NA-MIC Project Weeks
Back to Projects List
Mauritanian Project
Key Investigators
- Ahmedou Moulaye IDRISS (Faculté de médecine de Nouakchott, Université de Nouakchott, Mauritania)
- Sonia Pujol (Brigham and Women’s Hospital and Harvard Medical School, USA)
- Fatimetou Mohamed-Saleck (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- Moustapha Mohamed Saleck (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- Mohamed Mahmoud Septy Mohamed bamba (Faculté de médecine de Nouakchott, Université de Nouakchott, Mauritania)
- Mohamed Abdellahi Sidi Mohamed Blal (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- El Hacen Mohamed Soueilem (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- Mohamedou Ahmed Mahmoud (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- Mohamed Boullah (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
- Fatimetou Hademine (Faculté des Sciences et Techniques, Université de Nouakchott, Mauritania)
Project Description
The Mauritanian approach aims to make 3D Slicer accessible to a broader audience of users and researchers in the medical field. Various projects have been completed, initiated, or are currently under consideration, including:
- The anatomy atlases created by professors Idriss and Yahya in 2018 enabled medical students to visualize and quickly assimilate human body parts through the use of 3D Slicer.
- Other applications for generating 3D models of baby and expectant mother mannequins.
- Implementation and integration of three breast cancer segmentation methods by one of our researchers: An adaptive fuzzy C-means algorithm, An adaptive k-means algorithm, and an adaptive Otsu thresholding. Integration is underway.
- Processing and segmentation of medical images with 3D Slicer and Deep Learning in the context of an integrated approach for analyzing various medical data from Mauritania. This is the subject of an ongoing doctoral thesis.
- Other upcoming applications are planned.
Objective
- Generate 3D models of baby and expectant mother mannequins.
- Contribute to enriching breast cancer segmentation methods by introducing and comparing user choices in 3D Slicer.
- Explore cutting-edge techniques in medical image processing and segmentation, especially those based on Deep Learning, implement them, and evaluate performance.
- Anticipate additional upcoming applications.
Approach and Plan
- 3D Model of Baby and Expectant Mother
- Introduction of 3D Segmentation Methods:
- Segmentation of Medical Images from Mauritania:
- Other upcoming applications:
Progress and Next Steps
- Progress in generating 3D models of the baby and expectant mother.
- Progress in integrating the 3D segmentation methods.
- Progress of the doctoral thesis: Segmentation of Medical Images from Mauritania.
- Other upcoming projects.
Illustrations
We have developed an extension for 3D Slicer to perform medical image (volume) segmentation using the K-Means algorithm. Specifically, we have implemented an adaptive version of K-Means, which allows segmentation based on pixel intensity.
We encountered several challenges during volume processing and rendering, as well as in finding alternatives to libraries like scikit-learn, NumPy, and OpenCV to integrate them into the 3D Slicer API.
Ultimately, we successfully segmented the images using both the adaptive K-Means and the classic K-Means methods. However, these results still require improvement and testing on various types of medical images to ensure their reliability
Adaptive Algorithm Result
Classic Algorithm Result
