txt/2205.10663.txt

Transformer based Generative Adversarial
Network for Liver Segmentation
Ugur Demir*, Zheyuan Zhang*, Bin Wang, Matthew Antalek, Elif Keles,
Debesh Jha, Amir Borhani, Daniela Ladner, and Ulas Bagci
Northwestern University, IL 60201, USA
∗[email protected]
Abstract. Automated liver segmentation from radiology scans (CT,
MRI) can improve surgery and therapy planning and follow-up assess-
ment in addition to conventional use for diagnosis and prognosis. Al-
though convolutional neural networks (CNNs) have became the stan-
dard image segmentation tasks, more recently this has started to change
towards Transformers based architectures because Transformers are tak-
ing advantage of capturing long range dependence modeling capability in
signals, so called attention mechanism. In this study, we propose a new
segmentation approach using a hybrid approach combining the Trans-
former(s) with the Generative Adversarial Network (GAN) approach.
The premise behind this choice is that the self-attention mechanism of the
Transformers allows the network to aggregate the high dimensional fea-
ture and provide global information modeling. This mechanism provides
better segmentation performance compared with traditional methods.
Furthermore, we encode this generator into the GAN based architecture
so that the discriminator network in the GAN can classify the credibility
of the generated segmentation masks compared with the real masks com-
ing from human (expert) annotations. This allows us to extract the high
dimensional topology information in the mask for biomedical image seg-
mentation and provide more reliable segmentation results. Our model
achieved a high dice coecient of 0.9433, recall of 0.9515, and preci-
sion of 0.9376 and outperformed other Transformer based approaches.
The implementation details of the proposed architecture can be found
athttps://github.com/UgurDemir/tranformer_liver_segmentation .
Keywords: Liver segmentation ·Transformer ·Generative adversarial
network
1 Introduction
Liver cancer is among the leading causes of cancer-related deaths, accounting for
8.3% of cancer mortality [14]. The high variability in shape, size, appearance,
and local orientations makes liver (and liver diseases such as tumors, brosis)
challenging to analyze from radiology scans for which the image segmentation is
∗Those authors contribute equally to this paper.arXiv:2205.10663v2  [eess.IV]  28 May 20222 Demir, Zhang et al.
often necessary [3]. An accurate organ and lesion segmentation could facilitate
reliable diagnosis and therapy planning including prognosis [5].
As a solution to biomedical image segmentation, the literature is vast and
rich. The self-attention mechanism is nowadays widely used in the biomedical
image segmentation eld where long-range dependencies and context dependent
features are essential. By capturing such information, transformer based seg-
mentation architectures (for example, SwinUNet [2]) have achieved promising
performance on many vision tasks including biomedical image segmentation [7,
15].
In parallel to the all advances in Transformers, generative methods have
achieved remarkable progresses in almost all elds of computer vision too [4].
For example, Generative Adversarial Networks (GAN) [6] is a widely used tool
for generating one target image from one source image. GAN has been applied to
the image segmentation framework to distinguish the credibility of the generated
masks like previous studies [11, 9]. The high dimensional topology information
is an important feature for pixel levell classication, thus segmentation. For ex-
ample, the segmented mask should recognize the object location, orientation,
and scale prior to delineation procedure, but most current segmentation en-
gines are likely to provide false positives outside the target region or conversely
false negatives within the target region due to an inappropriate recognition of
the target regions. By introducing the discriminator architecture (as a part of
GAN) to distinguish whether the segmentation mask is high quality or not, we
could proactively screen poor predictions from the segmentation model. Fur-
thermore, this strategy can also allow us to take advantage of many unpaired
segmentation masks which can be easily acquired or even simulated in the seg-
mentation targets. To this end, in this paper, we propose a Transformer based
GAN architecture as well as a Transformer based CycleGAN architecture for au-
tomatic liver segmentation, a very improtant clinical precursor for liver diseases.
By combining two strong algorithms, we aim to achieve both good recognition
(localization) of the target region and high quality delineations.
2 Proposed method
We rst investigated the transformer architecture to solve the liver segmenta-
tion problem from radiology scans, CT in particular due to its widespread use
and being the rst choice in most liver disease quantication. The self-attention
mechanism of the Transformers has been demonstrated to be very eective ap-
proach when nding long range dependencies as stated before. This can be quite
benecial for the liver segmentation problem especially because the object of
interest (liver) is large and pixels constituting the same object are far from each
other. We also utilized an adversarial training approach to boost the segmenta-
tion model performance. For this, we have devised a conditional image generator
in a vanilla-GAN that learns a mapping between the CT slices and the segmen-
tation maps (i.e., surrogate of the truths or reference standard). The adversarial
training forces the generator model to predict more realistic segmentation out-Transformer based Generative Adversarial Network for Liver Segmentation 3
Ground 
Truth 
Predicted 
Mask 
Real 
Fake Discriminator Transformer 
Blocks Transformer 
(Generator) 
Decoder Encoder 
Transformer 
Blocks Image to Mask 
Generator 
Decoder Encoder Transformer 
Blocks Mask to Image 
Generator 
Decoder Encoder 
Real 
Mask 
Fake 
Mask Mask 
Discriminator Real 
Image 
Fake 
Image Image 
Discriminator 
Ground 
Truth 
Input Input 
Predicted 
Mask 
Input a) Transformer-GAN 
b) Transformer-CycleGAN 
Fig. 1: Block diagram of the Transformer GAN. (a) Vanilla GAN and (b) Cycle-
GAN with Transformer generator architectures.
comes. In addition to vanilla-GAN, we have also utilized the CycleGAN [17], [13]
approach to investigate the eect of cycle consistency constraint on the segmen-
tation task. Figure 1 demonstrates the general overview of the proposed method.
2.1 Transformer based GAN
Like other GAN architectures [10], Transformer based GAN architecture is com-
posed of two related sub-architectures: the generator and the discriminator. The
generator part could generate the segmentation mask from the raw image (i.e.,
segmentation task itself), while the discriminator tries to distinguish predictions
from the human annotated ground truth. GAN provides a better way to dis-
tinguish the high-dimensional morphology information. The discriminator can
provide the similarity between the predicted masks and the ground truth (i.e.,
surrogate truth) masks. Vanilla GAN considers the whole segmentation to decide
whether it is fake or not.
2.2 Transformer based CycleGAN
One alternative extension to the standard GAN approach is to use transformer
based segmentation model within the CycleGAN setup. Unlike a standard GAN,
CycleGAN consists of two generators and two discriminator networks. While the4 Demir, Zhang et al.
rst generator accepts the raw images as input and predicts the segmentation
masks, the second generator takes the predicted segmentation maps as input
and maps them back to the input image. The rst discriminator classies the
segmentation masks as either real or fake, and the second discriminator distin-
guishes the real and the reconstructed image. Figure 1 illustrates this procedure
with liver segmentation from CT scans.
To embed transformers within the CycleGAN, we utilized the encoder-decoder
style convolutional transformer model [13]. The premise behind this idea was
that the encoder module takes the input image and decreases the spatial dimen-
sions while extracting features with convolution layers. This allowed processing
of large-scale images. The core transformer module consisted of several stacked
linear layers and self-attention blocks. The decoder part increased the spatial
dimension of the intermediate features and makes the nal prediction. For the
discriminator network, we tried three convolutional architectures. The vanilla-
GAN discriminator evaluates the input image as a whole. Alternatively, we have
adopted PatchGAN discriminator architecture [8] to focus on small mask patches
to decide the realness of each region. It splits the input masks into NxN regions
and asses their quality individually. When we set the patch size to a pixel,
PatchGAN can be considered as pixel level discriminator. W have observed that
the pixel level discriminator tends to surpass other architecture for segmenta-
tion. Figure 1 demonstrates the network overview. In all of the experiments,
the segmentation model uses the same convolutional transformer and pixel level
discriminator architectures.
3 Experimental setup
We have used Liver Tumor Segmentation Challenge (LiTS)[1] dataset. LiTS
consists of 131 CT scans. This dataset is publicly available under segmentation
challenge website and approved IRB by the challenge organizers. More informa-
tion about the dataset and challenge can be found here1.
All our models were trained on NVIDIA RTX A6000 GPU after implemented
using the PyTorch [12] framework. We have used 95 samples for training and
36 samples for testing. All models are trained on the same hyperparameters
conguration with a learning rate of 2 e4, and Adam optimizer with beta1
being 0.5 and beta2 being 0.999. All of the discriminators use the pixel level
discriminator in both GAN and CycleGAN experiments. We have used recall,
precision, and dice coecient for quantitative evaluations of the segmentation.
Further, segmentation results were qualitatively evaluated by the participating
physicians. Our algorithms are available for public use.
4 Results
We presented the evaluation results in Table 1. Our best performing method
was Transformer based GAN architecture, achieved a highest dice coecient
1https://competitions.codalab.org/competitions/17094#learn_the_detailsTransformer based Generative Adversarial Network for Liver Segmentation 5
Input Segmentation 
 Input Segmentation 
Fig. 2: Transformer based GAN liver segmentation results. Green: True positive,
Red: False Positive, Blue: False Negative.
Table 1: Performance of Transformer based methods on the LITS dataset. [1]
Method Dice coecient Precision Recall
Transformer [16, 13] 0.9432 0.9464 0.9425
Transformer - CycleGAN (ours) 0.9359 0.9539 0.9205
Transformer - GAN (ours) 0.9433 0.9376 0.9515
of 0.9433 and recall rate of 0.9515. Similarly, our transformer based CycleGAN
architecture has the highest precision, 0.9539. With Transformer based GAN, we
achieved 0.9% improvement in recall and 0.01% improvement in dice coecient
with respect to the vanilla Transformers. It is to be noted that we have used also
post-processing technique which boosts the performance for "all" the baselines
to avoid biases one from each other.6 Demir, Zhang et al.
Figure 2 shows our qualitative results for the liver segmentation. We have ex-
amined all the liver segmentation results one-by-one and no failure were identied
by the participating physicians. Hence, visual results agreed with the quantita-
tive results as described in Table 1.
5 Conclusion
In this study, we explored the use of transformer-based GAN architectures for
medical image segmentation. Specically, we used a self-attention mechanism
and designed a discriminator for classifying the credibility of generated segmen-
tation masks. Our experimental result showed that the proposed new segmenta-
tion architectures could provide accurate and reliable segmentation performance
as compared to the baseline Transfomers. Although we have shown our results
in an important clinical problem for liver diseases where image-based quanti-
cation is vital, the proposed hybrid architecture (i.e., combination of GAN and
Transformers) can potentially be applied to various medical image segmentation
tasks beyond liver CTs as the algorithms are generic, reproducible, and carries
similarities with the other segmentation tasks in biomedical imaging eld. We
anticipate that our architecture can also be applied to medical scans within the
semi-supervised learning, planned as a future work.
Acknowledgement
This study is partially supported by NIH NCI grants R01-CA246704 and R01-
R01-CA240639.
References
1. Bilic, P., Christ, P.F., Vorontsov, E., Chlebus, G., Chen, H., Dou, Q., Fu, C.W.,
Han, X., Heng, P.A., Hesser, J., et al.: The liver tumor segmentation benchmark
(lits). arXiv preprint arXiv:1901.04056 (2019)
2. Cao, H., Wang, Y., Chen, J., Jiang, D., Zhang, X., Tian, Q., Wang, M.: Swin-
unet: Unet-like pure transformer for medical image segmentation. arXiv preprint
arXiv:2105.05537 (2021)
3. Chlebus, G., Schenk, A., Moltz, J.H., van Ginneken, B., Hahn, H.K., Meine, H.:
Automatic liver tumor segmentation in ct with fully convolutional neural networks
and object-based postprocessing. Scientic reports 8(1), 1{7 (2018)
4. Chuquicusma, M.J., Hussein, S., Burt, J., Bagci, U.: How to fool radiologists with
generative adversarial networks? a visual turing test for lung cancer diagnosis. In:
2018 IEEE 15th international symposium on biomedical imaging (ISBI 2018). pp.
240{244. IEEE (2018)
5. Cornelis, F., Martin, M., Saut, O., Buy, X., Kind, M., Palussiere, J., Colin, T.:
Precision of manual two-dimensional segmentations of lung and liver metastases
and its impact on tumour response assessment using recist 1.1. European radiology
experimental 1(1), 1{7 (2017)Transformer based Generative Adversarial Network for Liver Segmentation 7
6. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S.,
Courville, A., Bengio, Y.: Generative adversarial nets. Advances in neural infor-
mation processing systems 27(2014)
7. Huang, X., Deng, Z., Li, D., Yuan, X.: Missformer: An eective medical image
segmentation transformer. arXiv preprint arXiv:2109.07162 (2021)
8. Isola, P., Zhu, J.Y., Zhou, T., Efros, A.A.: Image-to-image translation with condi-
tional adversarial networks. CVPR (2017)
9. Khosravan, N., Mortazi, A., Wallace, M., Bagci, U.: PAN: Projective Adversar-
ial Network for Medical Image Segmentation. In: Medical Image Computing and
Computer Assisted Intervention { MICCAI 2019 - 22nd International Conference,
Proceedings (2019)
10. Liu, Y., Khosravan, N., Liu, Y., Stember, J., Shoag, J., Bagci, U., Jambawalikar,
S.: Cross-modality knowledge transfer for prostate segmentation from ct scans. In:
Domain adaptation and representation transfer and medical image learning with
less labels and imperfect data, pp. 63{71. Springer (2019)
11. Luc, P., Couprie, C., Chintala, S., Verbeek, J.: Semantic Segmentation using Ad-
versarial Networks. In: NIPS Workshop on Adversarial Training. Barcelona, Spain
(Dec 2016), https://hal.inria.fr/hal-01398049
12. Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen,
T., Lin, Z., Gimelshein, N., Antiga, L., et al.: Pytorch: An imperative style, high-
performance deep learning library. Advances in neural information processing sys-
tems32(2019)
13. Ristea, N.C., Miron, A.I., Savencu, O., Georgescu, M.I., Verga, N., Khan, F.S.,
Ionescu, R.T.: Cytran: Cycle-consistent transformers for non-contrast to contrast
ct translation. arXiv preprint arXiv:2110.06400 (2021)
14. Sung, H., Ferlay, J., Siegel, R.L., Laversanne, M., Soerjomataram, I., Jemal, A.,
Bray, F.: Global cancer statistics 2020: Globocan estimates of incidence and mor-
tality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians
71(3), 209{249 (2021)
15. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser,
 L., Polosukhin, I.: Attention is all you need. Advances in neural information pro-
cessing systems 30(2017)
16. Wu, H., Xiao, B., Codella, N., Liu, M., Dai, X., Yuan, L., Zhang, L.: Cvt: In-
troducing convolutions to vision transformers. In: Proceedings of the IEEE/CVF
International Conference on Computer Vision (ICCV). pp. 22{31 (October 2021)
17. Zhu, J.Y., Park, T., Isola, P., Efros, A.A.: Unpaired image-to-image translation
using cycle-consistent adversarial networks. In: Proceedings of the IEEE interna-
tional conference on computer vision. pp. 2223{2232 (2017)