Domain adversarial networks and intensity-based data augmentation for male pelvic organ segmentation in cone beam CT

Ensemble learning and tensor regularization for cone‐beam computed tomography‐based pelvic organ segmentation

“…[6][7][8][9] Realizing the intrinsic limitation associated with CBCT modality, transfer learning and common domain embedding techniques have been investigated to incorporate priors from other modalities or training instances. [10][11][12][13][14][15] These approaches can be roughly categorized as (1) direct augmentation of training data with another modality: an example being a 3D UNet trained with additional CT scans as to augment CBCT scans which improved the resultant Dice similarity coefficient (DSC) to 0.874 ± 0.096 and 0.814 ± 0.055 for the bladder and rectum, respectively, from 0.796 ± 0.128 and 0.680 ± 0.117 11 ; and (2) modality translation based on generative adversarial network (GAN) techniques. CT was synthesized from CBCT using Cycle-consistent GAN (CycleGAN) to improve the virtual input image quality, and the resulting bladder and rectum DSC was 0.916 ± 0.005 and 0.872 ± 0.201, respectively.…”

“…Automatic approaches have been extensively studied for segmentation, ranging from gradient‐driven boundary detection‐based approaches to active contours, and to more recent deep learning based approaches utilizing contextual structures which have witnessed much success 6–9 . Realizing the intrinsic limitation associated with CBCT modality, transfer learning and common domain embedding techniques have been investigated to incorporate priors from other modalities or training instances 10–15 . These approaches can be roughly categorized as (1) direct augmentation of training data with another modality: an example being a 3D UNet trained with additional CT scans as to augment CBCT scans which improved the resultant Dice similarity coefficient (DSC) to $$0.874\, \pm \,0.096$$ and $$0.814 \,\pm \, 0.055$$ for the bladder and rectum, respectively, from $$0.796 \pm 0.128$$ and $$ 0.680 \pm 0.117$$ 11 ; and (2) modality translation based on generative adversarial network (GAN) techniques.…”

Automatic Pancreas Segmentation Using Double Adversarial Networks With Pyramidal Pooling Module

“…The novel generative adversarial network proposed by Goodfellow et al [10] is a powerful structure that has been applied to several tasks, yielding significant results [20][21][22][23][24][25]. Moreover, various studies have applied generative adversarial networks to organ segmentation tasks [11][12][13][14][15][16][17][18][19][26][27][28][29][30][31][32][33][34][35]. Zhang et al [11] developed a CNN-based adversarial multi-residual and multi-scale pooling MRFenhanced network for multi-organ segmentation from CT images as well as for accurate contour generation in pelvic CT images.…”

“…The generator produces samples that are similar to the corresponding ground truths while the discriminator attempts to differentiate the synthetic volumes from the original images regardless of how similar they are. Brion et al [14] adopted two strategies, namely adversarial networks and intensity-based data augmentation, to train neural networks for male pelvic organ segmentation from cone beam CT images. Wang et al [26] proposed a semisymmetric structure based on a novel multi-level adversarial feature method to maintain the segmentation performance during domain adaptation, and experiments have shown that this model achieves state-of-the-art performance in meningioma segmentation.…”

“…There is considerable scope as well as a growing demand to upgrade existing methods and further improve the segmentation performance. A new architecture proposed by Goodfellow et al [10], namely the generative adversarial network, is widely used in the field of organ segmentation [11][12][13][14][15][16][17][18][19]. A generative adversarial network consists of two competing networks, i.e., a discriminator and a generator, where the generator attempts to produce samples to deceive the discriminator while the discriminator aims to distinguish the synthetic samples from the real images regardless of how similar they are.…”