SlideLab AI — NTD Vision

Track: Health & Public Safety

SlideLab AI is an AI-powered microscopy tool that assists healthcare workers in detecting malaria from stained blood-smear images. The system uses a deep learning model to classify cells as parasitized or uninfected, aiming to support diagnosis in resource-limited settings.

---

Team

• Adeniyi Micheal Oluwafemi
  Fellow ID: FE/23/44017546
  Role: Team Lead
  Contact: oluwafemiadeniyi772@gmail.com

• Ifeanyi Hyacint Muotoe
  Fellow ID: FE/25/9928453022

• Olutunde Stephen Anuoluwa
  Fellow ID: FE/23/85039993

---

Overview

Malaria diagnosis in many regions still relies heavily on manual microscopy, which is time-consuming and depends on the skill and availability of trained personnel. SlideLab AI provides an AI-assisted second opinion by analysing digitised blood-smear images and returning a prediction in real time.

The current version focuses on malaria, but the architecture is designed to be extended to other neglected tropical diseases (NTDs) in the future.

---

Features

• AI-assisted slide analysis
  Upload a stained blood-smear image and receive an instant prediction.

• Binary malaria classification
  Classifies images into two classes: parasitized and uninfected.

• Visual feedback
  Displays the uploaded image together with the predicted class and probability.

• Extensible design
  The pipeline can be adapted to support additional NTDs such as filariasis and loiasis.

---

Dataset

The model is trained on a publicly available malaria blood-smear image dataset from the
Lister Hill National Center for Biomedical Communications (LHNCBC) at the
U.S. National Library of Medicine.

• Thin and thick blood-smear microscopy images
• Expert-annotated labels for parasitized vs. uninfected cells
• Anonymised images collected under real clinical conditions

This dataset provides realistic examples of what diagnostic labs in malaria-endemic regions encounter.

---

Model

• Architecture: EfficientNetB0
• Input size: 180 × 180 pixels
• Number of classes: 2 (parasitized, uninfected)
• Preprocessing: tf.keras.applications.efficientnet.preprocess_input

During development, standard evaluation tools (confusion matrix, ROC curve, probability histograms) were used to assess performance and calibration.

---

Project Structure

Current repository layout:

• Malaria_Cell_Classification_Model.h5
  Trained malaria cell classification model.

• Malaria_Cell_Classification_Model_Notebook.ipynb
  Jupyter notebook used for model training, experiments, and analysis.

• StreamlitWebApp.py
  Main Streamlit application used for running the web interface and model inference.

• requirements.txt
  Python dependencies required to run the project.

• README.md
  Project documentation (this file).

• LICENSE
  MIT license for this project.

---

How to Run

1. Clone the repository

   git clone https://github.com/OfemiAdeniyi/malaria-app.git
   cd malaria-app

2. Create and activate a virtual environment (optional but recommended)

   python -m venv venv
   # Windows
   venv\Scripts\activate
   # macOS / Linux
   source venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. Start the Streamlit app

   streamlit run StreamlitWebApp.py

5. Use the app

   • Open the URL shown in the terminal (usually http://localhost:8501).
   • Upload a stained blood-smear image.
   • View the predicted class (parasitized or uninfected) and the associated probability.

---

Limitations and Disclaimer

• The current version is trained and evaluated on a specific public dataset and may not cover all slide preparation techniques or imaging conditions.
• This tool is intended for research and demonstration purposes only.
  It is not a certified medical device and should not be used as a standalone basis for clinical decision-making.

---

License

This project is licensed under the MIT License – see the LICENSE file for details.