# Virtual 3D surface imaging system for the validation of tumor tracking algorithm used in surface‐guided radiotherapy

## TL;DR

This paper introduces a virtual 3D imaging system to help develop and test tumor tracking algorithms used in radiotherapy.

## Contribution

A novel simulation system using PMP and Unity is developed to support SGRT tumor tracking algorithm validation.

## Key findings

- The simulation system's RMSE values are sub-millimeter (0.46-0.52 mm), matching real-world results.
- The system enables testing under varied conditions without relying on idealized data.
- The system supports SGRT algorithm development and reduces real-world measurement costs.

## Abstract

To develop a surface 3D reconstruction simulation system based on the Phase Measurement Profilometry (PMP) and the Unity physics engine and validate its feasibility for assisting in the development of the Surface‐Guided Radiotherapy (SGRT) tumor tracking algorithm.

The components, such as cameras and projectors, are set up in the Unity environment to enable structured‐light‐based surface 3D reconstruction simulation using the PMP. This process includes structured light projection, camera calibration, phase unwrapping, and point cloud reconstruction procedures. The influence of parameter settings on the effectiveness of 3D reconstruction is investigated, including different distances and angles between the camera and the measurement surface, as well as variations in light intensity. The simulation capabilities of the system are validated by comparing surface imaging of the same human torso model in a radiotherapy room environment and within the simulation system. Additionally, the simulation system is further utilized to acquire surface imaging data required for the SGRT tumor tracking algorithm. A comparison is made between this data and the idealized skin surface imaging data obtained directly from CT reconstruction segmentation to verify the system's supportive role in the development of the SGRT tumor tracking algorithm.

The effects of varying light intensity and object positioning in the simulation system are consistent with those reported in previous studies conducted in real‐world environments. The root mean square errors (RMSE) of the surface imaging point clouds from different perspectives between the simulation system and the actual environment are 0.46, 0.47, and 0.52 mm, all at sub‐millimeter levels. The validation in the development of the SGRT tumor tracking algorithm indicates that the simulation system enables the SGRT algorithm development to avoid relying on overly idealized surface imaging data.

The simulation system based on PMP and Unity has been proposed, enabling a broader range of measurement conditions to be set in the virtual environment, thereby saving costs for measurements in real‐world scenarios. This system can also be utilized to assist in the validation of the SGRT tumor tracking algorithms, thereby advancing progress in this field.