Brain image segmentation
Training a convolutional neural network to process grayscale brain slice images.
The project includes development of Matlab code for training a convolutional neural network and segmentation of grayscale brain slice images based on the trained model.
A certain complexity of development was related to a limited number of provided labeled images for training stage. That’s why additional measures for data augmentation were implemented to the algorithm to generate more training data based on small distortions of an original image. After that, parameters of a neural network pre-trained for image segmentation (using PASCAL VOC dataset) were fine-tuned in order to work on a particular dataset.
Finally, the accuracy of prediction for available images was estimated based on Dice’s coefficient.
Key functionality:
- Image processing
- Image segmentation
- Convolutional neural network
Similar Projects
Virtual try-on tool for makeup products
The system consists of a face detection and segmentation model and an algorithm that allows recoloring objects without losing their original texture.
Online sign language interpreter
AI algorithm that converts video of a person using sign language into a text transcript
Cardiac coherence training app
Mobile application with a set of exercises to practice breathing techniques.