Biomedical subjects

Xuepeng Sun

Publications and source records attributed to Xuepeng Sun.

4 recordsLinked to original sources

PubMed2005

Investigation of using a power function as a cost function in inverse planning optimization.

The purpose of this paper is to investigate the use of a power function as a cost function in inverse planning optimization. The cost function for each structure is implemented as an exponential power function of the deviation between the resultant dose and prescribed or constrained dose. The total cost function for all structures is a summation of the cost function of every structure. When the exponents of all terms in the cost function are set to 2, the cost function becomes a classical quadratic cost function. An independent optimization module was developed and interfaced with a research treatment planning system from the University of North Carolina for dose calculation and display of results. Three clinical cases were tested for this study with various exponents set for tumor targets and sensitive structures. Treatment plans with these exponent settings were compared, using dose volume histograms. The results of our study demonstrated that using an exponent higher than 2 in the cost function for the target achieved better dose homogeneity than using an exponent of 2. An exponent higher than 2 for serial sensitive structures can effectively reduce the maximum dose. Varying the exponent from 2 to 4 resulted in the most effective changes in dose volume histograms while the change from 4 to 8 is less drastic, indicating a situation of saturation. In conclusion, using a power function with exponent greater than 2 as a cost function can effectively achieve homogeneous dose inside the target and/or minimize maximum dose to the critical structures.

Algorithms↗

PubMed2004

A new smoothing procedure to reduce delivery segments for static MLC-based IMRT planning.

In the application of pixel-based intensity-modulated radiation therapy (IMRT) using the step-and-shoot delivery method, one major difficulty is the prolonged delivery time. In this study, we present an integrated IMRT planning system that involves a simple smoothing method to reduce the complexity of the beam profiles. The system consists of three main steps: (a) an inverse planning process based on a least-square dose-based cost function; (b) smoothing of the intensity maps; (c) reoptimization of the segment weights. Step (a) obtains the best plan with the lowest cost value using a simulated annealing optimization algorithm with discrete intensity levels. Step (b) takes the intensity maps obtained from (a) and reduces the complexity of the maps by smoothing the adjacent beamlet intensities. During this process each beamlet is assigned a structure index based on anatomical information. A smoothing update is applied to average adjacent beamlets with the same index. To control the quality of the plan, a predefined clinical protocol is used as an acceptance criterion. The smoothing updates that violate the criterion are rejected. After the smoothing process, the segment weights are reoptimized in step (c) to further improve the plan quality. Three clinical cases were studied using this system: a medulloblastoma, a prostate cancer, and an oropharyngeal carcinoma. While the final plans demonstrate a degradation of the original plan quality, they still meet the plan acceptance criterion. On the other hand, the segment numbers or delivery times are reduced by 40%, 20%, and 20% for the three cases, respectively.

Algorithms↗

PubMed2004

Effects of the intensity levels and beam map resolutions on static IMRT plans.

In this study we focus on how the intensity level and multileaf collimator (MLC) resolution affect the quality of IMRT plans using the static MLC delivery technique. The planning process is based on a least-square dose-based quadratic function and uses a simulated annealing algorithm to sample the discrete variables. Three clinical cases are studied empirically: a medulloblastoma, a prostate, and an oropharyngeal carcinoma. The intensity levels used are 3, 5, 10, 20, and continuous; the map resolution varies from 0.15-1.5 cm, with the leaf width equal to the step size. The influence of these two parameters are studied by comparing the cost value and the cost of delivery time from a trade-off point of view. An "efficient frontier" is drawn by connecting the plans with the lowest cost value at any given resolutions. For each case, a practical delivery region is defined by doubling the delivery time needed at a normal setting (five levels, 1.0 cm). Within this region, the "efficient frontier" demonstrates that the plans with five intensity levels are the most efficient comparing with plans with higher levels. This is a confirmation of the conclusion from Keller-Reichenbecher et al. [Int. J. Radiat. Oncol., Biol., Phys. 45, 1315-1324 (1999)]. It indicates that to further improve the plan quality with the minimal cost of extra delivery time, the most economical way is to improve the resolution rather than using higher intensity levels.

Algorithms↗

PubMed2003

Monte Carlo studies of three-dimensional O1 and O4 phi4 theory related to Bose-Einstein condensation phase transition temperatures.

The phase transition temperature for the Bose-Einstein condensation (BEC) of weakly interacting Bose gases in three dimensions is known to be related to certain nonuniversal properties of the phase transition of three-dimensional O(2) symmetric phi(4) theory. These properties have been measured previously in Monte Carlo lattice simulations. They have also been approximated analytically, with moderate success, by large N approximations to O(N) symmetric phi(4) theory. To begin investigating the region of validity of the large N approximation in this application, the same Monte Carlo technique developed for the O(2) model [P. Arnold and G. Moore, Phys. Rev. E 64, 066113 (2001)] to O(1) and O(4) theories has been applied. The results indicate that there might exist some theoretically unanticipated systematic errors in the extrapolation of the continuum value from lattice Monte Carlo results. The final results show that the difference between simulations and next-to-leading order large N calculations does not improve significantly from N=2 to N=4. This suggests that one would need to simulate yet larger N's to see true large N scaling of the difference. Quite unexpectedly (and presumably accidentally), the Monte Carlo result for N=1 seems to give the best agreement with the large N approximation among the three cases.

Journal Article↗