Integrating Machine Learning with Flow-Imaging Microscopy for Automated Monitoring of Algal Blooms
Bibliographic record
Abstract
Real-time monitoring of phytoplankton in freshwater systems is critical for early detection of harmful algal blooms (HABs) to enable efficient response by water management agencies. This manuscript presents an image processing pipeline developed to adapt ARTiMiS, a low-cost automated flow-imaging device, for real-time algal monitoring in natural freshwater systems. This pipeline addresses several challenges associated with autonomous imaging of aquatic samples, such as flow-imaging artifacts (i.e., out-of-focus and background objects), as well as strategies to efficiently identify novel objects that are not represented in the training data set; the latter is a common challenge with the application of deep learning approaches for image classification in environmental systems. The pipeline leverages a random forest model to identify out-of-focus particles with an accuracy of 89% and a custom background particle detection algorithm to identify and remove particles that erroneously appear in consecutive images with >97 ± 2.8% accuracy. Furthermore, a convolutional neural network (CNN), trained to classify taxonomical classes, achieved 95% accuracy in a closed set classification. Nonetheless, the supervised closed-set classifiers struggled with the accurate classification of objects when challenged with novel particles, which are common in complex natural environments; this limits real-time monitoring applications by requiring extensive manual oversight. To mitigate this, three methods incorporating classification with rejection were tested to improve model precision by flagging irrelevant or unknown classes. Combined, these advances present a fully integrated, end-to-end solution for real-time HAB monitoring in natural freshwater systems, which enhances the scalability of automated detection in dynamic aquatic environments.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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".