In air and in vivo measurement of the leaf open time in tomotherapy using the on‐board detector pulse‐by‐pulse data

Abstract

Purpose We developed an algorithm to measure the leaf open times (LOT) from the on‐board detector (OBD) pulse‐by‐pulse data in tomotherapy. We assessed the feasibility of measuring the LOTs in dynamic jaw mode and validated the algorithm on machine QA and clinical data. Knowledge of the actual LOTs is a basis toward calculating the delivered dose and performing efficient phantom‐less delivery quality assurance (DQA) controls of the multileaf collimator (MLC). In tomotherapy, the quality of the delivered dose depends on the correct performance of the MLC, hence on the accuracy of the LOTs. Materials and methods In the detector signal, the period of time during which a leaf is open corresponds to a high intensity region. The algorithm described here locally normalizes the detector signal and measures the FWHM of the high intensity regions. The Daily QA module of the TomoTherapy Quality Assurance (TQA) tool measures LOT errors. The Daily QA detector data were collected during 9 days on two tomotherapy units. The errors yielded by the method were compared to these reported by the Daily QA module. In addition, clinical data were acquired on the two units (25 plans in total), in air without attenuation material in the beam path and in vivo during a treatment fraction. The study included plans with fields of all existing sizes (1.05, 2.51, 5.05 cm). The collimator jaws were in dynamic mode (TomoEDGETM). The feasibility of measuring the LOTs was assessed with respect to the jaw aperture. Results The mean discrepancy between LOTs measured by the algorithm and those measured by TQA was of 0 ms, with a standard deviation of 0.3 ms. The LOT measured by the method had thus an uncertainty of 1 ms with a confidence level of 99%. In 5.05 cm dynamic jaw procedures, the detector is in the beam umbra at the beginning and at the end of the delivery. In such procedures, the algorithm could not measure the LOTs at jaw apertures between 7 and maximum 12.4 mm. Otherwise, no measurement error due to the jaw movement was observed. No LOT measurement difference between air and in vivo data was observed either. Conclusion The method we propose is reliable. It can equivalently measure the LOTs from data acquired in air or in vivo. It handles fully the static procedures and the 2.51 cm dynamic procedures. It handles partially the 5.05 cm dynamic procedures. The limitation was evaluated with respect to the jaw aperture.

A delivery quality assurance tool based on the actual leaf open times in tomotherapy

“…Again, these results give us a good confidence in the LOT measurement algorithm (see also Ref. [28]). The LOT differences standard deviations (Fig.…”

“…This was extensively investigated in a previous publication. 28 We previously reported that the LOT measurement algorithm works as well in air as in vivo. 28 Thus, the method proposed here would also be valid in vivo.…”

“…In a previous work, we presented an algorithm to measure the leaf open times from the raw detector data. 28 The algorithm performs a Richardson-Lucy deconvolution of the signal. This allows us to get a neat signal per leaf, without interleaf contamination.…”

“…We chose Gaussian distributions to approximate the bellshaped LOT discrepancies observed in the previous work. 28 No plan was generated with a mean difference of 0%. We modified only the LOTs, that is, a leaf that does not open in the reference plan does not open in any of the modified plans either.…”

“…27 In a previous article, we presented an algorithm to measure the leaf open times (LOTs) from the raw detector data. 28 That previous article concentrated only on the technical aspects of measuring the LOT from the detector data. Here we propose a DQA method for tomotherapy based on the measured LOTs to address the related clinical aspects.…”