MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Target Definition On Multi-Modality MRI Image Sets for Gliomas

J Feng1*, W Jiang2, Z Wang3, Y Sa4, (1) Tianjin University & Tianjin University Huanhu Hospital (2) Tianjin University Huanhu Hospital (3) Tianjin University Huanhu Hospital (4) Tianjin University, Tianjin

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Brain tumor segmentation is a challenging task due to the heterogeneous nature of tumor and appearance. Functional MRI such as apparent diffusion constant (ADC) and Fractional anisotropy (FA) could be incorporated with the anatomical MRI to achieve more accurate characterization of tumor extent. In this study, a new method was proposed and tested for segmentation on tri-modality MRI image sets.

Methods: Tri-modality MRI image sets including T2 weighted, ADC and FA from five patients with gliomas were used. A restricted potential field segmentation method was developed which uses the local potential field of pixels in the region of interest (ROI) for initial segmentation and then morphologically processed for accurate delineation of the GTV. The GTV contours were delineated manually by an experienced radiation oncologist and generated using the proposed method, respectively. The contours generated with the two methods were compared and the efficiencies were analyzed with the parameters of time complexity (O) and segmentation quality measure (Q).

Results: Time complexity with the new method was reduced from O(n^2) to O(n×25). The mean Q was 0.79(±0.06), 0.69(±0.09), 0.64(±0.09) and 0.75(±0.08) for T2, ADC, FA and with the fused image set of them, respectively.

Conclusion: The tri-modality MRI segmentation approach with restricted potential field segmentation method has the potential to accurately segment tumor region, which can lead to improved efficiency in target definition in SBRT and SRS treatment planning for gliomas.

Download ePoster [PDF]

Funding Support, Disclosures, and Conflict of Interest: The work is partially supported by GE Healthcare. The funding body did not play any roles in designing this study, in collection, analysis, and interpretation of data and in writing the manuscript.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email