MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Space-Variant Deconvolution of Cerenkov Light Images of a Curved Surface

E Brost, Y Watanabe*, University of Minnesota, Minneapolis, MN

Presentations

(Sunday, 7/14/2019)  

Room: ePoster Forums

Purpose: To develop a deconvolution technique that can be applied to images of Cerenkov photon emission acquired from curved surfaces during external beam radiotherapy, in order to calculate the primary beam fluence.

Methods: An optical phantom (11x11x7 cm³) with known radiological and optical properties was used to study space-variant deconvolution techniques applied to images of Cerenkov photon emission. The front surface of the optical phantom was milled into a cylinder with a 9.8 cm radius of curvature. The phantom was mounted to a TrueBeam linac and irradiated with various open field sizes. The subsequent Cerenkov photon emission was acquired using a commercial CMOS (Canon EOS 6D) in a darkened room. Raw Cerenkov photon images were perspective distorted. A checkerboard registration technique was used to map Cerenkov images to an objective square. The resulting square images were partitioned into regions bounded by the primary radiation angle of incidence, into regions corresponding to changes of 1.5 degrees. Each region was individually deconvolved with a Cerenkov scatter function (CSF) corresponding to the angle of incidence. Deconvolved regions were pieced back together to create a space-variant deconvolved image, which was compared against the Monte Carlo (Gamos v5.0.0) calculated beam fluence and Treatment Planning System (Pinnacle3) calculated dose. CSFs were calculated via Gamos with a simulated optical phantom with matched properties.

Results: Space-variant deconvolved Cerenkov photon images better matched the expected primary beam fluence in the penumbra region. Space-variant deconvolution introduced artifacts in flat regions of the beam. This was further improved by selective deconvolution in only the penumbra region. The space-variant deconvolution result is most accurate when including a margin around the deconvolution area.

Conclusion: Selective space-variant deconvolution of only high-gradient regions in a Cerenkov photon image improve the correspondence to the primary radiation beam fluence.

Keywords

Deconvolution, Optical Dosimetry, Optical Modeling

Taxonomy

IM- Optical : General (Most aspects)

Contact Email