3D vs. 2D MRI radiomics in skeletal Ewing sarcoma: Feature reproducibility and preliminary machine learning analysis on neoadjuvant chemotherapy response prediction

Salvatore Gitto; Valentina D A Corino; Alessio Annovazzi; Estevāo Milazzo Machado; Marco Bologna; Lorenzo Marzorati; Domenico Albano; Carmelo Messina; Francesca Serpi; Vincenzo Anelli; Virginia Ferraresi; Carmine Zoccali; Alberto Aliprandi; Antonina Parafioriti; Alessandro Luzzati; Roberto Biagini; Luca Mainardi; Luca Maria Sconfienza

doi:10.3389/fonc.2022.1016123

3D vs. 2D MRI radiomics in skeletal Ewing sarcoma: Feature reproducibility and preliminary machine learning analysis on neoadjuvant chemotherapy response prediction

Front Oncol. 2022 Dec 2:12:1016123. doi: 10.3389/fonc.2022.1016123. eCollection 2022.

Authors

Salvatore Gitto¹, Valentina D A Corino^{2

3}, Alessio Annovazzi⁴, Estevāo Milazzo Machado⁵, Marco Bologna², Lorenzo Marzorati², Domenico Albano⁶, Carmelo Messina^{1

6}, Francesca Serpi¹, Vincenzo Anelli⁷, Virginia Ferraresi⁸, Carmine Zoccali^{9

10}, Alberto Aliprandi¹¹, Antonina Parafioriti¹², Alessandro Luzzati⁶, Roberto Biagini¹⁰, Luca Mainardi², Luca Maria Sconfienza^{1

6}

Affiliations

¹ Dipartimento di Scienze Biomediche per la Salute, Università degli Studi di Milano, Milan, Italy.
² Department of Electronics, Information and Bioengineering (DEIB), Politecnico Di Milano, Milan, Italy.
³ Cardiotech Lab, Centro Cardiologico Monzino IRCCS, Milan, Italy.
⁴ Nuclear Medicine Unit, IRCCS Regina Elena National Cancer Institute, Rome, Italy.
⁵ International Medical School, Università degli Studi di Milano, Milan, Italy.
⁶ IRCCS Istituto Ortopedico Galeazzi, Milan, Italy.
⁷ Radiology and Diagnostic Imaging Unit, IRCCS Regina Elena National Cancer Institute, Rome, Italy.
⁸ Sarcomas and Rare Tumours Departmental Unit, IRCCS Regina Elena National Cancer Institute, Rome, Italy.
⁹ Department of Anatomical, Histological, Forensic and Musculoskeletal System Sciences, Sapienza University of Rome, Rome, Italy.
¹⁰ Oncological Orthopaedics Unit, IRCCS Regina Elena National Cancer Institute, Rome, Italy.
¹¹ Istituti Clinici Zucchi, Monza, Italy.
¹² Pathology Department, ASST Pini - CTO, Milan, Italy.

Abstract

Objective: The extent of response to neoadjuvant chemotherapy predicts survival in Ewing sarcoma. This study focuses on MRI radiomics of skeletal Ewing sarcoma and aims to investigate feature reproducibility and machine learning prediction of response to neoadjuvant chemotherapy.

Materials and methods: This retrospective study included thirty patients with biopsy-proven skeletal Ewing sarcoma, who were treated with neoadjuvant chemotherapy before surgery at two tertiary sarcoma centres. 7 patients were poor responders and 23 were good responders based on pathological assessment of the surgical specimen. On pre-treatment T1-weighted and T2-weighted MRI, 2D and 3D tumour segmentations were manually performed. Features were extracted from original and wavelet-transformed images. Feature reproducibility was assessed through small geometrical transformations of the regions of interest mimicking multiple manual delineations, and intraclass correlation coefficient >0.75 defined feature reproducibility. Feature selection also consisted of collinearity and significance analysis. After class balancing in the training cohort, three machine learning classifiers were trained and tested on unseen data using hold-out cross-validation.

Results: 1303 (77%) 3D and 620 (65%) 2D radiomic features were reproducible. 4 3D and 4 2D features passed feature selection. Logistic regression built upon 3D features achieved the best performance with 85% accuracy (AUC=0.9) in predicting response to neoadjuvant chemotherapy.

Conclusion: Compared to 2D approach, 3D MRI radiomics of Ewing sarcoma had superior reproducibility and higher accuracy in predicting response to neoadjuvant chemotherapy, particularly when using logistic regression classifier.

Keywords: Ewing sarcoma; artificial intelligence; machine learning; magnetic resonance imaging; radiomics; texture analysis.