Auxiliary Layers, Alpha Dropouts, and a Fusion Residual Block

Multi-Fused Residual Convolutional Neural Network (MFuRe-CNN)

• Author: Dr. Francis Jesmar P. Montalbo

• Affiliation: Batangas State University

• Email: francismontalbo@ieee.org

Personal Webpage

PLEASE CONTANCT ME IF YOU ARE HAVING TROUBLE. I CAN OFFER ASSITANCE

Graphical Abstract

CITATION:

F. J. P. Montalbo, "Diagnosing Gastrointestinal Diseases from Endoscopy Images through a Multi-Fused CNN with Auxiliary Layers, Alpha Dropouts, and a Fusion Residual Block," Biomedical Signal Processing and Control (BSPC), vol. 76, July, 2022, doi: 10.1016/j.bspc.2022.103683

F. J. P. Montalbo, "Fusing Compressed Deep ConvNets with a Self-Normalizing Residual Block and Alpha Dropout for a Cost-Efficient Classification and Diagnosis of Gastrointestinal Tract Diseases," MethodsX, In-Press, November 2022. doi: 10.1016/j.mex.2022.101925.

Paper link: FULL PAPER LINK (READ FIRST) Methods Paper: FULL PAPER LINK (READ FIRST)

Datasets used:

KVASIR Dataset

Paper to cite:

KVASIR: A Multi-Class Image Dataset for Computer Aided Gastrointestinal Disease Detection

ETIS-Larib-Polyp DB Dataset

Paper to cite:

Towards embedded detection of polyps in WCE images for early diagnosis of colorectal cancer

===========================================================================

❗For a faster method, you may download the already prepared dataset used in the given link below.

CLICK ME FOR THE PREPARED DATASET USED IN THIS STUDY. Download the data.rar and extract it to the mfurecnn/data

===========================================================================

How to use:

Disclaimer

❗If training the model, the dependencies included a tensorflow-gpu. You may change the tensorflow-gpu to tensorflow if no GPU is to be used. However, the results from the paper were produced using a GPU (RTX 3060 12gb) and may have slight differences

Dependencies included in the requirements.txt:

• jupyter==1.0.0
• keras==2.4.3
• matplotlib==3.4.1
• numpy==1.19.5
• opencv-python==3.4.11.41
• pandas==1.2.4
• Pillow==8.2.0
• scikit-learn==0.24.1
• scikit-image==0.18.1
• scikit-plot==0.3.7
• scipy==1.2.0
• tf-nightly-gpu==2.6.0 (Note: This is optional and can train even with just a CPU or tensorflow non-gpu variant. Nightly is used to compensate the new RTX 3060 card)

===========================================================================

General Instruction:

You may clone using git or download the repository and extract the files manually:

• Once cloned, CD into the folder and enter pip install -r requirements.txt.
• After installation of the dependecies, there are two options, either evaluate from the given weights (EASY and FAST) or train the model again (DIFFICULT and LENGTHY).
• Download the readily trained weights and dataset here ---> Dataset and Trained Weights
• Extract the data.rar in mfurecnn/data and the models.rar in mfurecnn/models ===========================================================================

❗ HOW TO USE:

First (easier):

• Clone this repository or download as zip.

• Install the requirements on a newly created environment to prevent issues with other existing ones.

• Directly open the 006-Evaluator.ipynb go to the Kernel tab above then proceed with Restart & Run All to evaluate.

• In cell #5 of the 006-Evaluator.ipynb the architecture and condition variables have values that can be changed to evaluate the other models. As architecture has a list of ['MFuRe', 'MFNR'] and the condition has ['alpha', 'standard', 'no']. Values can be interchanged as needed depending on the model to be evaluated.

• For testing, open the 007-Tester_with_gradcam.ipynb go to the Kernel tab above then proceed with Restart & Run All to test.

• In cell #9 of the 007-Tester_with_gradcam.ipynb the case has values that can be changed to evaluate the various cases in isolation 0 for normal, 1 for ulcer, 2 for poylp, 3 for esophagitis.

• The saliency analysis can be found in 008-gradcams.ipynb for further visualization with other CAM algorithms.

Second (difficult):

• Make sure to download the PREPARED dataset and extract it to a folder within the mfure_cnn/ like mfure_cnn/data/

• Clone this repository or download as zip.

• Install the requirements on a newly created environment to prevent issues with other existing ones.

• Open one of the trainers like 000-MFuReCNN_alpha_do.ipynb then go to the Kernel tab above then proceed with Restart & Run All to train. You may train the other models if needed that has notebook numbers from 000 to 005. If there are trained models that exist in the mfurecnn/models/ they will be overwritten depending on the model re-trained. BE CAREFUL.

• Once trained, you may now again use the 006-Evaluator.ipynb, 007-Tester_with_gradcam.ipynb, 008-gradcams.ipynb, and you are done. Make sure that the respective weights are present.

REMEMBER THIS IS A LONGER PROCESS (Second process) WHEN TESTING AND SIMULATING THE MODEL.

Score-CAM is credited to https://github.com/tabayashi0117/Score-CAM