Individualized gamma knife radiosurgery prescription dosing for pituitary adenomas: development and internal validation of a feedforward neural network model
Authors' conclusion
This proof-of-concept FNN generated individualized GKRS prescription dose estimates for pituitary adenomaswith acceptable prediction error, supporting the feasibility of data-driven dose recommendation tools and motivating external validation in larger, multi-institutional cohorts.
Study context
- Population
- Pituitary adenomas treated with Gamma Knife radiosurgery (GKRS) at a single center
- Participants
- 102 pituitary adenomas
- Outcome
- Prediction of GKRS prescription dose (Gy); test performance by MAE, RMSE and R²
Methodological detailssingle-centre, feedforward neural network, 80:20 random training/test split, permutation importance
- Single-centre
- Feedforward neural network
- 80:20 random training/test split
- Permutation importance
Performance
As reported in the abstract
Test-set performance of feedforward neural network (FNN) for GKRS prescription dose
Mean absolute error (MAE)
1.61 Gy
Root mean squared error (RMSE)
2.31 Gy
R²
0.51
Further findings
Predictor contribution (permutation importance)
- Age at treatment: greatest contribution, ΔMAE with permutation
- +0.89 Gy
- Followed by functional or secretory status, tumor volume, cavernous sinus invasion and pre-treatment Karnofsky Performance Status (KPS)
- Demographic variables and biopsy-related factors had minimal impact
Original publication
Individualized gamma knife radiosurgery prescription dosing for pituitary adenomas: development and internal validation of a feedforward neural network model. Pituitary. Published 9 February 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →