RADIOMICS FEATURE-BASED MACHINE LEARNING FOR PREOPERATIVE GRADING OF MENINGIOMAS: A STUDY USING AUTOML

Year 2025, Volume: 26 Issue: 2, 168 - 173, 28.04.2025

Emin Demırel , Çiğdem Özer Gökaslan

https://doi.org/10.18229/kocatepetip.1581078

Abstract

OBJECTIVE: One of the most common primary intracranial neoplasms is meningiomas. Correct preoperative classification of these tumors is crucial for appropriate management of patients and treatment decisions. In this current study, we aimed to develop a radiomic feature-based machine learning model to predict grade I and grade II patients using open source software.
MATERIAL AND METHODS: Meningioma-SEG-CLASS open source dataset was collected from 96 untreated patients who underwent surgical resection between 2010 and 2019. Radiomic features of tumors were extracted from segmentation data shared as open source. AutoGluon AutoML platform was used to develop our automated machine learning algorithms.
RESULTS: AutoGluon AutoML machine learning models developed after necessary feature selection processes showed the best performance compared to the ensemble L2 model. These results are acceptable with 0.8205 AUC and 0.8000 F1 score on the test set, indicating good generalization ability of the model.
CONCLUSIONS: This study suggests that extraction of radiomic features from various MR sequences may help grade meningiomas better than traditional radiologic tests. This facilitates noninvasive preoperative tumor prediction, enabling better surgical planning and management.

Keywords

Meningioma, Machine Learning, MRI.

RADYOMİK ÖZELLİK TABANLI MAKİNE ÖĞRENİMİ İLE MENİNGİOMALARIN PREOPERATİF DERECELENDİRİLMESİ: BİR AUTOML ÇALIŞMASI

Abstract

AMAÇ: En yaygın primer intrakranial neoplazmlardan biri menenjiomlardır. Bu tümörlerin ameliyat öncesi doğru sınıflandırılması, hastaları uygun şekilde yönetmede ve tedaviye karar vermede çok önemlidir. Bu güncel çalışmada, açık kaynaklı yazılım kullanarak grade I ve grade II hastaları tahmin etmek için radyomik özellik temelli makine öğrenme modeli geliştirmeyi amaçladık.
GEREÇ VE YÖNTEM: Meningioma -SEG-CLASS açık kaynaklı veri seti, 2010 ve 2019 yılları arasında cerrahi rezeksiyon geçiren 96 tedavi edilmemiş hastadan toplanmıştır. Segmentasyon verisi açık kaynak olarak paylaşılan tümörlerin radyomik özellikleri çıkartıldı. Otomatik makine öğrenimi algoritmalarımızı geliştirmek için AutoGluon AutoML platformu kullanıldı.
BULGULAR: Gerekli özellik seçimi işlemleri sonrasında geliştirilen AutoGluon AutoML makine öğrenme modellerinde, ansambl L2 modeli göre en iyi performans gösterdi. Bu sonuçlar, test setinde 0,8205 AUC ve 0,8000 F1 skoru ile kabul edilebilir olup, modelin iyi bir genelleme yeteneğine işaret ediyor.
SONUÇ: Bu araştırma, çeşitli MRG dizilerinden radyomik özelliklerin çıkarılması, geleneksel radyolojik testlerden daha iyi bir şekilde meningiomların derecelendirilmesine yardımcı olabilir. Bu, invaziv olmayan preoperatif tümör tahminini kolaylaştırarak daha iyi cerrahi planlama ve yönetimi sağlar.

Keywords

Menenjiom, Makine Öğrenmesi, MRG

References

• 1. Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol. 2016;131(6):803-20.
• 2. Goldbrunner R, Minniti G, Preusser M, et al. EANO guidelines for the diagnosis and treatment of meningiomas. Lancet Oncol. 2016;17(9):e383-91.
• 3. Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM. Meningioma grading: an analysis of histologic parameters. Am J Surg Pathol. 1997;21(12):1455-65.
• 4. Watts J, Box G, Galvin A, et al. Magnetic resonance imaging of meningiomas: a pictorial review. Insights Imaging. 2014;5(1):113-22.
• 5. Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441-6.
• 6. Aerts HJ, Velazquez ER, Leijenaar RT, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006.
• 7. Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563-77.
• 8. Park YW, Oh J, You SC, et al. Radiomics and machine learning may accurately predict the grade and histological subtype in meningiomas using conventional and diffusion tensor imaging. Eur Radiol. 2019;29(8):4068-76.
• 9. Coroller TP, Bi WL, Huynh E, et al. Radiographic prediction of meningioma grade by semantic and radiomic features. PLoS One. 2017;12(11):e0187908.
• 10. Vassantachart A, Cao Y, Shen Z, et al, W. Segmentation and Classification of Grade I and II Meningiomas from Magnetic Resonance Imaging: An Open Annotated Dataset (Meningioma-SEG-CLASS) (Version 1) The Cancer Imaging Archive. 2024;51(3):2334-44.
• 11. Clark K, Vendt B, Smith K, et al. The Cancer Imaging Archive (TCIA): maintaining and operating a public information repository. J Digit Imaging. 2013;26(6):1045-57.
• 12. Demircioğlu A. The effect of feature normalization methods in radiomics. Insights Imaging. 2024;15(1):2.
• 13. Peng H, Long F, Ding C. Feature selection based on mutual information criteria of max-dependency, max- relevance, and min-redundancy. IEEE Transactions on pattern analysis and machine intelligence. 2005;27(8):1226-38.
• 14. Erickson N, Mueller J, Shirkov A, et al. Autogluon-tabular: Robust and accurate automl for structured data. arXiv preprint arXiv:200306505. access date: 13.03.2020.
• 15. Oya S, Kim SH, Sade B, Lee JH. The natural history of intracranial meningiomas. J Neurosurg. 2011;114(5):1250- 6.
• 16. Unser M. Texture classification and segmentation using wavelet frames. IEEE Transactions on Image Processing. 1995;4(11):1549-60.
• 17. Feng C, Wang H, Lu N, et al. Log-transformation and its implications for data analysis. Shanghai Arch Psychiatry. 2014;26(2):105-9.