Dataset Information

Partitioning variability in animal behavioral videos using semi-supervised variational autoencoders.

ABSTRACT: Recent neuroscience studies demonstrate that a deeper understanding of brain function requires a deeper understanding of behavior. Detailed behavioral measurements are now often collected using video cameras, resulting in an increased need for computer vision algorithms that extract useful information from video data. Here we introduce a new video analysis tool that combines the output of supervised pose estimation algorithms (e.g. DeepLabCut) with unsupervised dimensionality reduction methods to produce interpretable, low-dimensional representations of behavioral videos that extract more information than pose estimates alone. We demonstrate this tool by extracting interpretable behavioral features from videos of three different head-fixed mouse preparations, as well as a freely moving mouse in an open field arena, and show how these interpretable features can facilitate downstream behavioral and neural analyses. We also show how the behavioral features produced by our model improve the precision and interpretation of these downstream analyses compared to using the outputs of either fully supervised or fully unsupervised methods alone.

SUBMITTER: Whiteway MR

PROVIDER: S-EPMC8489729 | biostudies-literature | 2021 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Partitioning variability in animal behavioral videos using semi-supervised variational autoencoders.

Whiteway Matthew R MR Biderman Dan D Friedman Yoni Y Dipoppa Mario M Buchanan E Kelly EK Wu Anqi A Zhou John J Bonacchi Niccolò N Miska Nathaniel J NJ Noel Jean-Paul JP Rodriguez Erica E Schartner Michael M Socha Karolina K Urai Anne E AE Salzman C Daniel CD Cunningham John P JP Paninski Liam L

PLoS computational biology 20210922 9

Recent neuroscience studies demonstrate that a deeper understanding of brain function requires a deeper understanding of behavior. Detailed behavioral measurements are now often collected using video cameras, resulting in an increased need for computer vision algorithms that extract useful information from video data. Here we introduce a new video analysis tool that combines the output of supervised pose estimation algorithms (e.g. DeepLabCut) with unsupervised dimensionality reduction methods t ...[more]

PMID: 34550974

Similar Datasets

Project description:Background Animal illness is a disturbance in an animal’s natural condition that disrupts or changes critical functions. Concern over animal illnesses stretches back to the earliest human interactions with animals and is mirrored in early religious and magical beliefs. Animals have long been recognized as disease carriers. Man has most likely been bitten, stung, kicked, and gored by animals for as long as he has been alive; also, early man fell ill or died after consuming the flesh of deceased animals. Man has recently learned that numerous invertebrates are capable of transferring disease-causing pathogens from man to man or from other vertebrates to man. These animals, which function as hosts, agents, and carriers of disease, play a significant role in the transmission and perpetuation of human sickness. Thus, there is a need to detect unhealthy animals from a whole group of animals. Methods In this study, a deep learning-based method is used to detect or separate out healthy-unhealthy animals. As the dataset contains a smaller number of images, an image augmentation-based method is used prior to feed the data in the deep learning network. Flipping, scale-up, sale-down and orientation is applied in the combination of one to four to increase the number of images as well as to make the system robust from these variations. One fuzzy-based brightness correction method is proposed to correct the brightness of the image. Lastly, semi-supervised generative adversarial network (SGAN) is used to detect the healthy-unhealthy animal images. As per our knowledge, this is the first article which is prepared to detect healthy-unhealthy animal images. Results The outcome of the method is tested on augmented COCO dataset and achieved 91% accuracy which is showing the efficacy of the method. Conclusions A novel two-fold animal healthy-unhealthy detection system is proposed in this study. The result gives 91.4% accuracy of the model and detects the health of the animals in the pictures accurately. Thus, the system improved the literature on healthy-unhealthy animal detection techniques. The proposed approach may effortlessly be utilized in many computer vision systems that could be confused by the existence of a healthy-unhealthy animal.

Project description:Cancer is a complex disease that deregulates cellular functions at various molecular levels (e.g., DNA, RNA, and proteins). Integrated multi-omics analysis of data from these levels is necessary to understand the aberrant cellular functions accountable for cancer and its development. In recent years, Deep Learning (DL) approaches have become a useful tool in integrated multi-omics analysis of cancer data. However, high dimensional multi-omics data are generally imbalanced with too many molecular features and relatively few patient samples. This imbalance makes a DL based integrated multi-omics analysis difficult. DL-based dimensionality reduction technique, including variational autoencoder (VAE), is a potential solution to balance high dimensional multi-omics data. However, there are few VAE-based integrated multi-omics analyses, and they are limited to pancancer. In this work, we did an integrated multi-omics analysis of ovarian cancer using the compressed features learned through VAE and an improved version of VAE, namely Maximum Mean Discrepancy VAE (MMD-VAE). First, we designed and developed a DL architecture for VAE and MMD-VAE. Then we used the architecture for mono-omics, integrated di-omics and tri-omics data analysis of ovarian cancer through cancer samples identification, molecular subtypes clustering and classification, and survival analysis. The results show that MMD-VAE and VAE-based compressed features can respectively classify the transcriptional subtypes of the TCGA datasets with an accuracy in the range of 93.2-95.5% and 87.1-95.7%. Also, survival analysis results show that VAE and MMD-VAE based compressed representation of omics data can be used in cancer prognosis. Based on the results, we can conclude that (i) VAE and MMD-VAE outperform existing dimensionality reduction techniques, (ii) integrated multi-omics analyses perform better or similar compared to their mono-omics counterparts, and (iii) MMD-VAE performs better than VAE in most omics dataset.

Dataset Information

Partitioning variability in animal behavioral videos using semi-supervised variational autoencoders.

Publications

Partitioning variability in animal behavioral videos using semi-supervised variational autoencoders.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets