Each file is a Python-based project that I worked on throughout 2023. IDEs, such as Pycharm and Jupyter Notebook, were used in the creations of these programs.
- Pycharm Version: PyCharm 2023.1.3 (Community Edition), Build #PC-231.9161.41, built on June 20, 2023, Windows 11.0, GC: G1 Young Generation, G1 Old Generation
- Python Version: Python 3.11
- Jupyter Notebook:
The motivation driving this project was the probability of whether a combination of social factors could either delay the onset of AD or prevent its development altogether, and whether this information could be used to provide insight into AD prediction. Using ML, an approach can be developed utilizing supervised learning to create a categorical model. By evaluating the mean accuracies of all feature-pairs generated by the algorithm, a conslusion can be made on which feature-pair(s) have a strong interaction that relates to the maintenance of brain health.
The data utilized was sourced from the Open Access Series of Imaging Studies (OASIS) website. OASIS has made available a comprehensive cross-sectional MRI image dataset that is freely accessible for educational purposes and scientific investigations. The data collection occurred at Washington University, St. Louis, USA, which involves younger and middle-aged individuals from the local community associated with the university. The data concerning older individuals and their dementia diagnosis, it was sourced from the Washington University Alzheimer’s Disease Research Center (ADRC). The dataset consists of 436 individuals across a wide range of 18 to 96 years of age. Within this dataset, 100 individuals from the age group of 45 years to 96 years were identified to have mild to moderate Alzheimer’s disease. Other subjects were from the age group of 18 years to 44 years. The 12 attributes of this data set are ID, M/F, Hand, Age, Educ: Education, SES: Social Economic Status, MMSE: Mini-Mental state evaluation, CDR: Clinical Dementia Rating, eTIV: Intra-cranial volume, nWBV: Normal Whole Brain Volume, ASF: Atlas Scaling Factor, and Delay. https://www.oasis-brains.org/
The data set sourced from OASIS was cleaned, underwent some feature engineering. Columns where values were equal to and greater than 50% of the observations were removed. All missing values were removed from the dataset.
Multiple classifier models (Random Forest, KNN, SVC, MLP, and Gaussian NB) were tested for performance in this program. The selected ML was selected based on the best mean accuracy.
The purpose of this program is to enhance the MRI images in order to create segmentation masks that allow for computational visualization of tumours in the prostate.
In this section, several things were done: volume pre-processing, testing of different filters for enhancement and segmentation, and the selection and improvement of a segmentation mask. In the pre-processing step, the volume was rescaled to the range 0-255. Due to the large number of grey pixels in the volume, enhancing filters were used to brighten the image. The Median filter, Mean filter, and the Intensity Window filters were tried and tested. From there, thresholding was performed for the segmentation phase using the Binary Threshold Filter, and a combination of the Binary and Median filters, and the Confidence Connected. Finally, a clean up was performed on the selected segmentation filter using a vector filter which implements binary morphological operations, such as Binary Morphological Opening and Binary Morphological Closing.
Main function that executes the functions in the 'utils.py' file.
The purpose of this program was to explore and visualize medical images in 3D Slicer and Simple ITK and to apply simple transformations to the images.