An artificial intelligence‐driven agent for real‐time head‐and‐neck IMRT plan generation using conditional generative adversarial network (cGAN)
نویسندگان
چکیده
Purpose To develop an artificial intelligence (AI) agent for fully automated rapid head-and-neck intensity-modulated radiation therapy (IMRT) plan generation without time-consuming dose-volume-based inverse planning. Methods This AI was trained via implementing a conditional generative adversarial network (cGAN) architecture. The generator, PyraNet, is novel deep learning that implements 28 classic ResNet blocks in pyramid-like concatenations. discriminator customized four-layer DenseNet. first generates multiple two-dimensional projections at nine template beam angles from patient’s three-dimensional computed tomography (CT) volume and structures. These are then stacked as four-dimensional inputs of which fluence maps the corresponding generated simultaneously. Finally, predicted automatically postprocessed by Gaussian deconvolution operations imported into commercial treatment planning system (TPS) integrity check visualization. built tested upon 231 oropharyngeal IMRT plans TPS library. 200/16/15 were assigned training/validation/testing, respectively. Only primary sequential boost regime studied. All normalized to 44 Gy prescription (2 Gy/fx). A Harr wavelet loss adopted map comparison during training PyraNet. For test cases, isodose distributions qualitatively evaluated overall dose distributions. Key dosimetric metrics compared Wilcoxon signed-rank tests with significance level 0.05. Results 15 successfully generated. Isodose gradients outside PTV comparable those plans. After coverage normalization, Dmean left parotid (DAI = 23.1 ± 2.4 Gy; DTPS 2.0 Gy), right 23.8 3.0 23.9 2.3 oral cavity 24.7 6.0 4.3 Gy) statistical significance. achieved results maximum 0.01cc brainstem 15.0 2.1 15.5 2.7 cord + 5mm 27.5 25.8 1.9 clinically relevant differences, but body Dmax 121.1 3.9 109.0 0.9 higher than results. needed ~3 s predicting plan. Conclusions With execution, developed can generate complex acceptable dosimetry quality. approach holds great potential clinical applications preplanning decision-making real-time
منابع مشابه
Image De-raining Using a Conditional Generative Adversarial Network
Severe weather conditions such as rain and snow adversely affect the visual quality of images captured under such conditions thus rendering them useless for further usage and sharing. In addition, such degraded images drastically affect performance of vision systems. Hence, it is important to solve the problem of single image de-raining/de-snowing. However, this is a difficult problem to solve ...
متن کاملConditional generative adversarial nets for convolutional face generation
We apply an extension of generative adversarial networks (GANs) [8] to a conditional setting. In the GAN framework, a “generator” network is tasked with fooling a “discriminator” network into believing that its own samples are real data. We add the capability for each network to condition on some arbitrary external data which describes the image being generated or discriminated. By varying the ...
متن کاملConditional Generative Adversarial Nets
Generative Adversarial Nets [8] were recently introduced as a novel way to train generative models. In this work we introduce the conditional version of generative adversarial nets, which can be constructed by simply feeding the data, y, we wish to condition on to both the generator and discriminator. We show that this model can generate MNIST digits conditioned on class labels. We also illustr...
متن کاملGenerative Adversarial Network based on Resnet for Conditional Image Restoration
The GANs promote an adversarive game to approximate complex and jointed example probability. The networks driven by noise generate fake examples to approximate realistic data distributions. Later the conditional GAN merges prior-conditions as input in order to transfer attribute vectors to the corresponding data. However, the CGAN is not designed to deal with the high dimension conditions since...
متن کاملAdversarial Examples Generation and Defense Based on Generative Adversarial Network
We propose a novel generative adversarial network to generate and defend adversarial examples for deep neural networks (DNN). The adversarial stability of a network D is improved by training alternatively with an additional network G. Our experiment is carried out on MNIST, and the adversarial examples are generated in an efficient way compared with wildly-used gradient based methods. After tra...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Medical Physics
سال: 2021
ISSN: ['2473-4209', '1522-8541', '0094-2405']
DOI: https://doi.org/10.1002/mp.14770