A Fully Convolutional Network for Rodent Brain Lesion Segmentation. The main developer has been Juan Miguel Valverde.
RatLesNetv2 is a convolutional neural network implemented in Python+Pytorch to segment rodent brain lesions. The code of RatLesNetv2 is simplified to make it readable and accessible to a wide audience.
This implementation of RatLesNetv2 allows combining several models trained separately. This script will generate a prediction per model and a prediction combined with majority voting. Post-processing, i.e. removing small independently connected components (holes and islands) is also available.
References:
Juan Miguel Valverde, Artem Shatillo, Riccardo de Feo, Olli Gröhn, Alejandra Sierra, Jussi Tohka: RatLesNetv2: A Fully Convolutional Network for Rodent Brain Lesion Segmentation https://arxiv.org/abs/2001.09138
Juan Miguel Valverde, Artem Shatillo, Riccardo de Feo, Olli Gröhn, Alejandra Sierra, Jussi Tohka. Automatic Rodent Brain MRI Lesion Segmentation with Fully Convolutional Networks. Machine Learning in Medical Imaging, Lecture Notes in Computer Science 2019. (Earlier version)
Multi-task U-Net for the simultaneous segmentation and skull-stripping of mouse brain MRI . This convolutional neural network is designed to perform skull-stripping and region segmentation on mouse brain MRI. This network is trained on coronal T2 mouse brain MRI delineated with a bounding box, and so for the network to function correctly MRI volumes need to be cropped to a bounding box around the brain. To automate this task we include a lightweight auxiliary network. The main developer has been Riccardo De FEo.
Reference:
Riccardo De Feo, Artem Shatillo, Alejandra Sierra, Juan Miguel Valverde, Olli Gröhn, Federico Giove, Jussi Tohka. Automated skull-stripping and segmentation with Multi-Task U-Net in large mouse brain MRI databases .bioRxiv 2020.02.25.964015; doi: https://doi.org/10.1101/2020.02.25.964015