Neonatal Mouse Heart Segmentation with 3D UNet

Our project focuses on accurately segmenting three critical cardiac classes: Atrium, Ventricle, and Trabeculae. Leveraging advanced techniques in data preprocessing and augmentation, we have enhanced the robustness and generalization capabilities of the model. Our efforts have culminated in an impressive IoU Score of 0.84, affirming the model's exceptional accuracy in segmenting these distinct cardiac structures.

Project Highlights

1. 3D UNet Model: We have designed and fine-tuned a 3D UNet model tailored to the specific challenges of neonatal mouse heart segmentation. This model leverages the power of deep learning to provide precise and reliable segmentation results.

2. Three Essential Classes: The model is trained to segment three vital cardiac structures: Atrium, Ventricle, and Trabeculae. This enables detailed analysis and understanding of the neonatal mouse heart's anatomy.

3. Data Preprocessing: To improve model performance, we have implemented robust data preprocessing techniques. This includes data normalization, alignment, and quality enhancement, ensuring that the model receives the best input for accurate segmentation.

4. Data Augmentation: Data augmentation is a key factor in enhancing model generalization. Our project includes various data augmentation strategies to expose the model to a wide range of variations, ultimately improving its ability to handle diverse real-world scenarios.

5. Impressive IoU Score: The model's performance is validated by achieving an IoU Score of 0.84, showcasing its high accuracy in accurately segmenting neonatal mouse heart structures.

Getting Started

To explore and utilize this project, follow these steps:

i. Clone this repository to your local machine. ii. Review the code and documentation to understand the model architecture and training process. iii. Access the preprocessed and augmented dataset used for training and evaluation. iv. Experiment with the model on your own data or use our provided dataset for segmentation tasks. v. Explore the model's results and evaluate its performance using the included evaluation metrics.