Premium
WE‐E‐AUD C‐05: Predictability of Patient Specific Prostate Margins From Real‐Time Intrafraction Motion Measurements
Author(s) -
Litzenberg D,
Balter J,
Levine L,
Hadley S
Publication year - 2008
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2962786
Subject(s) - isocenter , nuclear medicine , medicine , margin (machine learning) , fraction (chemistry) , mathematics , jackknife resampling , proton therapy , statistics , imaging phantom , radiation therapy , surgery , computer science , chemistry , organic chemistry , machine learning , estimator
Purpose: To investigate the ability to predict individualized PTV margins for prostate treatment based on limited real‐time intrafraction motion data. Method and Materials: Under IRB approved protocols, 35 patients with 3 transponders implanted in the prostate were studied. Transponders were placed at the apex, right‐ and left‐base under ultrasound guidance. Isocenter was chosen relative to the centroid of the transponders and the patients were initially positioned using the electromagnetic system. The relative position of the transponders was monitored continuously during each fraction, at 10 Hz. The probability distribution of absolute displacements from isocenter was found in each direction for the first fraction. Cumulative margins, Mn, were found using van Herk's formula ( 2.5 Σ + 0.7 σ ' ̇ ) after each fraction, including all fractions (n=N), to determine the best retrospective PTV margin for each individual patient. Metrics from the first fraction's probability distribution were tested for correlation (Pearson's r, Pr) with the final cumulative margin. These metrics included the position in each direction which was >= 50% and 95% (R50, R95) of absolute deviations from isocenter. The percentage of patients for which ( | MN − Mn | < = 1 mm ) was found after n fractions. Results: The correlation coefficients in the Left‐Right, Anterior‐Posterior, and Superior‐Inferior directions were Pr50 = (0.275, 0.422, 0.177), Pr95 = (0.435, 0.489, 0.168). The percentage of patients within 1 mm of their final margin after n=2, 3, 5, 10, 20 days was LR = (94.3, 97.1, 97.1, 100.0, 100.0)%, AP = (40.0, 57.1, 74.3, 91.4, 97.1)%, and SI = (34.3, 37.1, 65.7, 91.4)%. Conclusion: R50 and R95 from a single fraction of measured intrafraction displacements are poorly correlated with a given patient's final individualized margins. In addition, 15 to 20 fractions are required to estimate margins within 1 mm for 90% of patients. Conflict of Interest: Supported by NIHP01CA59827 and Calypso Medical.