Deep learning model shows promise for detecting and grading sesamoiditis in horse radiographs
Bibliographic record
Abstract
OBJECTIVE: The objective of this study was to develop a robust machine-learning approach for efficient detection and grading of sesamoiditis in horses using radiographs, specifically in data-limited conditions. SAMPLE: A dataset of 255 dorsolateral-palmaromedial oblique (DLPMO) and dorsomedial-palmarolateral oblique (DMPLO) equine radiographs were retrospectively acquired from Hagyard Equine Medical Institute. These images were anonymized and classified into 3 categories of sesamoiditis severity (normal, mild, and moderate). METHODS: This study was conducted from February 1, 2023, to August 31, 2023. Two RetinaNet models were used in a cascaded manner, with a self-attention module incorporated into the second RetinaNet's classification subnetwork. The first RetinaNet localized the sesamoid bone in the radiographs, while the second RetinaNet graded the severity of sesamoiditis based on the localized region. Model performance was evaluated using the confusion matrix and average precision (AP). RESULTS: The proposed model demonstrated a promising classification performance with 92.7% accuracy, surpassing the base RetinaNet model. It achieved a mean average precision (mAP) of 81.8%, indicating superior object detection ability. Notably, performance metrics for each severity category showed significant improvement. CLINICAL RELEVANCE: The proposed deep learning-based method can accurately localize the position of sesamoid bones and grade the severity of sesamoiditis on equine radiographs, providing corresponding confidence scores. This approach has the potential to be deployed in a clinical environment, improving the diagnostic interpretation of metacarpophalangeal (fetlock) joint radiographs in horses. Furthermore, by expanding the training dataset, the model may learn to assist in the diagnosis of pathologies in other skeletal regions of the horse.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".