Bilgisayarlı Tomografi Görüntülerinden Akciğer Nodülü Tespit Arayüzünün Tasarımı ve Geliştirilmesi

Abstract

Akciğer kanseri dünyada ölüme neden olan hastalıkların başında gelmektedir. Akciğer kanserinin erken teşhisi, tedavisi kadar önemlidir. Bu nedenle akciğer tomografi görüntüsünden nodüllerin yerini ve boyutunu tespit edecek bir derin öğrenme modeli olan LinkNet mimarisinin kullanımı ile bu çalışma uygulanmıştır. Bu çalışmada, akciğer Bilgisayarlı Tomografi (BT) görüntülerinde nodül saptanan 110 görüntü ve 343 nodül modeli kullanılmıştır. Bu çalışma dahilinde herhangi bir kamuya açık veri seti kullanılmamış ve etik kurul izni dahilinde tomografi görüntüleri elde edilmiştir. Ön işleme aşamasında OTSU yöntemi ile akciğer Hounsfield Unit (HU) eşik değerine göre eşikleme yapılmış ve akciğer görüntülerine segmentasyon uygulanmıştır. Uzman radyolog tarafından işaretlenen nodüllerin ROI'lerinin (Region of Interest) XML (Extensible Markup Language) dosyaları ayıklanarak görüntülere dönüştürülmüştür. LinkNet ve U-Net mimarileri ile eğitilmiş şablon nodüller kullanılarak 16 farklı mimari için karşılaştırma ve başarı yüzdesi tabloları sunulmuştur. Model, LinkNet'i kullanarak geçerli için %69'luk birleşim (IoU) puanı ve çalışma için %94'lük bir IoU puanı elde edilmiştir. Sonuç bölümünde 274 nodül verisi, validasyon bölümünde 69 nodül verisi kullanılmıştır. Deneysel sonuçlar, radyolog tarafından gözden kaçırılabilecek nodüllerin yaptığımız CAD (Bilgisayar Destekli Tasarım) ile tespit edilebileceğini ve akciğer kanseri tanısında faydalı olacağını göstermektedir.

Keywords

• Akciğer Nodül Tespiti
• BT Görüntüsü
• Otsu Eşikleme
• U-Net
• LinkNet

Lung Nodule Detection Interface Design and Development From Computerized Tomography Images

Abstract

Lung cancer is one of the leading diseases that cause death in the world. Early diagnosis of lung cancer is as important as its treatment. Therefore, we propose the LinkNet architecture, which is a deep learning model that will detect the location and size of nodules from the lung tomography image. The study was conducted with 110 patients and 343 nodules with nodules detected in Lung Computed Tomography (CT) Images. In the study, no public dataset was used and tomography images were obtained from the hospital. In the pre-processing stage, thresholding is made according to the lung Hounsfield Unit (HU) threshold value with the Otsu method and the lung is segmented. The XML (Extensible Markup Language) files of ROIs(Region of Interest) of the nodules previously marked by the radiologist are extracted and converted into images. Using template nodules trained with LinkNet and U-Net architectures, comparison and success percentage tables for 16 different architectures were presented. Using LinkNet, the model achieved an intersection of union (IoU) score of 69% for valid and an IoU score of 94% for the train. 274 nodule data were used in the train section and 69 nodule data were used in the validation section. Experimental results show that nodules that may be overlooked by a radiologist can be detected with CAD (Computer- Aided Design) performed and will be useful in the diagnosis of lung cancer.

Keywords

• Lung Nodule Detection
• CT Image
• Otsu Thresholding
• U-Net
• LinkNet

References

1. [1] G. Aresta et al., “iW-Net: an automatic and minimalistic interactive lung nodule segmentation deep network,” Scientific Reports, vol. 1, no. 1, pp. 1-9, 2019.
2. [2] G. A. Borkan, S.G. Gerzof, A.H. Robbins, D.E. Hults, C.K. Silbert, and J.E. Silbert, “Assessment of abdominal fat content by computed tomography,” The American Journal of Clinical Nutrition, vol.36, no.1, pp.172-177, 1982.
3. [3] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A.Torre, and A. Jemal, “Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries,” CA: a cancer journal for clinicians, vol. 68, no.6, pp. 394-424, 2018.
4. [4] M. Sari and S. Vatansever, “Current trends in the incidence of non-small cell lung cancer in Turkey: lung cancer aging,” EJMI, vol.4, no.2, pp.169–172, 2020.
5. [5] H. Macmahon et al.,“Guidelines for management of incidental pulmonary nodules detected on CT images: from The Fleischner Society 2017,” Radiology, vol.284, no.1, pp. 228-243, 2017.
6. [6] M. Javaid, M. Javid, M. Z. U. Rehman, and S. I. A. Shah, “A novel approach to cad system for the detection of lung nodules in CT images,” Computer Methods and Programs in Biomedicine,vol. 135, no.2016, pp. 125-139,2016.
7. [7] I. W. Harsono,S. Liawatimena, and T. W. Cenggoro, “Lung nodule detection and classification from thorax CT-scan using retinanet with transfer learning,” Journal of King Saud University-Computer and Information Sciences,vol.34, no.3, pp.567-577, 2022.
8. [8] J. Mukherjee, M. Kar, A. Chakrabarti, and S. Das,”A soft-computing based approach towards automatic detection of pulmonary nodule,” Biocybernetics and Biomedical Engineering,vol. 40, no. 3, pp. 1036-1051, 2020.

9. [9] A. Gupta, O. Märtens,Y. Le Moullec, and T. Saar, “A tool for lung nodules analysis based on segmentation and morphological operation,” In 2015 IEEE 9th International Symposium on Intelligent Signal Processing (WISP) Proceedings, Siena, Italy, 2015, pp. 1-5.
10. [10] C. F. J. Kuo et al., “Automatic lung nodule detection system using ımage processing techniques ın computed tomography,” Biomedical Signal Processing and Control,vol. 56, no.2020, pp. 101659, 2020.
11. [11] C. Zhao, J. Han,Y. Jia, and F. Gou, “Lung nodule detection via 3D U-Net and contextual convolutional neural network,” In 2018 International Conference on Networking and Network Applications (Nana), Xi'an, China, 2018, pp.356-361.
12. [12] T. N. Raju,”The nobel chronicles. 1979: allan macleod cormack (b 1924); and sir godfrey newbold hounsfield (B 1919),” Lancet (London, England), vol. 354, no. 9190, pp. 1653,1999.
13. [13] W. Xue-guang, C. Shu-hong” An improved image segmentation algorithm based on two-dimensional Otsu method,” Information Sciences Letters, vol.1, no.3, pp. 2, 2012.
14. [14] A. Rosset, L. Spadola, and O. Ratib, “Osirix: an open-source software for navigating in multidimensional DICOM images,” Journal of Digital Imaging, vol.17, no.3, pp. 205-216, 2004.
15. [15] O. Ronneberger, P. Fischer,and T. Brox, “U-Net: convolutional networks for biomedical image segmentation,” In International Conference On Medical Image Computing and Computer-Assisted Intervention, Munich, Germany, 2015, pp.1-8.
16. [16] A. Chaurasia and E. Culurciello,” Linknet: Exploiting encoder representations for efficient semantic segmentation,” In 2017 IEEE Visual Communications and Image Processing (VCIP), St. Petersburg, FL, USA, 2017, pp.1-4.
17. [17] Y. Tan, L. H. Schwartz, and B. Zhao, ”Segmentation of lung lesions on CT scans using watershed, active contours, and Markov random field” Medical Physics, vol. 40, no.4, pp. 043502, 2013.
18. [18] C. Lassen, C. Jacobs, J. M. Kuhnigk, B.Van Ginneken, and E. M. Van Rikxoort,”Robust semi-automatic segmentation of pulmonary subsolid nodules in chest computed tomography scans,”Physics in Medicine & Biology, vol.60, no.3, pp. 1307, 2015.
19. [19] B. Wu, Z. Zhou, J. Wang, and Y. Wang,” Joint learning for pulmonary nodule segmentation, attributes and malignancy prediction,” In 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018) ,Washington, USA, 2018, pp.1109-1113.