Dataset Information

A comparative study of cell classifiers for image-based high-throughput screening.

ABSTRACT: BACKGROUND: Millions of cells are present in thousands of images created in high-throughput screening (HTS). Biologists could classify each of these cells into a phenotype by visual inspection. But in the presence of millions of cells this visual classification task becomes infeasible. Biologists train classification models on a few thousand visually classified example cells and iteratively improve the training data by visual inspection of the important misclassified phenotypes. Classification methods differ in performance and performance evaluation time. We present a comparative study of computational performance of gentle boosting, joint boosting CellProfiler Analyst (CPA), support vector machines (linear and radial basis function) and linear discriminant analysis (LDA) on two data sets of HT29 and HeLa cancer cells. RESULTS: For the HT29 data set we find that gentle boosting, SVM (linear) and SVM (RBF) are close in performance but SVM (linear) is faster than gentle boosting and SVM (RBF). For the HT29 data set the average performance difference between SVM (RBF) and SVM (linear) is 0.42 %. For the HeLa data set we find that SVM (RBF) outperforms other classification methods and is on average 1.41 % better in performance than SVM (linear). CONCLUSIONS: Our study proposes SVM (linear) for iterative improvement of the training data and SVM (RBF) for the final classifier to classify all unlabeled cells in the whole data set.

SUBMITTER: Abbas SS

PROVIDER: S-EPMC4287552 | biostudies-literature | 2014

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A comparative study of cell classifiers for image-based high-throughput screening.

Abbas Syed Saiden SS Dijkstra Tjeerd M H TM Heskes Tom T

BMC bioinformatics 20141021

<h4>Background</h4>Millions of cells are present in thousands of images created in high-throughput screening (HTS). Biologists could classify each of these cells into a phenotype by visual inspection. But in the presence of millions of cells this visual classification task becomes infeasible. Biologists train classification models on a few thousand visually classified example cells and iteratively improve the training data by visual inspection of the important misclassified phenotypes. Classific ...[more]

PMID: 25336059

Similar Datasets

Project description:Studies on monolayer cultures and whole-animal models for the prediction of the response of native human tissue are associated with limitations. Therefore, more and more laboratories are tending towards multicellular spheroids grown in vitro as a model of native tissues. In addition, they are increasingly used in a wide range of biofabrication methodologies. These 3D microspheroids are generated through a self-assembly process that is still poorly characterised, called cellular aggregation. Here, a system is proposed for the automated, non-invasive and high throughput monitoring of the morphological changes during cell aggregation. Microwell patterned inserts were used for spheroid formation while an automated microscope with 4x bright-field objective captured the morphological changes during this process. Subsequently, the acquired time-lapse images were automatically segmented and several morphological features such as minor axis length, major axis length, roundness, area, perimeter and circularity were extracted for each spheroid. The method was quantitatively validated with respect to manual segmentation on four sets of ± 60 spheroids. The average sensitivities and precisions of the proposed segmentation method ranged from 96.67-97.84% and 96.77-97.73%, respectively. In addition, the different morphological features were validated, obtaining average relative errors between 0.78-4.50%. On average, a spheroid was processed 73 times faster than a human operator. As opposed to existing algorithms, our methodology was not only able to automatically monitor compact spheroids but also the aggregation process of individual spheroids, and this in an accurate and high-throughput manner. In total, the aggregation behaviour of more than 700 individual spheroids was monitored over a duration of 16 hours with a time interval of 5 minutes, and this could be increased up to 48,000 for the described culture format. In conclusion, the proposed system has the potential to be used for unravelling the mechanisms involved in spheroid formation and monitoring their formation during large-scale manufacturing protocols.

Dataset Information

A comparative study of cell classifiers for image-based high-throughput screening.

Publications

A comparative study of cell classifiers for image-based high-throughput screening.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets