Interpreting response to TMZ therapy in murine GL261 glioblastoma by combining Radiomics, Convex-NMF and feature selection in MRI/MRSI data analysis

Machine learning (ML) methods have shown great potential for the analysis of data involved in medical decisions. However, for these methods to be incorpored in the medical pipeline, they must be made interpretable not only to the data analyst, but also to the medical expert. In this work, we have applied a combination of feature transformation, selection and classification using ML and statistical methods to differentiate between control (untreated) and Temozolomide (TMZ)-treated tumour tissue from a glioblastoma (brain tumour) murine model. As input, we have used T2 weighted magnetic resonance images (MRI) and spectroscopic imaging (MRSI). Radiomics features have been extracted from the MRI dataset, while convex Non-negative Matrix Factorization (Convex-NMF) was used to extract sources from the MRSI dataset. Exhaustive feature selection has revealed parsimonious feature subsets that facilitate the expert interpretation of results while retaining a high discriminatory ability.

[1]  M. Götz,et al.  Radiomic Profiling of Glioblastoma: Identifying an Imaging Predictor of Patient Survival with Improved Performance over Established Clinical and Radiologic Risk Models. , 2016, Radiology.

[2]  Paulo J. G. Lisboa,et al.  Convex Non-Negative Matrix Factorization for Brain Tumor Delimitation from MRSI Data , 2012, PloS one.

[3]  C Arús,et al.  MRSI‐based molecular imaging of therapy response to temozolomide in preclinical glioblastoma using source analysis , 2016, NMR in biomedicine.

[5]  Dr Markand Patel,et al.  Radiomic evaluation of treatment response in patients with glioblastoma: a pilot study , 2018, Neuro-Oncology.

[6]  Margarida Julià-Sapé,et al.  Metronomic treatment in immunocompetent preclinical GL261 glioblastoma: effects of cyclophosphamide and temozolomide , 2017, NMR in biomedicine.

[7]  Margarida Julià-Sapé,et al.  Metabolomics of Therapy Response in Preclinical Glioblastoma: A Multi-Slice MRSI-Based Volumetric Analysis for Noninvasive Assessment of Temozolomide Treatment , 2017, Metabolites.

[8]  Tsuyoshi Murata,et al.  {m , 1934, ACML.

[9]  Andre Dekker,et al.  Radiomics: the process and the challenges. , 2012, Magnetic resonance imaging.

[10]  G. Fuller,et al.  Multicenter study demonstrates radiomic features derived from magnetic resonance perfusion images identify pseudoprogression in glioblastoma , 2019, Nature Communications.

[11]  Brendan J. McCullough,et al.  Issues in Response Assessment of Brain Tumor Chemotherapy , 2018 .

[12]  John B. Shoven,et al.  I , Edinburgh Medical and Surgical Journal.

[13]  Risto A. Kauppinen,et al.  1H MRS detects polyunsaturated fatty acid accumulation during gene therapy of glioma: Implications for the in vivo detection of apoptosis , 1999, Nature Medicine.

[14]  Yuzhuo Su,et al.  Spectrum separation resolves partial‐volume effect of MRSI as demonstrated on brain tumor scans , 2008, NMR in biomedicine.

[15]  Sarah L Vowler,et al.  Analysis of heterogeneity in T2-weighted MR images can differentiate pseudoprogression from progression in glioblastoma , 2017, PloS one.

[16]  Kook Jin Ahn,et al.  Analysis of peritumoral hyperintensity on pre-operative T2-weighted MR images in glioblastoma: Additive prognostic value of Minkowski functionals , 2019, PloS one.

[17]  Arvind Rao,et al.  Radiomics in glioblastoma: current status, challenges and potential opportunities , 2016 .

[18]  Chris H. Q. Ding,et al.  Convex and Semi-Nonnegative Matrix Factorizations , 2010, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[19]  Alfredo Vellido,et al.  Societal Issues Concerning the Application of Artificial Intelligence in Medicine , 2018, Kidney Diseases.

[20]  Paulo J. G. Lisboa,et al.  Discriminant Convex Non-negative Matrix Factorization for the classification of human brain tumours , 2013, Pattern Recognit. Lett..

[21]  Lluís A. Belanche Muñoz,et al.  Feature selection algorithms: a survey and experimental evaluation , 2002, 2002 IEEE International Conference on Data Mining, 2002. Proceedings..

[22]  C Arús,et al.  Robust discrimination of glioblastomas from metastatic brain tumors on the basis of single‐voxel 1H MRS , 2012, NMR in biomedicine.

[23]  Torsten Mattfeldt,et al.  CHARACTERIZATION OF MAMMARY GLAND TISSUE USING JOINT ESTIMATORS OF MINKOWSKI FUNCTIONALS , 2011 .

[24]  G. G. Stokes "J." , 1890, The New Yale Book of Quotations.

[25]  Ying Qin,et al.  High-order radiomics features based on T2 FLAIR MRI predict multiple glioma immunohistochemical features: A more precise and personalized gliomas management , 2020, PloS one.

[26]  Margarida Julià-Sapé,et al.  Extraction of artefactual MRS patterns from a large database using non‐negative matrix factorization , 2019, NMR in biomedicine.

[27]  Tamim Niazi,et al.  Radiomics in Glioblastoma: Current Status and Challenges Facing Clinical Implementation , 2019, Front. Oncol..

[28]  Sarah E Bohndiek,et al.  Analysis of image heterogeneity using 2D Minkowski functionals detects tumor responses to treatment , 2014, Magnetic resonance in medicine.

[29]  Margarida Julià-Sapé,et al.  Anti‐tumour immune response in GL261 glioblastoma generated by Temozolomide Immune‐Enhancing Metronomic Schedule monitored with MRSI‐based nosological images , 2020, NMR in biomedicine.

[30]  Alfredo Vellido,et al.  Label noise in subtype discrimination of class C G protein-coupled receptors: A systematic approach to the analysis of classification errors , 2015, BMC Bioinformatics.

[31]  Alfredo Vellido,et al.  The importance of interpretability and visualization in machine learning for applications in medicine and health care , 2019, Neural Computing and Applications.