In search of a brain microbiome: A machine learning search pipeline for electron microscopy images of brain tissue
Matelsky, J. K.; Shih, C.; Hijazi, K.; Johnson, E. C.
Show abstract
The healthy human brain has long been considered a sterile environment, with the blood brain barrier preventing the formation of a bacterial brain microbiome. Recent electron microscopy (EM) imaging of brain tissue has, however, provided the first preliminary evidence of bacteria in otherwise healthy brain slices. Whether due to contamination, disease, or a previously unknown relationship of bacteria to healthy brain tissue, novel tools are needed to detect and search for bacteria in nanoscale, volumetric EM images. While computer vision tools are widely used in cell segmentation and object detection problems in EM imaging, no bacteria detection tool or dataset exists. Overcoming the rarity of training data, this work presents the first pipeline for training a bacteria detection network for EM images, leveraging existing deep networks for object detection. A deployment and proofreading pipeline is presented, along with characterization of deployment to public EM image datasets. While bacteria in healthy brain tissue were not discovered in this work, this tool presents an opportunity for large scale bacteria search in EM imaging for both scientific discovery and experimental quality control, and serves more generally as a framework for sparse object detection in large imagery datasets.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Deep Learning Classification of Lipid Droplets in Quantitative Phase Images 96%
- Semantic Segmentation of HeLa Cells: An Objective Comparison between one Traditional Algorithm and Three Deep-Learning Architectures 95%
- microbeSEG: A deep learning software tool with OMERO data management for efficient and accurate cell segmentation 94%
Similar papers in this journal
- THINGSvision: a Python toolbox for streamlining the extraction of activations from deep neural networks 93%
- ShuTu: Open-Source Software for Efficient and Accurate Reconstruction of Dendritic Morphology 92%
- Systems Neuroscience Computing in Python (SyNCoPy): A Python Package for Large-scale Analysis of Electrophysiological Data 92%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.