This project focuses on the classification of kidney diseases using deep learning and computer vision techniques. We have developed a complete pipeline encompassing data ingestion, transformation, model training, and evaluation. The project leverages technologies such as Next.js, TypeScript, TailwindCSS for the frontend, and Python, Flask, mlflow, DVC for the backend.
Frontend URL: https://kidney-disease-classification.vercel.app/
- (Model Notebooks can be found in Research Folder)
- (Model Results can be found in Results Folder)
- Comprehensive data processing and transformation pipeline.
- Multiple deep learning, Transformers and machine learning models for comparative analysis.
- Complete backend implementation in Flask with logging, custom exception handling, and class-based modular coding following OOPs principles.
- Frontend developed using Next.js, TailwindCSS, TypeScript and Shadcn.UI for a seamless user experience.
CT KIDNEY DATASET: Normal-Cyst-Tumor and Stone
The dataset comprises 12,446 unique data points, including 3,709 cyst, 5,077 normal, 1,377 stone, and 2,283 tumor findings. It was collected from PACS from various hospitals in Dhaka, Bangladesh. After thorough selection and anonymization processes, the images were converted to a lossless jpg format and verified by medical professionals.
Dataset Link: https://www.kaggle.com/datasets/nazmul0087/ct-kidney-dataset-normal-cyst-tumor-and-stone/data
- Nextjs 14
- TypeScript
- TailwindCSS
- Shadcn.UI
- Python
- DVC
- MLFlow
- Deep Learning & Computer Vision
- Transformers & HuggingFace
- Flask
- OOP Principles
- TNT transformer
- DEIT transformer
- ConVIT transformer
- VIT Transformer
- ResNet50
- InceptionV3
- MobileNetV2
- DenseNet121
- Xception
- EfficientNet Series (B0 to B3)
- VGG16 and VGG19
- Traditional ML Models (Random Forest, SVM, KNN, Naive Bayes, Decision Tree, Logistic Regression, Gradient Boosting, AdaBoost, Extra Trees)
- Ensemble Methods (Hard Voting, Soft Voting, Stacking)
- Update config.yaml
- Update secrets.yaml [Optional]
- Update params.yaml
- Update the entity
- Update the configuration manager in src config
- Update the components
- Update the pipeline
- Update the main.py
- Update the dvc.yaml
- app.py
Clone the repository
git clone https://github.com/Prriyanshu9898/Kidney-Disease-Classification-Using-Deep-Learning.gitcd Kidney-Disease-Classification-Using-Deep-Learning
python -m venv envenv\Scripts\activatepip install -r requirements.txtpython app.pypython main.pycd frontendnpm installnpm run devnpm run build- mlflow ui
MLFLOW_TRACKING_URI=https://dagshub.com/Priyanshu9898/Kidney-Disease-Classification-Deep-Learning-Project.mlflow
MLFLOW_TRACKING_USERNAME=Priyanshu9898
MLFLOW_TRACKING_PASSWORD=1dc505ed931b2af16eacead37f82f256c16d99fe
python script.py
Run this to export as env variables:
export MLFLOW_TRACKING_URI=https://dagshub.com/Priyanshu9898/Kidney-Disease-Classification-Deep-Learning-Project.mlflow
export MLFLOW_TRACKING_USERNAME=Priyanshu9898
export MLFLOW_TRACKING_PASSWORD=1dc505ed931b2af16eacead37f82f256c16d99fe
| Model | Accuracy | Precision | Recall | F1 Score |
|---|---|---|---|---|
| Stacking | 100.0% | 100.0% | 100.0% | 100.0% |
| K-Nearest Neighbors | 99.92% | 99.88% | 99.94% | 99.91% |
| Extra Trees | 99.84% | 99.9% | 99.64% | 99.77% |
| ConVIT transformer | 99.84% | 99.9% | 99.64% | 99.77% |
| Soft Voting | 99.6% | 99.74% | 99.09% | 99.4% |
| InceptionV3 | 99.52% | 99.52% | 99.52% | 99.52% |
| Random Forest | 99.44% | 99.66% | 98.85% | 99.24% |
| Hard Voting | 99.36% | 99.58% | 98.6% | 99.07% |
| SVM | 99.28% | 99.38% | 98.79% | 99.07% |
| EfficientNetB3 | 98.96% | 98.96% | 98.96% | 98.95% |
| Xception | 98.83% | 98.88% | 98.83% | 98.82% |
| EfficientNetB2 | 98.67% | 98.69% | 98.67% | 98.67% |
| Gradient Boosting | 98.47% | 98.53% | 97.08% | 97.75% |
| TNT transformer | 98.31% | 98.31% | 98.31% | 98.31% |
| EfficientNetB1 | 98.27% | 98.31% | 98.27% | 98.26% |
| Logistic Regression | 98.15% | 97.68% | 97.2% | 97.42% |
| VIT transformer | 96.43% | 96.3% | 94.58% | 95.33% |
| ResNet50 | 96.26% | 96.34% | 96.26% | 96.24% |
| DEIT transformer | 96.22% | 96.22% | 96.22% | 96.22% |
| DenseNet121 | 95.94% | 95.96% | 95.94% | 95.87% |
| MobileNetV2 | 95.42% | 95.72% | 95.42% | 95.29% |
| EfficientNetB0 | 95.3% | 95.61% | 95.3% | 94.98% |
| Decision Tree | 92.77% | 90.63% | 91.3% | 90.95% |
| VGG16 | 91.68% | 91.83% | 91.68% | 91.43% |
| AdaBoost | 85.3% | 81.99% | 78.92% | 80.22% |
| VGG19 | 83.97% | 83.23% | 83.97% | 82.57% |
| Naive Bayes | 52.69% | 60.53% | 45.11% | 37.68% |
Insert gif or link to demo
Add badges from somewhere like: shields.io
