Department of Radiation Oncology, University of Florida, Gainesville, Florida, USA.
J Appl Clin Med Phys. 2021 Oct;22(10):161-168. doi: 10.1002/acm2.13411. Epub 2021 Sep 6.
The use of the ionization chamber array ICProfiler (ICP) is limited by its relatively poor detector spatial resolution and the inherent volume averaging effect (VAE). The purpose of this work is to study the feasibility of reconstructing VAE-free continuous photon beam profiles from ICP measurements with a machine learning technique.
In- and cross-plane photon beam profiles of a 6 MV beam from an Elekta linear accelerator, ranging from 2 × 2 to 10 × 10 cm at 1.5 cm, 5 cm, and 10 cm depth, were measured with an ICP. The discrete measurements were interpolated with a Makima method to obtain continuous beam profiles. Artificial neural networks (ANNs) were trained to restore the penumbra of the beam profiles. Plane-specific (in- and cr-plane) ANNs and a combined ANN were separately trained. The performance of the ANNs was evaluated using the penumbra width difference (PWD, the difference between the penumbra widths of the reconstructed and the reference profile). The plane-specific and the combined ANNs were compared to study the feasibility of using a single ANN for both in- and cross-plane.
The profiles reconstructed with all the ANNs had excellent agreement with the reference. For in-plane, the ANNs reduced the PWD from 1.6 ± 0.7 mm at 1.5 cm depth to 0.1 ± 0.1 mm, from 1.8 ± 0.6 mm at 5.0 cm depth to 0.1 ± 0.1 mm, and from 2.4 ± 0.1 mm at 10.0 cm depth to 0.0 ± 0.0 mm; for cross-plane, the ANNs reduced the PWD from 1.2 ± 0.4 mm at 1.5 cm depth, 1.2 ± 0.3 mm at 5.0 cm depth, and 1.6 ± 0.1 mm at 10.0 cm depth, to 0.1 ± 0.1 mm.
This study demonstrated the feasibility of using simple ANNs to reconstruct VAE-free continuous photon beam profiles from discrete ICP measurements. A combined ANN can restore the penumbra of in- and cross-plane beam profiles of various fields at different depths.
电离室阵列 ICP （ICProfiler）的使用受到其相对较差的探测器空间分辨率和固有体积平均效应（VAE）的限制。本工作旨在研究使用机器学习技术从 ICP 测量中重建无 VAE 的连续光子束轮廓的可行性。
使用 Elekta 直线加速器的 6 MV 光束在 1.5cm、5cm 和 10cm 深度处从 2×2 到 10×10cm 的内-和外-平面光子束轮廓，用 ICP 进行测量。离散测量用 Makima 方法插值以获得连续的光束轮廓。训练人工神经网络（ANN）以恢复光束轮廓的半影。分别训练特定于平面的（内-和外-平面）ANN 和组合 ANN。使用半影宽度差（PWD，重建轮廓和参考轮廓的半影宽度之间的差异）评估 ANN 的性能。比较特定于平面和组合 ANN，以研究使用单个 ANN 用于内-和外-平面的可行性。
所有 ANN 重建的轮廓与参考轮廓吻合良好。对于内-平面，ANN 将 1.5cm 深度处的 PWD 从 1.6±0.7mm 降低至 0.1±0.1mm，将 5.0cm 深度处的 PWD 从 1.8±0.6mm 降低至 0.1±0.1mm，将 10.0cm 深度处的 PWD 从 2.4±0.1mm 降低至 0.0±0.0mm；对于外-平面，ANN 将 1.5cm 深度处的 PWD 从 1.2±0.4mm、5.0cm 深度处的 PWD 从 1.2±0.3mm 和 10.0cm 深度处的 PWD 从 1.6±0.1mm 降低至 0.1±0.1mm。
本研究证明了使用简单的 ANN 从离散的 ICP 测量中重建无 VAE 的连续光子束轮廓的可行性。组合 ANN 可以恢复不同深度和不同射野的内-和外-平面光束轮廓的半影。
