CINNAMON-GUI is an advanced digital pathology tool that leverages the power of convolutional neural networks (CNN) and multimodal learning to provide an intuitive graphical user interface for research professionals. This tool is designed to enhance the analysis and interpretation of pathological samples, making it easier to study various diseases. Cinnamon-GUI has evolved from its initial implementation with ipywidgets for Jupyter Lab and Notebook to a fully-fledged Shiny app in Python. This transition brings several advantages, notably an enhanced user interface and improved interactivity. The Shiny framework allows for dynamic and responsive web applications, making it more straightforward for users to interact with machine learning models and visualize results in real time. This update ensures that Cinnamon-GUI remains at the forefront of digital pathology tools, providing a more robust and user-friendly experience for researchers and clinicians. The user interface, built with Shiny components, provides a responsive and interactive experience. Images and plots are generated dynamically using base64 encoding, ensuring efficient memory usage and performance. You can choose two primary operational pathways within the Cinnamon-GUI platform: 1) Data Processing and Model Training Workflow, where users can load pickle files from datasets to train models and generate reports post-testing, and 2) Biospecimen Annotation and Analysis Workflow, where users can load annotated Pap smear images, with annotations applied using the integrated Labelme software. This workflow supports loading the corresponding JSON annotation files and converting them and the images into a pickle dataset format for subsequent cell type prediction and classification. Additionally, this pathway facilitates the creation and expansion of datasets, enhancing the utility of Cinnamon-GUI for diverse research applications. Installation Download the latest version of CINNAMON-GUI from our official repository (https://github.com/lunanfoldomics/Cinnamon-GUI/). Follow the installation guide provided in this README.md document to set up the software on your system. Cinnamon-GUI has undergone rigorous testing in MacOS environments for M1 processors. Virtual Environment To run the code, we recommend using a conda virtual environment. You can create a virtual environment named cinnamongui and install the required dependencies by executing the following commands: # Create a new virtual environment cinnamongui conda create -n cinnamongui python=3.9 Activate the virtual environment conda activate cinnamongui # On Windows source activate cinnamongui # On macOS and Linux To install the necessary packages for Cinnamon-GUI, you can create a requirements.txt file with the following content. This file lists all the dependencies that need to be installed: shiny os io re shutil pandas numpy json tensorflow Pillow matplotlib scikit-learn datetime pickle base64 tempfile asyncio labelme You can install these dependencies by running: pip install -r requirements.txt Deactivate the virtual environment conda deactivate # On Windows source deactivate # On macOS and Linux Running Cinnamon-Gui cd dashboard-tips python cinnamon-gui.py
About
Cinnamon-GUI is a state-of-the-art software tool designed to redefine digital pathology. Licensed under the AGPLv3, this user-friendly interface harnesses the power of Convolutional Neural Networks (CNNs) and pre-trained models to classify biomedical images with unparalleled precision. Key Features Empowering Image Classification: Cinnamon-GUI excels in classifying images of cells from diverse biospecimens, such as cervical pap smears, and accurately identifying precancerous cells. Advanced Research Capabilities: For researchers and medical professionals, Cinnamon-GUI offers a comprehensive suite of features, including advanced feature mapping. This feature aids in identifying critical patterns in cell evolution. Training Customization: Users can customize the architecture of the Convolutional Neural Network using the intuitive Training Customization interface, allowing for tailored model training to meet specific research needs. Versatility: While initially tailored to cervical pathology, Cinnamon-GUI is adaptable to any pathology, making it a versatile tool for a wide range of biomedical research applications.
