Automatic high content screening using deep learning
Bibliographic record
Abstract
Recently, deep learning algorithms have been used with success in a variety of domains. Deep learning has proven to be a very helpful tool for discovering complicated structures in high-dimensional and big datasets. In this work, five deep learning models inspired by AlexNet, VGG, and GoogleNet are developed to predict mechanism of actions (MOAs) based on phenotypic screens of a number of cells in dimly lit and noisy images. We demonstrate that our models can predict the MOA for a compendium of drugs that alter cells through single cell or cell population views without any segmentation and feature extraction steps. According to these results, our models do not need to fully realize single-cell measurements to profile samples because they use the morphology of specific phenomena in the cell population samples. We used an imbalanced High Content Screening big dataset to predict MOAs with the main goal of understanding how to work properly with deep learning algorithms on imbalanced datasets when sampling methods, like Oversampling, Undersampling, and Synthetic Minority Over-sampling (SMOTE) algorithms are used for balancing the dataset. Based on our findings, it is now clear that the SMOTE sampling algorithm must be part of the deep learning algorithms when confronting imbalanced datasets. High Content Screening technologies have to deal with screening thousands of cells to provide a number of parameters for each cell, such as nuclear size, nuclear morphology, DNA replication, etc. The success of High Content Screening (HCS) systems depends on automatic image analysis. Recently, deep learning algorithms have overcome object recognition challenges on tasks with a single centered object per image. Present deep learning algorithms have not been applied to images that include multiple specific complex objects, such as microscopic images of many objects such as cells in these images.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".