Non-contrast CT Liver Segmentation Using CycleGAN Data Augmentation from Contrast Enhanced CT

Song, Chongchong; He, Baochun; Chen, Hongyu; Jia, Shuangfu; Chen, Xiaoxia; Jia, Fucang

doi:10.1007/978-3-030-61166-8_13

Cited by 5 publications

(9 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reference segmentations of arteries and veins were manually performed by medical experts on our paired pathological dataset. Segmentation performances To further demonstrate the realness of the images generated by our method, similarly to [4,23,27,33], we compared the performance of a segmentation network when using either a real image and a fake image, or both real images. Given the restricted dataset, all tests were done with the Leave-One-Patient-Out (L-O-P-O) method using the 3D nnU-Net [12].…”

Section: Blood Vessel Segmentation Using Cect and Ctmentioning

confidence: 99%

“…Given the restricted dataset, all tests were done with the Leave-One-Patient-Out (L-O-P-O) method using the 3D nnU-Net [12]. Results show that replacing a real CT modality with a synthetic one produced with CycleGAN and the PBS method, as in [23,27], is not sufficient to achieve performances as good as when using both real modalities. By contrast, the synthetic CT images produced by our method achieve the highest Dice score and the lowest Hausdorff distance, with the best combination of precision and recall.…”

Section: Blood Vessel Segmentation Using Cect and Ctmentioning

confidence: 99%

“…In [23,27,33] the authors show that the combined use of ceCT and contrast-free (CT) CT images is able to deal with the heterogeneity of ceCT images and thus improves segmentation. However, in order to limit ionising radiations, clinicians often acquire only one CT modality.…”

Section: Introductionmentioning

confidence: 99%

“…In the absence of paired data sets, unsupervised translation methods, based on CycleGAN [32] and UNIT [17], have been proposed [7,14,21,31]. Some authors have also already considered applying Cycle-GAN [23,27] or UNIT [33] to artificially remove or add contrast medium on CT images. CycleGAN [32] is an evolution of Generative Adversarial Network (GAN) [10], which introduces a second neural network that tries to solve the inverse task, namely reconstructing the input.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Anatomically constrained CT image translation for heterogeneous blood vessel segmentation

Barbera¹,

Boussaid²,

Maso³

et al. 2022

Preprint

View full text Add to dashboard Cite

Anatomical structures such as blood vessels in contrast-enhanced CT (ceCT) images can be challenging to segment due to the variability in contrast medium diffusion. The combined use of ceCT and contrast-free (CT) CT images can improve the segmentation performances, but at the cost of a double radiation exposure. To limit the radiation dose, generative models could be used to synthesize one modality, instead of acquiring it. The CycleGAN approach has recently attracted particular attention because it alleviates the need for paired data that are difficult to obtain. Despite the great performances demonstrated in the literature, limitations still remain when dealing with 3D volumes generated slice by slice from unpaired datasets with different fields of view. We present an extension of CycleGAN to generate high fidelity images, with good structural consistency, in this context. We leverage anatomical constraints and automatic region of interest selection by adapting the Self-Supervised Body Regressor. These constraints enforce anatomical consistency and allow feeding anatomically-paired input images to the algorithm. Results show qualitative and quantitative improvements, compared to stateof-the-art methods, on the translation task between ceCT and CT images (and vice versa).

show abstract

Section: Blood Vessel Segmentation Using Cect and Ctmentioning

confidence: 99%

Section: Blood Vessel Segmentation Using Cect and Ctmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Anatomically constrained CT image translation for heterogeneous blood vessel segmentation

Barbera¹,

Boussaid²,

Maso³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Previously, to reduce radiation exposure or achieve other purposes in CT reconstruction, a few studies , Hu et al 2016, Koike et al 2020, Liugang et al 2020 have explored the feasibility of using deep learning to generate noncontrast CT from contrast-enhanced CT and validated their results in multiple potential clinical applications, such as organ segmentation and radiation treatment planning. However, their noncontrast reconstruction methods suffered from image distortion and structure loss in the generated noncontrast CT (Song et al 2020, Liugang et al 2020. In addition, previous studies explicitly selected almost paired noncontrast and contrast CT images to train their models, without considering the real clinical situation, in which a noncontrast CT image and a contrast-enhanced CT image cannot be 100% aligned pixel to pixel, due to involuntary physiological motion during image acquisition within a short time interval before and after injection of a contrast agent.…”

Section: Introductionmentioning

confidence: 99%

Contrast-enhanced to noncontrast CT transformation via an adjacency content-transfer-based deep subtraction residual neural network

Gu¹,

Liu²,

Zhou³

et al. 2021

Phys. Med. Biol.

View full text Add to dashboard Cite

To reduce overall patient radiation exposure in some clinical scenarios (since cancer patients need frequent follow-ups), noncontrast CT is not used in some institutions. However, although less desirable, noncontrast CT could provide additional important information. In this article, we propose a deep subtraction residual network based on adjacency content transfer to reconstruct noncontrast CT from contrast CT and maintain image quality comparable to that of a CT scan originally acquired without contrast. To address the slight structural dissimilarity of the paired CT images (noncontrast CT and contrast CT) due to involuntary physiological motion, we introduce a contrastive loss network derived from the adjacency content-transfer strategy. We evaluate the results of various similarity metrics (MSE, SSIM, NRMSE, PSNR, MAE) and the fitting curve (HU distribution) of the output mapping to estimate the reconstruction performance of the algorithm. To build the model, we randomly select a total of 15,405 CT paired images (noncontrast CT and contrast-enhanced CT) for training and 10,270 CT paired images for testing. The proposed algorithm preserves the robust structures from the contrast-enhanced CT scans and learns the noncontrast attenuation pattern from the noncontrast CT scans. During the evaluation, the deep subtraction residual network achieves higher MSE, MAE, NRMSE, and PSNR scores (by 30%) than those of the baseline models (BEGAN, CycleGAN, Pixel2Pixel) and better simulates the HU curve of noncontrast CT attenuation. After validation based on an analysis of the experimental results, we can report that the noncontrast CT images reconstructed by our proposed algorithm not only preserve the high-quality structures from the contrast-enhanced CT images, but also mimic the CT attenuation of the originally acquired noncontrast CT images.

show abstract

Deep Learning Approaches for Contrast Removal from Contrast-enhanced CT

Ganß,

De Benetti,

Brosch-Lenz

et al. 2023

Informatik Aktuell

View full text Add to dashboard Cite

Non-contrast CT Liver Segmentation Using CycleGAN Data Augmentation from Contrast Enhanced CT

Cited by 5 publications

References 9 publications

Anatomically constrained CT image translation for heterogeneous blood vessel segmentation

Anatomically constrained CT image translation for heterogeneous blood vessel segmentation

Contrast-enhanced to noncontrast CT transformation via an adjacency content-transfer-based deep subtraction residual neural network

Deep Learning Approaches for Contrast Removal from Contrast-enhanced CT

Contact Info

Product

Resources

About