Deep Learning for Diagnosis of Lung Cancer from CT Images

Abstract

Early cancer diagnosis with Deep Learning (DL) techniques has been the most emphasized subject among researchers recently. In addition, as seen in many studies, the use of DL in the field of medicine is gaining more importance today. Researchers generally use DL techniques to diagnose cancer and cancer types in the health field. In the diagnosis of lung cancer, due to the inconsistency of Computed Tomography (CT) images, experts have disagreements in making the right decision. DL decision-making mechanisms that can diagnose these and similar diseases early and accurately and provide more reliable results have become an option. Studies show that lung cancer is among the leading causes of death worldwide. Lung cancer caused an estimated 1.76 million deaths in 2019. It has been observed that as the causes of lung cancer increase, the average mortality rate increases by more than 80%. It has been remarked that the rate of cancer-related deaths decreases if the cases are diagnosed and treated early. Accurate detection of the disease plays an important role in its treatment. In this study, the 6053 lung CT data set was processed with the DL techniques. It is tried to decide whether the patient has cancer and if it is cancer, it is benign or malignant. In the lung CT dataset, after the image processing stages, feature extraction is performed, and the data obtained are used as input data in DL. In this study, two methods are proposed: VGG-16, Inception v4, MobileNet v3 are used in the first method, while the AlexNet method is used in the second method. This study differs from other commonly used techniques in that it has two stages. It was determined that the experimental results showed high performance and AlexNet gave 0.96 accuracies, MobileNet v3 0.81, VGG-16 0.84, Inception v4 0.86 accuracies. Thus, preliminary information can be obtained about whether there is cancer in the CT images of lung patients, and if it is cancer, at what stage the disease is.

Keywords

• Computed Tomography (CT)
• Lung cancer
• Deep Learning (DL)
• VGG-16
• MobileNet v3
• AlexNet
• Inception v4
• Convolutional Neural Networks (CNN)

Derin Öğrenme Metodu Kullanarak BT Görüntülerinden Akciğer Kanseri Teşhisi

Abstract

Derin Öğrenmenin (DÖ) teknikleriyle erken kanser tanısı son dönemlerde araştırmacılar arasında en çok üzerinde durulan konu olmuştur. Ayrıca pek çok araştırmada görüldüğü üzere DÖ’nün tıp alanında kullanımı günümüzde daha da önem kazanmaktadır. Araştırmacılar sağlık alanında kanser ve kanser türlerini teşhis etmede genellikle DÖ tekniklerinden yararlanmaktadır. Akciğer kanseri tanısında Bilgisayarlı Tomografi (BT) görüntülerinin net olmamasından dolayı, doğru karar vermede uzmanlar görüş ayrılıkları yaşamaktadır. Bu ve benzeri hastalıkları erken ve doğru tanılayabilen ve daha güvenilir sonuçlar verebilen DÖ karar verme mekanizmaları bir seçenek haline gelmiştir. Yapılan araştırmalara göre akciğer kanseri, dünya çapındaki ölümlerin önde gelen nedenleri arasındadır. Akciğer kanseri, sadece 2019 yılında tahmini 1,76 milyon insanın ölümüne sebep olmuştur. Akciğer kanserinin sebepleri arttıkça bu hastalıktan ölüm oranının %80'in üzerine çıktığı gözlemlenmiştir. Olgular erken tanı konup, tedavi edilirse kanser kaynaklı ölümlerin oranının azalmakta olduğu görülmüştür. Hastalığın doğru saptanması, tedavi edilmesinde önemli rol oynamaktadır. Bu çalışmada DÖ tekniği ile, 6053 akciğer tomografi veri seti üzerinde işlem yapılmıştır. Hastanın kanser olup olmadığına, kanser ise bunun iyi huylu (benign) ya da kötü huylu (malign) olduğuna karar verilmesine çalışılmaktadır. Akciğer BT veri kümesinde görüntü işleme aşamalarının ardından öznitelik çıkarımı yapılıp elde edilen veriler DÖ ’de girdi verisi olarak kullanılmaktadır. Bu çalışmada iki metot önerilmiştir: Birinci yöntemde VGG-16, Inception v4, MobileNet v3 kullanılırken ikinci yöntemde AlexNet yöntemi uygulanmaktadır. İki farklı aşamanın sebebi verinin farklı oranlarda bölünmesidir. Bu çalışma, iki aşamalı olması yönüyle yaygın kullanılan diğer tekniklerden farklıdır. Deneysel sonuçların yüksek performans gösterdiği ve AlexNet’in 0.96, MobileNet v3’ün 0.81, VGG-16 0.84, Inception v4’ün ise 0.86 doğrulukta sonuç verdiği belirlenmiştir. Böylece akciğer hastalarının BT görüntülerinde kanser olup olmadığı, kanser ise hastalığın hangi aşamada olduğu konusunda ön bilgi elde edilebilmektedir.

Keywords

• Bilgisayarlı Tomografi
• Akciğer kanseri
• Derin Öğrenme
• VGG-16
• MobileNet v3
• AlexNet
• Inception v4
• Evrişimli Sinir Ağları

References

1. [1] E. Cengil and A. Çinar, “A Deep Learning Based Approach to Lung Cancer Identification, ” 2018 Int. Conf. Artif. Intell. Data Process. IDAP 2018, 2019, doi: 10.1109/IDAP.2018.8620723.
2. [2] Paul, D., Su, R., Romain, M., Sébastien, V., Pierre, V., & Isabelle, G. (2017). Feature selection for outcome prediction in oesophageal cancer using genetic algorithm and random forest classifier. Computerized Medical Imaging and Graphics, 60, 42-49.
3. [3] Y. Xu et al., “Deep learning predicts lung cancer treatment response from serial medical imaging, ” Clin. Cancer Res., vol. 25, no. 11, pp. 3266–3275, 2019, doi: 10.1158/1078-0432.CCR-18-2495.
4. [4] M. F. Serj, B. Lavi, G. Hoff, and D. P. Valls, “A Deep Convolutional Neural Network for Lung Cancer Diagnostic, ” pp. 1–10, 2018.
5. [5] K. Munir, H. Elahi, A. Ayub, F. Frezza, and A. Rizzi, “Cancer diagnosis using deep learning: A bibliographic review, ” Cancers (Basel)., vol. 11, no. 9, pp. 1–36, 2019, doi: 10.3390/cancers11091235.
6. [6] S. K. Lakshmanaprabu, S. N. Mohanty, K. Shankar, N. Arunkumar, and G. Ramirez, “Optimal deep learning model for classification of lung cancer on CT images, ” Futur. Gener. Comput. Syst., vol. 92, pp. 374–382, 2019, doi: 10.1016/j.future.2018.10.009.
7. [7] H. Park and C. Monahan, “Genetic Deep Learning for Lung Cancer Screening, ” 2019.
8. [8] D. Learning, E. Detection, and L. Cancer, “Deep Learning – Early Detection of Lung Cancer with CNN, ” no. Lidc, pp. 2–4, 2019.

9. [9] S. Kumar, “Importance of Artificial Intelligence – Machine Learning & Deep Learning Prediction in Cancer Diagnosis using Logistic Regression, ” vol. 5, no. November, 2019.
10. [10] J. L. Causey et al., “Lung cancer screening with low-dose CT scans using a deep learning approach, ” 2019.
11. [11] L. Ebner et al., “Lung nodule detection by microdose CT versus chest radiography (standard and dual-energy subtracted), ” Am. J. Roentgenol., vol. 204, no. 4, pp. 727–735, 2015, doi: 10.2214/AJR.14.12921.
12. [12] S. S. Singh, A. Dahal, L. Shrestha, and S. D. Jois, “Genotype Driven Therapy for Non-Small Cell Lung Cancer: Resistance, Pan Inhibitors and Immunotherapy, ” Curr. Med. Chem., 2019, doi: 10.2174/0929867326666190222183219.
13. [13] M. J. Ahn et al., “136O: Osimertinib combined with durvalumab in EGFR-mutant non-small cell lung cancer: Results from the TATTON phase Ib trial, ” J. Thorac. Oncol., vol. 11, no. 4, p. S115, 2016, doi: 10.1016/S1556-0864(16)30246-5.
14. [14] A. Sharma and R. Rani, “An optimized framework for cancer classification using deep learning and genetic algorithm, ” J. Med. Imaging Heal. Informatics, 2017, doi: 10.1166/jmihi.2017.2266.
15. [15] J. Kuruvilla and K. Gunavathi, “Lung cancer classification using neural networks for CT images, ” Comput. Methods Programs Biomed., 2014, doi: 10.1016/j.cmpb.2013.10.011.
16. [16] Y. Wang, “Interactive Machine Learning with Applications in Health Informatics, ” 2018.
17. [17] Q. Z. Song, L. Zhao, X. K. Luo, and X. C. Dou, “Using Deep Learning for Classification of Lung Nodules on Computed Tomography Images, ” J. Healthc. Eng., vol. 2017, 2017, doi: 10.1155/2017/8314740.
18. [18] S. U. R. Mir, I. S. A. Ahmed, S. Arnold, and R. J. Craven, “Elevated progesterone receptor membrane component 1/sigma-2 receptor levels in lung tumors and plasma from lung cancer patients, ” Int. J. Cancer, 2012, doi: 10.1002/ijc.26432.
19. [19] M. S. Rahman, P. C. Shill, and Z. Homayra, “A New Method for Lung Nodule Detection Using Deep Neural Networks for CT Images, ” 2nd Int. Conf. Electr. Comput. Commun. Eng. ECCE 2019, pp. 1–6, 2019, doi: 10.1109/ECACE.2019.8679439.
20. [20] Z. Zhong et al., “3D fully convolutional networks for co-segmentation of tumors on PET-CT images, ” Proc. - Int. Symp. Biomed. Imaging, vol. 2018-April, no. Isbi, pp. 228–231, 2018, doi: 10.1109/ISBI.2018.8363561.
21. [21] W. Alakwaa, M. Nassef, and A. Badr, “Lung cancer detection and classification with 3D convolutional neural network (3D-CNN), ” Int. J. Biol. Biomed. Eng., vol. 11, no. November, pp. 66–73, 2017, doi: 10.14569/ijacsa.2017.080853.
22. [22] Q. Z. Song, L. Zhao, X. K. Luo, and X. C. Dou, “Using Deep Learning for Classification of Lung Nodules on Computed Tomography Images, ” J. Healthc. Eng., 2017, doi: 10.1155/2017/8314740.
23. [23] T. Pandiangan, I. Bali, and A. R. J. Silalahi, “Early lung cancer detection using artificial neural network, ” Atom Indones., 2019, doi: 10.17146/aij.2019.860.
24. [24] T. Heeneman and M. Business Analytics, “Lung nodule detection by using Deep Learning, ” no. January, 2018.
25. [25] A. M. Mesleh, “Lung cancer detection using multi-layer neural networks with independent component analysis: A comparative study of training algorithms, ” Jordan J. Biol. Sci., 2017.
26. [26] R. Dey, Z. Lu, and Y. Hong, “Diagnostic classification of lung nodules using 3D neural networks, ” Proc. - Int. Symp. Biomed. Imaging, vol. 2018-April, pp. 774–778, 2018, doi: 10.1109/ISBI.2018.8363687.
27. [27] S. Li et al., “Predicting lung nodule malignancies by combining deep convolutional neural network and handcrafted features, ” Phys. Med. Biol., 2019, doi: 10.1088/1361-6560/ab326a.
28. [28] S. Shen, S. X. Han, D. R. Aberle, A. A. Bui, and W. Hsu, “An interpretable deep hierarchical semantic convolutional neural network for lung nodule malignancy classification, ” Expert Syst. Appl., 2019, doi: 10.1016/j.eswa.2019.01.048.
29. [29] A. Shaffie et al., “A generalized deep learning-based diagnostic system for early diagnosis of various types of pulmonary nodules, ” Technol. Cancer Res. Treat., 2018, doi: 10.1177/1533033818798800.
30. [30] Deye, F., Vincent Michel, P., & Ehrmann, S. (2013). Changes in cardiac arrest patients†TM temperature management after the 2013 †œTTM― trial: Results from an international survey. Annals of Intensive, 6(1), 737-750.
31. [31] S. Engineering, “DEEP NEURAL NETWORKS FOR HUMAN MOTION ANALYSIS IN BIOMECHANICS APPLICATIONS by Deep Neural Networks for Human Motion Analysis in Biomechanics Applications By RAHIL MEHRIZI Dissertation Director : Kang Li, ” 2019.
32. [32] H. Polat and H. D. Mehr, “Classification of pulmonary CT images by using hybrid 3D-deep convolutional neural network architecture, ” Appl. Sci., vol. 9, no. 5, 2019, doi: 10.3390/app9050940.
33. [33] E. Matsuyama and D.-Y. Tsai, “Automated Classification of Lung Diseases in Computed Tomography Images Using a Wavelet Based Convolutional Neural Network, ” J. Biomed. Sci. Eng., vol. 11, no. 10, pp. 263–274, 2018, doi: 10.4236/jbise.2018.1110022.
34. [34] D. Zhang, L. Zou, X. Zhou, and F. He, “Integrating Feature Selection and Feature Extraction Methods with Deep Learning to Predict Clinical Outcome of Breast Cancer, ” IEEE Access, 2018, doi: 10.1109/ACCESS.2018.2837654.
35. [35] A. Masood et al., “Computer-Assisted Decision Support System in Pulmonary Cancer detection and stage classification on CT images, ” J. Biomed. Inform., 2018, doi: 10.1016/j.jbi.2018.01.005.
36. [36] The Cancer Imaging Archive (TCIA). 2020. TCIA Collections - The Cancer Imaging Archive (TCIA). [online] Available at: [Accessed 31 May 2020].
37. [37] P. Kaur and R. Bhatia, “A Review on Lung Cancer Detection Using PET/CT Scan, ” Int. J. Adv. Res. Comput. Sci. Softw. Eng., vol. 7, no. 5, pp. 977–981, 2017, doi: 10.23956/ijarcsse/v7i5/0120.
38. [38] K. H. Yu et al., “Predicting non-small cell lung cancer prognosis by fully automated microscopic pathology image features, ” Nat. Commun., 2016, doi: 10.1038/ncomms12474.
39. [39] C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine Learning. 2018.
40. [40] A. M. Santos, A. O. De Carvalho Filho, A. C. Silva, A. C. De Paiva, R. A. Nunes, and M. Gattass, “Automatic detection of small lung nodules in 3D CT data using Gaussian mixture models, Tsallis entropy and SVM, ” Eng. Appl. Artif. Intell., 2014, doi: 10.1016/j.engappai.2014.07.007.
41. [41] R. Helen, N. Kamaraj, K. Selvi, and V. Raja Raman, “Segmentation of pulmonary parenchyma in CT lung images based on 2D Otsu optimized by PSO, ” in 2011 International Conference on Emerging Trends in Electrical and Computer Technology, ICETECT 2011, 2011, doi: 10.1109/ICETECT.2011.5760176.
42. [42] J. Bhatt, M. Joshi, and M. Sharma, “Early detection of lung cancer from CT images: nodule segmentation and classification using deep learning, ” 2018, doi: 10.1117/12.2309530.
43. [43] E. Adetiba and O. O. Olugbara, “Lung cancer prediction using neural network ensemble with histogram of oriented gradient genomic features, ” Sci. World J., vol. 2015, 2015, doi: 10.1155/2015/786013.
44. [44] Noorda, R. A., Naranjo Ornedo, V., & Pons, V. (2017). Performance of Common Clustering Methods in Segmenting Vascular Pathologies in Capsule Endoscopy Images. International Journal of Computer Assisted Radiology and Surgery, 12(1), S22-S23.
45. [45] P. Mohamed Shakeel, M. I. Desa, and M. A. Burhanuddin, “Improved watershed histogram thresholding with probabilistic neural networks for lung cancer diagnosis for CBMIR systems, ” Multimed. Tools Appl., 2019, doi: 10.1007/s11042-019-7662-9.
46. [46] Y. Xie, J. Zhang, S. Liu, W. Cai, and Y. Xia, “Lung nodule classification by jointly using visual descriptors and deep features, ” in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2017, doi: 10.1007/978-3-319-61188-4_11.
47. [47] Y. Liu, H. Wang, Y. Gu, and X. Lv, “Image classification toward lung cancer recognition by learning deep quality model, ” J. Vis. Commun. Image Represent., 2019, doi: 10.1016/j.jvcir.2019.06.012.
48. [48] S. Li et al., “Predicting lung nodule malignancies by combining deep convolutional neural network and handcrafted features, ” Phys. Med. Biol., vol. 64, no. 17, 2019, doi: 10.1088/1361-6560/ab326a.
49. [49] F. Sahito, P. Zhiwen, J. Ahmed, and R. A. Memon, “Wavelet-integrated deep networks for single image super-resolution, ” Electron., vol. 8, no. 5, pp. 1–16, 2019, doi: 10.3390/electronics8050553.
50. [50] K. H.B, T. Sarode, and S. Natu, “Performance Comparison Of Hybrid Wavelet Transforms Formed Using Dct, Walsh, Haar and DKT in Watermarking, ” Int. J. Comput. Sci. Inf. Technol., vol. 7, no. 1, pp. 41–58, 2015, doi: 10.5121/ijcsit.2015.7105.
51. [51] R. V. M. Da Nóbrega, S. A. Peixoto, S. P. P. Da Silva, and P. P. R. Filho, “Lung Nodule Classification via Deep Transfer Learning in CT Lung Images, ” in Proceedings - IEEE Symposium on Computer-Based Medical Systems, 2018, doi: 10.1109/CBMS.2018.00050.
52. [52] P. Nardelli et al., “Pulmonary Artery-Vein Classification in CT Images Using Deep Learning, ” IEEE Trans. Med. Imaging, 2018, doi: 10.1109/TMI.2018.2833385.