Akciğer Histopatoloji Görüntülerinden Çıkarılan Derin Özellikleri Kullanan Makine Öğrenmesi Sınıflandırıcıları ile Akciğer Kanseri Tespiti

Yıl 2021, , 1552 - 1562, 31.12.2021

Emine Uçar

https://doi.org/10.17798/bitlisfen.983291

Öz

Kanser dünyada ve ülkemizde gözlenme sıklığı giderek artan sağlık sorunlarının başında gelmekte ve her yıl milyonlarca insan kanser nedeniyle hayatını kaybetmektedir. Histopatolojik tanı, kanser türünün teşhisinde ve tedavi stratejisinin belirlenmesinde önemli bir rol oynamaktadır. Bu çalışmada akciğer histopatoloji görüntüleri kullanılarak derin öğrenme yöntemlerine dayalı bir otomatik model geliştirilmesi amaçlanmıştır. Geliştirilen modelde öncelikle DenseNet201, MobileNetV2, VGG16, NASNetLarge, Xception, InceptionV3, VGG19, EfficientNetB7 ve ResNet152 gibi önceden eğitilmiş derin öğrenme mimarileri kullanılarak özellik çıkarımı gerçekleştirilmiş ve daha sonra Adaboost, Çok katmanlı algılayıcı, Rastgele orman ve Destek vektör makinesi gibi makine öğrenmesi yöntemleri ile sınıflandırılmıştır. Ardından sınıflandırıcılardan elde edilen değerlendirme sonuçlarına göre en iyi performansa sahip ilk üç derin öznitelik birleştirilerek makine öğrenmesi sınıflandırıcılarına girdi olarak kullanılmıştır. Deneysel sonuçlar en iyi özniteliklerin birlikte kullanılmasının sınıflandırma başarısına olumlu yönde katkı sağladığını göstermiştir. Test veri setinden elde edilen sonuçlar, önerilen hibrit yaklaşımın %97.22 ortalama sınıflandırma başarısı ile akciğer histopatoloji görüntülerinden adenokarsinom, skuamöz hücreli karsinom ve normal dokuların otomatik sınıflandırmasında etkili olduğunu göstermiştir.

Anahtar Kelimeler

Akciğer kanseri tespiti, sınıflandırma, derin öğrenme, makine öğrenmesi.

Lung Cancer Detection with Machine Learning Classifiers using Deep Features Extracted from Lung Histopathology Images

Öz

Cancer is one of the health problems with an increasing incidence in the world and in our country, and millions of people die every year due to cancer. Histopathological diagnosis plays an important role in diagnosing the type of cancer and determining the treatment strategy. In this study, it is aimed to develop an automatic model based on deep learning methods using lung histopathology images. In the developed model, firstly feature extraction was performed using pre-trained deep learning architectures such as DenseNet201, MobileNetV2, VGG16, NASNetLarge, Xception, InceptionV3, VGG19, EfficientNetB7 and ResNet152, and then classified with machine learning methods such as Adaboost, Multi-layer perceptron, Random forest and Support vector machines. Afterwards, according to the evaluation results obtained from the classifiers, the first three deep features with the best performance were combined and used as input to the machine learning classifiers. Experimental results showed that using the best features together contributes positively to the classification success. Results from the test dataset showed that the proposed hybrid approach was effective in automatic classification of adenocarcinoma, squamous cell carcinoma and benign tissues from lung histopathology images, with an average classification accuracy of 97.22%.

Anahtar Kelimeler

Lung cancer detection, classification, deep learning, machine learning

Kaynakça

• Torre LA, Siegel RL, Jemal A. Lung cancer statistics. Lung Cancer Pers Med 2016:1–19.
• Rosamaria P, Daniela P, Rosanna L, Michele M, Annamaria C, Pamela P, et al. KRAS-driven lung adenocarcinoma and B cell infiltration: novel insights for immunotherapy. Cancers (Basel) 2019;11:1145.
• Gan Z, Zou Q, Lin Y, Huang X, Huang Z, Chen Z, et al. Construction and validation of a seven-microRNA signature as a prognostic tool for lung squamous cell carcinoma. Cancer Manag Res 2019;11:5701.
• Ranschaert ER, Morozov S, Algra PR. Artificial intelligence in medical imaging: opportunities, applications and risks. Springer; 2019.
• Janowczyk A, Madabhushi A. Deep learning for digital pathology image analysis: A comprehensive tutorial with selected use cases. J Pathol Inform 2016;7.
• Abdullah DM, Ahmed NS, others. A Review of most Recent Lung Cancer Detection Techniques using Machine Learning. Int J Sci Bus 2021;5:159–73.
• Dandl E, Çakiroğlu M, Ekşi Z, Özkan M, Kurt ÖK, Canan A. Artificial neural network-based classification system for lung nodules on computed tomography scans. 2014 6th Int. Conf. soft Comput. pattern Recognit., 2014, p. 382–6.
• Chauhan D, Jaiswal V. An efficient data mining classification approach for detecting lung cancer disease. 2016 Int. Conf. Commun. Electron. Syst., 2016, p. 1–8.
• Faisal MI, Bashir S, Khan ZS, Khan FH. An evaluation of machine learning classifiers and ensembles for early stage prediction of lung cancer. 2018 3rd Int. Conf. Emerg. Trends Eng. Sci. Technol., 2018, p. 1–4.
• Nasser IM, Abu-Naser SS. Lung cancer detection using artificial neural network. Int J Eng Inf Syst 2019;3:17–23.
• Thallam C, Peruboyina A, Raju SST, Sampath N. Early Stage Lung Cancer Prediction Using Various Machine Learning Techniques. 2020 4th Int. Conf. Electron. Commun. Aerosp. Technol., 2020, p. 1285–92.
• Shen W, Zhou M, Yang F, Yang C, Tian J. Multi-scale convolutional neural networks for lung nodule classification. Int. Conf. Inf. Process. Med. imaging, 2015, p. 588–99.
• Rao P, Pereira NA, Srinivasan R. Convolutional neural networks for lung cancer screening in computed tomography (CT) scans. 2016 2nd Int. Conf. Contemp. Comput. Informatics, 2016, p. 489–93.
• Alakwaa W, Nassef M, Badr A. Lung cancer detection and classification with 3D convolutional neural network (3D-CNN). Lung Cancer 2017;8:409.
• Song Q, Zhao L, Luo X, Dou X. Using deep learning for classification of lung nodules on computed tomography images. J Healthc Eng 2017;2017.
• Shakeel PM, Burhanuddin MA, Desa MI. Lung cancer detection from CT image using improved profuse clustering and deep learning instantaneously trained neural networks. Measurement 2019;145:702–12.
• Abbas MA, Bukhari SUK, Syed A, Shah SSH. The Histopathological Diagnosis of Adenocarcinoma & Squamous Cells Carcinoma of Lungs by Artificial intelligence: A comparative study of convolutional neural networks. MedRxiv 2020.
• Borkowski AA, Bui MM, Thomas LB, Wilson CP, DeLand LA, Mastorides SM. Lung and colon cancer histopathological image dataset (lc25000). ArXiv Prepr ArXiv191212142 2019.
• Christodoulidis S, Anthimopoulos M, Ebner L, Christe A, Mougiakakou S. Multisource transfer learning with convolutional neural networks for lung pattern analysis. IEEE J Biomed Heal Informatics 2016;21:76–84.
• Tajbakhsh N, Shin JY, Gurudu SR, Hurst RT, Kendall CB, Gotway MB, et al. Convolutional neural networks for medical image analysis: Full training or fine tuning? IEEE Trans Med Imaging 2016;35:1299–312.
• Pan SJ, Yang Q. A survey on transfer learning. IEEE Trans Knowl Data Eng 2009;22:1345–59.
• Freund Y, Schapire RE. A decision-theoretic generalization of on-line learning and an application to boosting. J Comput Syst Sci 1997;55:119–39.
• Duda RO, Hart PE, others. Pattern classification. John Wiley & Sons; 2006.
• Breiman L. Random forests. Mach Learn 2001;45:5–32.
• Cortes C, Vapnik V. Support-vector networks. Mach Learn 1995;20:273–97.