Although current research aims to improve deep learning networks by applying knowledge about the healthy human brain and vice versa, the potential of using such networks to model and study neurodegenerative diseases remains largely unexplored. In this work, we present an in-depth feasibility study modeling progressive dementia in silico with deep convolutional neural networks. Therefore, networks were trained to perform visual object recognition and then progressively injured by applying neuronal as well as synaptic injury. After each iteration of injury, network object recognition accuracy, saliency map similarity between the intact and injured networks, and internal activations of the degenerating models were evaluated. The evaluation revealed that cognitive function of the network progressively decreased with increasing injury load whereas this effect was much more pronounced for synaptic damage. The effects of neurodegeneration found for the in silico model are especially similar to the loss of visual cognition seen in patients with posterior cortical atrophy.

Neuron morphology gives rise to distinct axons and dendrites and plays an essential role in neuronal functionality and circuit dynamics. In rat hippocampal neurons, morphological development occurs over roughly one week in vitro. This development has been qualitatively described as occurring in 5 stages. Still, there is a need to quantify cell growth to monitor cell culture health, understand cell responses to sensory cues, and compare experimental results and computational growth model predictions. To address this need, embryonic rat hippocampal neurons were observed in vitro over six days, and their processes were quantified using both standard morphometrics (degree, number of neurites, total length, and tortuosity) and new metrics (distance between change points, relative turning angle, and the number of change points) based on the Change-Point Test to track changes in path trajectories. Of the standard morphometrics, the total length of neurites per cell and the number of endpoints were significantly different between 0.5, 1.5, and 4 days in vitro, which are typically associated with Stages 2-4. Using the Change-Point Test, the number of change points and the average distance between change points per cell were also significantly different between those key time points. This work highlights key quantitative characteristics, both among common and novel morphometrics, that can describe neuron development in vitro and provides a foundation for analyzing directional changes in neurite growth for future studies.

We present MedicDeepLabv3+, a convolutional neural network that is the first completely automatic method to segment cerebral hemispheres in magnetic resonance (MR) volumes of rats with ischemic lesions. MedicDeepLabv3+ improves the state-of-the-art DeepLabv3+ with an advanced decoder, incorporating spatial attention layers and additional skip connections that, as we show in our experiments, lead to more precise segmentations. MedicDeepLabv3+ requires no MR image preprocessing, such as bias-field correction or registration to a template, produces segmentations in less than a second, and its GPU memory requirements can be adjusted based on the available resources. We optimized MedicDeepLabv3+ and six other state-of-the-art convolutional neural networks (DeepLabv3+, UNet, HighRes3DNet, V-Net, VoxResNet, Demon) on a heterogeneous training set comprised by MR volumes from 11 cohorts acquired at different lesion stages. Then, we evaluated the trained models and two approaches specifically designed for rodent MRI skull stripping (RATS and RBET) on a large dataset of 655 MR rat brain volumes. In our experiments, MedicDeepLabv3+ outperformed the other methods, yielding an average Dice coefficient of 0.952 and 0.944 in the brain and contralateral hemisphere regions. Additionally, we show that despite limiting the GPU memory and the training data, our MedicDeepLabv3+ also provided satisfactory segmentations. In conclusion, our method, publicly available at https://github.com/jmlipman/MedicDeepLabv3Plus , yielded excellent results in multiple scenarios, demonstrating its capability to reduce human workload in rat neuroimaging studies.