Model Methodology, Validation and Limits
This page documents how the system produces a result, what it was measured at, and where it should not be trusted. Gender is never requested from the user — it is inferred from the image.
Medically reviewed by Dr. Öğr. Üyesi Mehmet Ali Gedik — Radiologist, Kütahya Şehir Hastanesi
Last reviewed:
1. Routing (user first, model as fallback)
The under/over 18 branch decides which specialist model runs and in which unit the result is reported, so it is asked to the user first: a declared branch is binding. If the user does not answer, a dedicated gate model classifies the radiograph as under or over 18 years (threshold: 216 months) and routing follows that. The gate is reliable but not infallible near the threshold — this is exactly why the declaration takes precedence.
- Minor specialist: bone age in months. Two independent systems run in parallel; in each, a coarse router first makes a rough estimate, then the age-band expert covering that estimate produces the final value (divide and conquer). The two systems' outputs are combined.
- Adult specialist: age in years, from a single hybrid model.
2. Architecture
Multi-task convolutional networks (EfficientNet-B3 and ConvNeXt V2-Tiny backbones, 512 px input; 384 px for the gate). Age is learned as a probability distribution rather than a single number, and the prediction is the expectation of that distribution. Horizontal-flip test-time augmentation is applied to stabilise the output. All models are image-only: no clinical data, patient identity or user-supplied gender enters the inference.
3. Internal validation results
Each model's own validation score, read directly from the trained checkpoints. MAE = mean absolute error.
| Model | Backbone | Input | Score | Gender acc. |
|---|---|---|---|---|
| Gate (under/over 18) | ConvNeXt V2-T | 384 px | AUC 0.994 · accuracy 96.3% | 92.3% |
| Minor — coarse router | EfficientNet-B3 | 512 px | MAE 11.0 months | 87.1% |
| Minor — band A (io) | EfficientNet-B3 | 512 px | MAE 8.0 months | 65.1% |
| Minor — band B (io) | EfficientNet-B3 | 512 px | MAE 11.1 months | 80.9% |
| Minor — band C (io) | EfficientNet-B3 | 512 px | MAE 10.1 months | 90.4% |
| Minor — 2nd system router | ConvNeXt V2-T | 512 px | MAE 10.4 months | 86.9% |
| Minor — band A (cnx) | ConvNeXt V2-T | 512 px | MAE 6.9 months | 76.4% |
| Minor — band B (cnx) | ConvNeXt V2-T | 512 px | MAE 10.0 months | 80.9% |
| Minor — band C (cnx) | ConvNeXt V2-T | 512 px | MAE 9.4 months | 86.3% |
| Adult | EfficientNet-B3 | 512 px | MAE 3.9 years | 94.3% |
How to read this table
- These are per-model figures on an internal validation split. They are NOT the accuracy of the full pipeline: the minor result comes from two systems combined, so its error differs from any single row.
- Internal validation always flatters a model. Real-world performance on your own patient population will differ and is the only number that should guide clinical use.
- The low gender accuracy of some band experts is not a defect: the routing model's gender output is used when a band expert is uncertain.
4. Why adult accuracy is lower
The adult model's error is measured in years, not months, and this is a physiological limit rather than a software shortcoming: hand and wrist growth plates close around 18-19 years, after which the radiograph carries very little age-discriminating signal. No model can recover information the image does not contain. Treat over-18 output as a coarse estimate, and do not use it where a legally decisive age determination is required.
5. Validity scope
Expected performance applies to well-positioned PA hand-wrist radiographs resembling the training and validation data. Atypical anatomy, skeletal dysplasia, surgical implants, casts, severe artifacts, burned-in text, cropped images or incorrect positioning may all degrade the result — sometimes without any visible sign that the estimate is wrong.
6. Intended use
This is a clinical decision-support tool, not a diagnostic device, and it has no regulatory clearance as one. Every result must be reviewed by a physician or radiology specialist together with clinical findings. We recommend that institutions run an acceptance test on their own sample before putting it into routine use.
7. Continuous measurement
Users can report the patient's real age after an analysis. These reports let us measure error against ground truth in actual use, and we intend to publish those figures on this page as enough data accumulates. Note they will be voluntary — and therefore biased — so the trend over time matters more than the absolute value.
References
The clinical concept of skeletal maturity assessment this tool automates rests on the following standard works. They describe the reference methods, not this system's training data.
- Greulich WW, Pyle SI. Radiographic Atlas of Skeletal Development of the Hand and Wrist. 2nd ed. Stanford University Press, 1959.
- Tanner JM, Healy MJR, Goldstein H, Cameron N. Assessment of Skeletal Maturity and Prediction of Adult Height (TW3 Method). 3rd ed. WB Saunders, 2001.
- Halabi SS et al. The RSNA Pediatric Bone Age Machine Learning Challenge. Radiology. 2019;290(2):498-503.
Questions about the methodology? Contact us · See also the FAQ.