Reproducibility – README

Project Title
Acute Lymphoblastic Leukemia Cancer Diagnosis in Children and Adults using Transforming Blood Fluorescence Microscopy Imaging
Description
This work use ResNet-CNN sequential architecture that provides the basis for a novel approach of 
classifying acute lymphoblastic leukemia. This method utilizes transfer learning, customized layers, 
and a limited number of ReLU units to extract crucial features from blood microscopy data. 
The experiments are conducted using jupyter notebook and NVIDIA GEforce GPU.

Dataset
The dataset used in this study is publicly available at:
- Database name: Acute Lymphoblastic Leukemia (ALL) image dataset; Kaggle
- Link: https://doi.org/10.34740/KAGGLE/DSV/2175623
- Dataset reference: Mehrad Aria, Mustafa Ghaderzadeh, Davood Bashash, Hassan Abolghasemi, Farkhondeh Asadi, and Azamossadat Hosseini, “Acute Lymphoblastic Leukemia (ALL) image dataset.” Kaggle, (2021). DOI: 10.34740/KAGGLE/DSV/2175623

Code Information
Installation and Setup
1. Setup Environment
2. Ensure Python is installed.
3. Ensure tensorflow is installed
4. Install the required libraries using:
5. GPU Connection

  

Running the Code

Step 1: Install pakages

pip install tensorflow 
pip install keras 
pip install opencv-python 
pip install numpy 
pip install pandas 
pip install matplotlib 
pip install scikit-learn


Step 2: Datset loading
        Preprocessing
        Data splitting
        Reshaping Data


Step 3: Requirements
To run a model for acute lymphoblastic leukemia cancer diagnosis in children and adults efficiently, we need a system with at least an NVIDIA GPU (e.g., Geforce RTX with 8GB VRAM), though 8–24GB VRAM 
GPUs like RTX 3060/3090 for 
higher performance and larger batch sizes. The system should have at least 8GB RAM and CPU Intel i5/i7.
other requirements
- numpy, pandas
- opencv-python, Pillow, scikit-image
- tensorflow, keras, grad-cam, scikit-learn, matplotlib, seaborn
- pip install numpy pandas opencv-python Pillow scikit-image tensorflow keras scikit-learn matplotlib seaborn


Methodology
Step 4: Model Training
- Trainined proposed model
- Evaluation
- Testing
- Results- confusion matrix, performance metrics, AUC, comulative gain
Detection reults
- Accuracy metrics for each model
- Training and testing accuracy as well as loss curves
- Confusion matrices for classification
- XAI 

Citation:Mehrad Aria, Mustafa Ghaderzadeh, Davood Bashash, Hassan Abolghasemi, Farkhondeh Asadi, and Azamossadat Hosseini, “Acute Lymphoblastic Leukemia (ALL) image dataset.” Kaggle, (2021). DOI: 10.34740/KAGGLE/DSV/2175623

License:
Not applicable