Prostate Gland Segmentation using Semantic Segmentation Models U-Net and LinkNet

Prostate Gland Segmentation using Semantic Segmentation Models U-Net and LinkNet
  IJETT-book-cover           
  
© 2022 by IJETT Journal
Volume-70 Issue-12
Year of Publication : 2022
Author : M. N. Rajesh, B. S. Chandrasekar
DOI : 10.14445/22315381/IJETT-V70I12P224

How to Cite?

M. N. Rajesh, B. S. Chandrasekar, "Prostate Gland Segmentation using Semantic Segmentation Models U-Net and LinkNet," International Journal of Engineering Trends and Technology, vol. 70, no. 12, pp. 252-271, 2022. Crossref, https://doi.org/10.14445/22315381/IJETT-V70I12P224

Abstract
The segmentation and classification of the prostate lesion or malignant growth through manual observation are highly challenging. Machine learning-based semantic segmentation architecture was used to segment the diseases based on lesion appearance and characteristics automatically. Still, those models consume more energy and processing time and will lead to reduced scalability and reliability. To tackle these limitations, deep learning (DL) based semantic segmentation architecture can be implemented, which has more advantages in discriminating the features of the lesions efficiently and accurately. This paper proposes DL-based semantic segmentation models such as LinkNet, U-Net, PSPNet and FPN. These proposed segmentation models are integrated with convolutional neural network (CNN) based backbone architectures like ResNet-34 and SE-ResNet-34. Initially, the nearest neighbour interpolation technique is employed as preprocessing technique for scaling the image. Next, normalization of intensity is employed to minimize the variations in the intensity distributions of the image. Normalized image is used for processing various settings of LinkNet and U-Net architectures. Furthermore, the proposed model uses hyper-parameter optimization with optimizers such as Adam, Adamax, Stochastic Gradient Descent, RMSProp, and Nadam for both U-Net and LinkNet to minimize the complexity of the network and enhance computing efficiency. Experimental results have been evaluated using Python on Google Colab with NCI-ISBI 2013 dataset. Performance analysis of the proposed model is assessed in terms of the Intersection of the Union (IoU) score. The LinkNet with SE-ResNet-34 model optimized using Adamax generated the best result with 0.7454453 IoU score, following U-Net with SE-ResNet-34 optimized using Adamax generated 0.738271933 IoU.

Keywords
Deep learning, FPN, Prostate cancer, PSP-Net, LinkNet, ResNet-34, Semantic segmentation, SE-ResNet-34, U-Net.

References
[1] Narinder Singh Punn, and Sonali Agarwal, “Modality Specifics U Net Variant for Bio Medical Image Segmentations: A Survey,” Artificial Intelligence Review, vol. 55, pp. 5845–5889, 2022. Crossref, https://doi.org/10.1007/s10462-022-10152-1
[2] David Gillespie et al., “Deep Learning in Magnetic Resonances Prostate Segmentations: A Review and New Perspectives,” Image and Video Processing, Arxiv Preprint arXiv:2011.07795. Crossref, https://doi.org/10.48550/arXiv.2011.07795
[3] Haozhe Jia et al., “Atlas Registration and Ensembled Deep Convolution Neural Networks-Based Prostate Segmentations Using Magnetic Resonances Imaging,” Neurocomputing, vol. 274, pp. 1359–1368, 2018. Crossref, https://doi.org/10.1016/j.neucom.2017.09.084
[4] Geert Litjens et al., “Evaluations of Prostate Segmentations Algorithm for MRIS: the PROMISE 12 Challenges,” Medical Image Analysis, vol. 18, no. 2, pp. 359–373, 2014. Crossref, https://doi.org/10.1016/j.media.2013.12.002
[5] Qikui. Zhu et al., “Deeply Supervised CNNS for Prostate Segmentations,” 2017 International Joint Conference on Neural Networks (IJCNN), pp. 178–184, 2017. Crossref, http://dx.doi.org/10.1109/IJCNN.2017.7965852
[6] Nazia Hameed et al., “An Intelligent Computers-Aided Scheme for Classifying Multiple Skin Lesion,” Computers, vol. 8, no. 3, pp. 62, 2019. Crossref, https://doi.org/10.3390/computers8030062
[7] A.Murugan, S.A.H.Nair, and K.P.S.Kumar, “Detections of Skin Cancers Using SVM, Random Forests and KNN Classifier,” Journal of Medical Systems, vol. 43, no. 8, pp. 269, 2019. Crossref, https://doi.org/10.1007/s10916-019-1400-8
[8] R.D.Seeja, and A.Suresh, “Deep Learning-Based Skin Lesions Segmentations and Classifications of Melanoma Using Support Vector Machines (SVM),” Asia Pacific Journal of Cancer Prevention, vol. 21, no. 5, pp. 1554–61, 2019. Crossref, https://doi.org/10.31557%2FAPJCP.2019.20.5.1555
[9] S.Liao et al., “Representations Learning: A Unified Deep Learning Framework for Automatic Prostate MR Segmentations,” Medical Image Computing and Computer-Assisted Intervention: MICCAI-International Conference on Medical Image Computing and Computer-Assisted Intervention, vol. 16, pp. 254–261, 2013. Crossref, https://doi.org/10.1007/978-3-642-40763-5_32
[10] A.Krizhevsky, I.Sutskever, and G.E.Hinton, “Imagenet Classifications with Deep Convolution Neural Networks,” Communication of the ACM, vol. 61, no. 6, pp. 84-901, 2017.
[11] Achim Hekler et al., “Superior Skin Cancers Classifications by the Combinations of Human and Artificial Intelligences,” European Journal of cancer, vol. 120, pp. 114–121, 2019. Crossref, https://doi.org/10.1016/j.ejca.2019.07.019
[12] Titus Josef Brinker et al., “Skin Cancers Classifications Using Convolution Neural Network: Systematic Review,” Journal of Medical Internet Research, vol. 20, no. 10, pp. E11936, 2018. Crossref, https://doi.org/10.2196/11936
[13] S. R. Guha, and S. R. Haque, “Performance Comparison of Machine Learning-Based Classification of Skin Diseases from Skin Lesion Images,” Proceedings of International Conferences on Communication, Computing and Electronics System, vol. 637, pp. 15–25, 2020. Crossref, https://doi.org/10.1007/978-981-15-2612-1_2
[14] Lei Bi, Dagan Feng, and Jinman Kim, “Dual-Paths Adversarial Learnings for Fully Convolution Network (FCN)-Based Medical Images Segmentations,” The Visual Computer, vol. 34, pp. 1042–52, 2018. Crossref, https://doi.org/10.1007/s00371-018-1519-5
[15] Lei Bi et al., “Dermoscopic Image Segmentation Via Multistage Fully Convolutional Networks,” IEEE Transactions on Biomedical Engineering, vol. 64, no. 9, pp. 2065–2074, 2017. doi: 10.1109/TBME.2017.2712771.
[16] Behnaz Abdollahi, Naofumi Tomita, and Saeed Hassanpour, “Data Augmentations in Training Deep Learning Model for Medical Images Analyses,” Intelligent Systems Reference Library, vol. 186, pp. 167–180, 2020. Crossref, http://dx.doi.org/10.1007/978-3-030-42750-4_6
[17] Keerti Maithil, and Tasneem Bano Rehman, "Urban Remote Sensing Image Segmentation Using Dense U-Net+," SSRG International Journal of Computer Science and Engineering, vol. 9, no. 3, pp. 21-28, 2022. Crossref, https://doi.org/10.14445/23488387/IJCSEV9I3P103
[18] Z.Khan et al., “Evaluations of Deep Neural Network for Semantics Segmentations of Prostate in T2W-MRI,” Sensors, vol. 20, No. 11, pp. 1-17, 2020. Crossref, https://doi.org/10.3390/s20113183
[19] Hamid Moradi, and Amir Hossein Foruzan, “Integrations of Dynamic Multi-Atlas and Deep Learning Technique to Improve Segmentations of the Prostate in MR-Images,” International Journal of Image and Graphics, vol. 22, no.4, pp. 1-17, 2022. Crossref, https://doi.org/10.1142/S0219467822500310
[20] Justin Lo et al., “Cross Attentions Squeeze Excitations Networks (CASE-Net) for Whole-Body Fetal MRI Segmentations,” Sensor, vol. 21, no. 13, pp. 4490, 2021. Crossref, https://doi.org/10.3390/s21134490
[21] Quande Liu et al., “MS-Net: Multi-Sites Networks for Improving Prostate Segmentations with Heterogeneous MRI Data,” IEEE Transactions on Medical Imaging, vol. 39, no. 9, pp. 2712-2725, 2020. Crossref, https://doi.org/10.1109/TMI.2020.2974574
[22] PrafulHambarde et al., “Prostate Lesions Segmentations in MR Image Using Radiomic Based Deeply Supervised U-Net,” Biocyberneticsand Biomedical Engineering, vol. 40, no. 4, pp. 1421–1435, 2020. Crossref, https://doi.org/10.1016/j.bbe.2020.07.011
[23] Zhiqiang Tian et al., “Psnet: Prostate Segmentations on MRI Based on A Convolution Neural Networks,” Journal of Medical Imaging, vol. 5, no. 2, pp. 1-6, 2018. Crossref, https://doi.org/10.1117%2F1.JMI.5.2.021208
[24] LeonardoRundo et al., “USE-Net: Incorporating Squeeze-and-Excitations Block Into U-net for Prostates Zonal Segmentations of Multi-Institutional MRI Data Sets,” Neurocomputing, vol. 365, pp. 31-43, 2019. Crossref, https://doi.org/10.1016/j.neucom.2019.07.006
[25] Fred Prior et al., “The Public Cancers Radiology Imaging Collection of the Cancers Imaging Archives,” Scientific Data, vol. 4, no. 170124, pp. 1-7, 2017. Crossref, https://doi.org/10.1038/sdata.2017.124
[26] A.Voulodimos et al., “Deep Learning for Computers Vision: A Brief Review,” Computational Intelligence and Neuroscience, vol. 2018, pp. 1–13, 2018. Crossref, https://doi.org/10.1155/2018/7068349
[27] BiLei et al., “Stepwise Integrations of Deep Class-Specifics Learning for Dermoscopy Images Segmentations,” Pattern Recognition, vol. 85, pp. 78-89, 2019. Crossref, https://doi.org/10.1016/j.patcog.2018.08.001
[28] Abhishek Chaurasia, and Eugenio Culurciello, “Linknet: Exploiting Encoders Representation for Efficient Semantics Segmentations,” IEEE Visual Communication and Images Processing, pp. 1-4, 2017. Crossref, https://doi.org/10.1109/VCIP.2017.8305148
[29] Hengshuang Zhao et al., “Pyramid Scenes Parsing Networks,” IEEE Conferences on Computer Visions and Patterns Recognitions, pp. 2881-2890, 2017. Crossref, https://doi.org/10.1109/CVPR.2017.660
[30] Wenchao Zhang et al., “Global Contexts Aware RCNN for Objects Detections,” Neural Computer Application, vol. 33, pp. 11627– 11639, 2021. Crossref, https://doi.org/10.48550/arXiv.2012.02637
[31] M.A. Aghdam, A.Sharifi, and M.M.Pedram, “Combinations of rs-fMRI and sMRI Data to Discriminate Autisms Spectrums Disorder in Young Children Using Deep Belief Networks,” Journal of Digital Imaging, vol. 31, No. 6, pp. 895–903, 2018. Crossref, https://doi.org/10.1007/s10278-018-0093-8
[32] Yulian Zhu et al., “MRI-Based Prostate Cancers Detections with High Level Representations and Hierarchical Classifications,” Medical physics, vol. 44, no. 3, pp. 1028–1039, 2017. Crossref, https://doi.org/10.1002/mp.12116
[33] Jie Hu et al., “Squeeze-and-Excitations Network,” Proceedings of IEEE Conferences on Computer Visions and Patterns Recognitions, pp. 7132-7141, 2018. Crossref, https://doi.org/10.1109/CVPR.2018.00745
[34] Olaf Ronneberger et al., “U-Net: Convolution Network for Biomedical Images Segmentations,” Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, vol. 9351, pp. 234-241, 2015. Crossref, https://doi.org/10.1007/978-3-319-24574- 4_28
[35] Ö.Çiçek et al., “3D U-Net: Learning Dense Volumetric Segmentations From Sparse Annotations,” International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016, vol. 9901, pp. 424–432, 2016. Crossref, https://doi.org/10.1007/978-3-319-46723-8_49
[36] Sitanaboina S L Parvathi, and Harikiran Jonnadula, “A Hybrid Semantic Model for MRI Kidney Objects Segmentations With Stochastic Feature and Edge Detections Technique,” International Journal of Engineering Trends and Technology, vol. 70, no. 9, pp. 411-420, 2022. Crossref, https://doi.org/10.14445/22315381/IJETT-V70I9P242
[37] P. Poornima, and S. Saranya, “Review on Image Segmentations Techniques to Detect Outlier in Blood Sample,” International Journal of Engineering Trends and Technology, vol. 53, no. 2, pp. 64-73, 2017. Crossref, https://doi.org/10.14445/22315381/IJETT-V53P212
[38] C. Narmatha, and P.M. Surendra, “A Review on Prostate Cancer Detection Using Deep Learning Techniques,” A Review on Prostate Cancer Detection Using Deep Learning Techniques, vol. 1, no. 2, pp. 26-33, 2020.
[39] M. B. Sudhan et al., “Segmentation and Classification of Glaucoma Using U-Net with Deep Learning Model,” Journal of Healthcare Engineering, vol. 2022, no. 1601354, pp. 1–10, 2022. Crossref, https://doi.org/10.1155/2022/1601354
[40] S. Sridhar et al., “A Torn ACL Mapping in Knee MRI Images Using Deep Convolution Neural Network with Inception,” Journal of Healthcare Engineering, vol. 2022, no. 7872500, pp. 1-9, 2022. Crossref, https://doi.org/10.1155/2022/7872500
[41] A. K. Kumaraswamy, and Chandrasekar.M.Patil, “Automatic Prostate Segmentations of Magnetic Resonances Imaging Using Resnet,” Magnetic Resonance Materials in Physics, Biology and Medicine, vol. 35, pp. 621–630, 2022. Crossref, https://doi.org/10.1007/s10334-021-00979-0