Multiview motion tracking based on a cartesian robot to monitor Caenorhabditis elegans in standard Petri dishes

Automation of Caenorhabditis elegans lifespan assay using a simplified domain synthetic image-based neural network training strategy

Performing lifespan assays with Caenorhabditis elegans (C. elegans) nematodes manually is a time consuming and laborious task. Therefore, automation is necessary to increase productivity. In this paper, we propose a method to automate the counting of live C. elegans using deep learning. The survival curves of the experiment are obtained using a sequence formed by an image taken on each day of the assay. Solving this problem would require a very large labeled dataset; thus, to facilitate its generation, we propose a simplified image-based strategy. This simplification consists of transforming the real images of the nematodes in the Petri dish to a synthetic image, in which circular blobs are drawn on a constant background to mark the position of the C. elegans. To apply this simplification method, it is divided into two steps. First, a Faster R-CNN network detects the C. elegans, allowing its transformation into a synthetic image. Second, using the simplified image sequence as input, a regression neural network is in charge of predicting the count of live nematodes on each day of the experiment. In this way, the counting network was trained using a simple simulator, avoiding labeling a very large real dataset or developing a realistic simulator. Results showed that the differences between the curves obtained by the proposed method and the manual curves are not statistically significant for either short-lived N2 (p-value log rank test 0.45) or long-lived daf-2 (p-value log rank test 0.83) strains.

Towards generalization for Caenorhabditis elegans detection

The nematode Caenorhabditis elegans (C. elegans) is of significant interest for research into neurodegenerative diseases, aging, and drug screening. However, conducting these assays manually is a tedious and time-consuming process. This paper proposes a methodology to achieve a generalist C. elegans detection algorithm, as previous work only focused on dataset-specific detection, tailored exclusively to the characteristics and appearance of the images in a given dataset. The main aim of our study is to achieve a solution that allows for robust detection, regardless of the image-capture system used, with the potential to serve as a basis for the automation of numerous assays. These potential applications include worm counting, worm tracking, motion detection and motion characterization. To train this model, a dataset consisting of a wide variety of appearances adopted by C. elegans has been curated and dataset augmentation methods have been proposed and evaluated, including synthetic image generation. The results show that the model achieves an average precision of 89.5% for a wide variety of C. elegans appearances that were not used during training, thereby validating its generalization capabilities.

Automatic segmentation of Caenorhabditis elegans skeletons in worm aggregations using improved U-Net in low-resolution image sequences

Pose estimation of C. elegans in image sequences is challenging and even more difficult in low-resolution images. Problems range from occlusions, loss of worm identity, and overlaps to aggregations that are too complex or difficult to resolve, even for the human eye. Neural networks, on the other hand, have shown good results in both low-resolution and high-resolution images. However, training in a neural network model requires a very large and balanced dataset, which is sometimes impossible or too expensive to obtain. In this article, a novel method for predicting C. elegans poses in cases of multi-worm aggregation and aggregation with noise is proposed. To solve this problem we use an improved U-Net model capable of obtaining images of the next aggregated worm posture. This neural network model was trained/validated using a custom-generated dataset with a synthetic image simulator. Subsequently, tested with a dataset of real images. The results obtained were greater than 75% in precision and 0.65 with Intersection over Union (IoU) values.

Adult-restricted gene knock-down reveals candidates that affect locomotive healthspan in C. elegans

Understanding how we can age healthily is a challenge at the heart of biogerontological interest. Whereas myriad genes are known to affect the lifespan of model organisms, effects of such interventions on healthspan-the period of life where an animal is considered healthy, rather than merely alive-are less clear. To understand relationships between life- and healthspan, in recent years several platforms were developed with the purpose of assessing both readouts simultaneously. We here relied on one such platform, the WorMotel, to study effects of adulthood-restricted knock-down of 130 Caenorhabditis elegans genes on the locomotive health of the animals along their lifespans. We found that knock-down of six genes affected healthspan while lifespan remained unchanged. For two of these, F26A3.4 and chn-1, knock-down resulted in an improvement of healthspan. In follow-up experiments we showed that knockdown of F26A3.4 indeed improves locomotive health and muscle structure at old age.

Analysis of a C. elegans lifespan prediction method based on a bimodal neural network and uncertainty estimation

In recent decades, assays with the nematode Caenorhabditis elegans (C. elegans) have enabled great advances to be made in research on aging. However, performing these assays manually is a laborious task. To solve this problem, numerous C. elegans assay automation techniques are being developed to increase throughput and accuracy. In this paper, a method for predicting the lifespan of C. elegans nematodes using a bimodal neural network is proposed and analyzed. Specifically, the model uses the sequence of images and the count of live C. elegans up to the current day to predict the lifespan curve termination. This network has been trained using a simulator to avoid the labeling costs of training such a model. In addition, a method for estimating the uncertainty of the model predictions has been proposed. Using this uncertainty, a criterion has been analyzed to decide at what point the assay could be halted and the user could rely on the model's predictions. The method has been analyzed and validated using real experiments. The results show that uncertainty is reduced from the mean lifespan and that most of the predictions obtained do not present statistically significant differences with respect to the curves obtained manually.

A new training strategy for spatial transform networks (STN’s)

Spatial transform networks (STN) are widely used since they can transform images captured from different viewpoints to obtain an objective image. These networks use an image captured from any viewpoint as input and the desired image as a label. Usually, these images are segmented, but this could lead to convergence problems if the percentage of overlap between the segmented images is quite low. In this paper, we propose a new training method to facilitate the convergence of a STN in these cases, even when there is no overlap between the object’s projections in the two images. This new strategy is based on the incorporation of the distance transformation images to the training, thus increasing the useful image information to provide gradients in the loss function. This new training strategy has been applied to a real case, with images of Caenorhabditis elegans, and to a simulated case, which uses artificial images to ensure that there is no overlap between the images used for the assays. In the assays carried out with these datasets, we have shown that the training convergence is strengthened, reaching a precision level for IoU metric of 0.862 and 0.984, respectively, and the computational cost has been maintained compared to the assay with segmented images, for the real case.