Controlling the Scanning Electron Microscope with Deep Learning for Automated Acquisition and Image Quality Improvement
Bibliographic record
Abstract
Controlling image acquisition hardware with image processing software is the future of microscopy.The efficiency of microscopy techniques is currently limited by manual intervention and the need for microscopist supervision during acquisitions.More importantly, proper interpretation of image data requires quantitative analysis with computational methods.Automating simple and complex tasks on the scanning electron microscope (SEM) specifically, is feasible by replacing microscopist involvement with appropriate software algorithms and deep learning.Guided acquisitions on the SEM with hybrid methods is the solution to higher throughput and to generating better data, faster.In this work, computational approaches to microscopy are presented.Foremost, two workflows for automated acquisitions are described, for imaging and image processing in Dragonfly, an image processing software.First, a feedforward workflow sequentially positions the beam in a predetermined grid layout, stitches the images, segments the feature of interest and computes measurements for quantitative analysis.Second, a workflow with a feedback loop between imaging and analysis is developed to enable dynamic imaging with feature segmentation and feature tracking.Guiding the hardware to image areas of interest with smart beam positioning is an optimized solution for acquiring only relevant data and minimizing acquisition duration.Then, microscope routines, prior to acquisitions, are enhanced by integrating computational methods, focused on artificial intelligence.Microscopists spend a considerable amount of time tuning microscope parameters and aligning the beam until a desired image quality is observed.To assist with interpretation of image quality, a regression model is trained to optimize SEM v
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 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.006 | 0.003 |
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".