Prior user


Algorithm Logo

About

Editor:
User Mugshot Dbence 
Contact email:
Image Version:
aabdaa05-0ff3-48ff-b2d1-02b19ef6b536 — Aug. 10, 2026
Model Version:
2411dc3c-42ac-4995-ae25-fe22b92ea9ef — Aug. 6, 2026

Summary

Physics Prior

We have updated the model with a special gaussian tube, which is a continous line along the gantry angle and a gaussian decay in every other direction. This forces the model to concentrate in this section. Additionally we added a similar prior, but it estimates a possible endpoint for the beam energy deposition, making it easier to learn.

The model backbone is just a simple Encoder->Transformer->Decoder pipeline inspired by DoTA.

Mechanism

Since this algorithm was designed for the DoseRad Challange it takes CT images, initial energy and ray source/target pairs that are used to generate the priors.


Interfaces

This algorithm implements all of the following input-output combinations:

Inputs Outputs
1
    Radiation-Dose Calculation Source CT Image 1
    Radiation-Dose Calculation Source CT Image 2
    Radiation-Dose Calculation Source CT Image 3
    Radiation-Dose Calculation Source CT Image 4
    Radiation-Dose Calculation Source CT Image 5
    Radiation-Dose Calculation Source CT Image 6
    Radiation-Dose Calculation Source CT Image 7
    Radiation-Dose Calculation Source CT Image 8
    Radiation-Dose Calculation Source CT Image 9
    Radiation-Dose Calculation Source CT Image 10
    Stacked Proton Beam-Level Metadata
    Stacked Radiation-Dose Map 1
    Stacked Radiation-Dose Map 2
    Stacked Radiation-Dose Map 3
    Stacked Radiation-Dose Map 4
    Stacked Radiation-Dose Map 5
    Stacked Radiation-Dose Map 6
    Stacked Radiation-Dose Map 7
    Stacked Radiation-Dose Map 8
    Stacked Radiation-Dose Map 9
    Stacked Radiation-Dose Map 10

Validation and Performance

Uses and Directions

This algorithm was developed for research purposes only.

Warnings

Common Error Messages

Information on this algorithm has been provided by the Algorithm Editors, following the Model Facts labels guidelines from Sendak, M.P., Gao, M., Brajer, N. et al. Presenting machine learning model information to clinical end users with model facts labels. npj Digit. Med. 3, 41 (2020). 10.1038/s41746-020-0253-3