Evaluation of a Deep Active Learning Model for the Segmentation of Canine Thoracic Radiographs
Bibliographic record
Abstract
With increasing interest in artificial intelligence (AI) for veterinary medical imaging, there will be an increasing need for the segmentation of medical images. Image segmentation-the process of delineating anatomical structures in medical images-is a critical step for enabling analysis and decision support in veterinary radiology. Manual segmentation of medical images is a time-consuming and tedious task associated with user variation. Many segmentation tasks require a radiologist's expertise. To date, there have been limited evaluations of segmentation methods in veterinary medicine. It is unknown whether novice evaluators can segment radiographs with similar accuracy to experts. The present study aimed to evaluate the performance of an AI segmentation tool in enhancing the accuracy and reducing the time of canine radiograph segmentation of novice, intermediate, and expert users when using an internally developed software that allows both AI-assisted semiautomated and manual segmentation. The AI model was trained using 50 thoracic radiographs from patients referred to the Ontario Veterinary College between January 2020 and July 2021. The intersection over union scores (IoU) for the abdomen, heart, and spinous process labels were higher when all cohorts used the semiautomated method (0.98, 0.98, and >0.74, respectively) versus the manual method (>0.93, >0.94, and >0.42, respectively). The Hausdorff distance for the structure labels was significantly lower when the participants used the semiautomated method than the manual method (p < .0001). The intraobserver intraclass correlation coefficients (ICC) for the semiautomatic and manual methods were 0.81 and 0.36, respectively. In conclusion, the semiautomated tool effectively assisted users with segmenting canine thoracic radiographs.
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 | 0.003 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| 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 | 0.000 |
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".