Meeting Banner
Abstract #3365

Identifying Constituent Tumor Tissue Subclasses in HR-MAS Spectra Using Advanced Blind Source Separation Techniques

Anca Ramona Croitor Sava1, Diana Maria Sima1, Bernardo Celda2,3, Sabine Van Huffel1

1ESAT-SCD-Biomed, Katholieke Universiteit Leuven, Heeverle, Leuven, Belgium; 2Departamento de Qumica-Fsica, Facultad de Qumica, Universitat de Valencia, Valencia, Spain; 3CIBER-BBN, ISC-III, Universitat de Valencia, Valencia, Spain


Glial tumors have proved to be very heterogeneous, both in the malignancy grade and in the tumor tissue type. We analyze the mixture of different tumor tissue types (necrotic, high cellular and border tumor tissue) within HR-MAS spectra by separating between the different sources that contribute to the profile of each spectrum. Non-negative matrix factorization and independent component analysis are used to extract the constituent source profiles and their abundance distributions within all samples. Thus each feature vector is represented as a linear combination of profiles corresponding to constituent tissue types.