Evaluation of Automated Segmentation of Intracranial Radiosensitive Structures in iPlan Image and its Effects on iPlan Dose PDF

Keywords

iPlan treatment planning system, SRS/SRT, radiosensitive structures, radiotherapy, ×‑ray.

Abstract

The BrainLab iPlan Treatment Planning System (TPS) was used to model a clinac ×-ray photon beam with the BrainLab stereotactic radiosurgery (SRS) apparatus. The head and neck region of the anthropomorphic Rando phantom was scanned using both CT and MRI imaging modalities. Imaged slices were fused together for better contrast using the automatic image fusion provided with iPlan RT image software. A 1.25 cm3 tumour was assumed and defined at the back of the brain. Nine intracranial radiosensitive structures were segmented manually and automatically by using the atlas-based auto-segmentation tool implemented in iPlan RT image software. The iPlan RT dose software was then employed to estimate the dose received by 50% of each of these structures and the tumour. Volumes and doses of automatically and manually segmented structures were then compared. Generally, it was found that iPlan RT image overestimates the volume of intracranial structures except two of them, the right and the left eye, were underestimated. The dose received by radiosensitive structures exposed to direct ×-ray beam were affected by segmentation discrepancies, while the off-beam structures were not. It was found that auto-segmentation helped in reducing the time required for segmentation by considerable amounts with acceptable accuracy. Finally, an important recommendation is to explore the possibility of predefining the radiological properties of different types of tumour cells in the code for quick and accurate auto-segmentation of Gross Tumour Volume (GTV), for better dose estimation.