MRI texture analysis in differentiating luminal A and luminal B breast cancer molecular subtypes - a feasibility study

BMC Medical Imaging

Research Article

Kirsi Holli-Helenius¹  Pia Boström²  Nina Brück³  Ilkka Koskivuo³  Annukka Salminen⁴  Irina Rinta-Kiikka⁴  Riitta Parkkola⁵ 

[1] Department of Medical Physics, Medical Imaging Centre and Hospital Pharmacy, Pirkanmaa Hospital District, Post Box 2000, 33521, Tampere, Finland;Department of Pathology, University of Turku and Turku University Hospital, Turku, Finland;Department of Plastic and General Surgery Turku University Hospital, Turku, Finland;Department of Radiology, Tampere University Hospital, Tampere, Finland;Department of Radiology, University of Turku and Turku University Hospital, Turku, Finland;

关键词: magnetic resonance imaging (MRI);    texture analysis (TA);    breast cancer;    invasive ductal carcinoma (IDC);    volumetric analysis;    prognostic factors;   

DOI  :  10.1186/s12880-017-0239-z

 received in 2017-05-10, accepted in 2017-12-15,  发布年份 2017

来源: Springer

【 摘 要 】

BackgroundThe aim of this study was to use texture analysis (TA) of breast magnetic resonance (MR) images to assist in differentiating estrogen receptor (ER) positive breast cancer molecular subtypes.MethodsTwenty-seven patients with histopathologically proven invasive ductal breast cancer were selected in preliminary study. Tumors were classified into molecular subtypes: luminal A (ER-positive and/or progesterone receptor (PR)-positive, human epidermal growth factor receptor type 2 (HER2) -negative, proliferation marker Ki-67 < 20 and low grade (I)) and luminal B (ER-positive and/or PR-positive, HER2-positive or HER2-negative with high Ki-67 ≥ 20 and higher grade (II or III)). Co-occurrence matrix -based texture features were extracted from each tumor on T1-weighted non fat saturated pre- and postcontrast MR images using TA software MaZda. Texture parameters and tumour volumes were correlated with tumour prognostic factors.ResultsTextural differences were observed mainly in precontrast images. The two most discriminative texture parameters to differentiate luminal A and luminal B subtypes were sum entropy and sum variance (p = 0.003). The AUCs were 0.828 for sum entropy (p = 0.004), and 0.833 for sum variance (p = 0.003), and 0.878 for the model combining texture features sum entropy, sum variance (p = 0.001). In the LOOCV, the AUC for model combining features sum entropy and sum variance was 0.876. Sum entropy and sum variance showed positive correlation with higher Ki-67 index. Luminal B types were larger in volume and moderate correlation between larger tumour volume and higher Ki-67 index was also observed (r = 0.499, p = 0.008).ConclusionsTexture features which measure randomness, heterogeneity or smoothness and homogeneity may either directly or indirectly reflect underlying growth patterns of breast tumours. TA and volumetric analysis may provide a way to evaluate the biologic aggressiveness of breast tumours and provide aid in decisions regarding therapeutic efficacy.

【 授权许可】

CC BY   
© The Author(s). 2017