Applying pytorch toolkit to plan optimization for circular cone based robotic radiotherapy

Radiation Oncology

Table 1 Condition functions of typical objectives

Objectives	g
Max dose	\(H(D_{i} - D_{i}^{0} )\)
Min dose	\(H(D_{0} - D_{i}^{0} )\)
Uniform dose	1
Max DV	\(\left\{ {\begin{array}{*{20}c} {H(D_{i} - D_{i}^{0} )} & {{\text{head}}} \\ 0 & {\text{others }} \\ \end{array} } \right.\)
Min DV	\(\left\{ {\begin{array}{*{20}c} {H(D_{i}^{0} - D_{i} )} & {{\text{tail}}} \\ 0 & {{\text{others}}} \\ \end{array} } \right.\)

H is the Heaviside function. For dose volume (DV) objectives, head and tail refer to the voxels at the top and end position sorted by dose in descending order

ISSN: 1748-717X