Deep DoseNet: a deep neural network for accurate dosimetric transformation between different spatial resolutions
1Department of Radiation Oncology, Stanford University, Stanford, CA 94305-5847, United States of America.
None:
The purpose of this work is to introduce a novel deep learning strategy to obtain highly accurate dose plan by transforming from a dose distribution calculated using a low-cost algorithm (or algorithmic settings). 25 168 slices of dose distribution are calculated using Eclipse treatment planning system V15.6 (Varian Medical Systems, Palo Alto, CA) on ten patient CTs whose treatment sites ranging from lung, brain, abdomen and pelvis, with a grid size of 1.25 × 1.25 × 1.25 mm using both anisotropic analytical algorithm (AAA) in 5 mm resolution and Acuros XB algorithm (AXB) in 1.25 mm resolution. The AAA dose slices, and the corresponding down sampled CT slices are combined to form a tensor with a size of 2 × 64 × 64, working as the input to the deep learning-based dose calculation network (deep DoseNet), which outputs the calculated Acuros dose with a size of 256 × 256. The deep DoseNet (DDN) consists of a feature extraction component and an upscaling part. The DDN converges after ~100 epochs with a learning rate of [Formula: see text], using ADAM. We compared up sampled AAA dose and DDN output with that of AXB. For the evaluation set, the average mean-square-error decreased from 4.7 × [Formula: see text] between AAA and AXB to 7.0 × 10-5 between DDN and AXB, with an average improvement of ~12 times. The average Gamma index passing rate at 3mm3% improved from 76% between AAA and AXB to 91% between DDN and AXB. The average calculation time is less than 1 ms for a single slice on a NVIDIA DGX workstation. DDN, trained with a large amount of dosimetric data, can be employed as a general-purpose dose calculation acceleration engine across various dose calculation algorithms.
