Multiparametric radiomics methods for breast cancer tissue characterization using radiological imaging

Abstract

Background and purpose: Multiparametric radiological imaging is vital for detection, characterization, and diagnosis of many different diseases. Radiomics provide quantitative metrics from radiological imaging that may infer potential biological meaning of the underlying tissue. However, current methods are limited to regions of interest extracted from a single imaging parameter or modality, which limits the amount of information available within the data. This limitation can directly affect the integration and applicable scope of radiomics into different clinical settings, since single image radiomics are not capable of capturing the true underlying tissue characteristics in the multiparametric radiological imaging space. To that end, we developed a multiparametric imaging radiomic (mpRad) framework for extraction of first and second order radiomic features from multiparametric radiological datasets.

Methods: We developed five different radiomic techniques that extract different aspects of the inter-voxel and inter-parametric relationships within the high-dimensional multiparametric magnetic resonance imaging breast datasets. Our patient cohort consisted of 138 breast patients, where, 97 patients had malignant lesions and 41 patients had benign lesions. Sensitivity, specificity, receiver operating characteristic (ROC) and areas under the curve (AUC) analysis were performed to assess diagnostic performance of the mpRad parameters. Statistical significance was set at p < 0.05.

Results: The mpRad features successfully classified malignant from benign breast lesions with excellent sensitivity and specificity of 82.5% and 80.5%, respectively, with Area Under the receiver operating characteristic Curve (AUC) of 0.87 (0.81-0.93). mpRad provided a 9-28% increase in AUC metrics over single radiomic parameters.

Conclusions: We have introduced the mpRad framework that extends radiomic analysis from single images to multiparametric datasets for better characterization of the underlying tissue biology.

Keywords: ADC; Breast cancer; Diffusion; Entropy; Gray-level co-occurrence matrix (GLCM); Informatics; Machine learning; Magnetic resonance imaging; Multiparametric imaging; Radiomics; Texture.

Fig. 1

Illustration of the mpRad framework applied to different organs for analysis of different pathologies

Fig. 2

Illustration of the five different types of multiparametric radiomics (mpRad) framework features based on first and second order statistical analysis. Left: Construction of representative breast tissue signatures on normal and lesion tissue. Right: mpRad features defined as the radiomic tissue signature first order statistics (TSFOS), tissue signature probability matrix (TSPM), and the tissue signature co-occurrence matrix (TSCM) features evaluate the complex interactions between different tissue signatures. The tissue signature complex interaction network (TSCIN) first order statistics and tissue signature relationship matrix (TSRM) features evaluate the inter-parameter complex interactions. The straight yellow arrows indicate the lesion tissue and the curved yellow arrow show glandular tissue

Fig. 3

The radiomic feature maps (RFM) obtained from single and multiparametric radiomics (mpRad) analysis in a patient with a malignant lesion. The straight yellow arrow highlights the lesion location. The curved arrow demonstrates a benign cyst in the breast. a Multiparametric MRI parameters used for the mpRad framework. b Single radiomic gray-level co-occurrence matrix (GLCM) entropy features maps from each MRI parameter. c The mpRad RFMs tissue signature co-occurrence matrix (TSCM) and tissue signature complex interaction network (TSCIN) radiomic features. Note, the improved tissue delineation between the different tissue types using the mpRad framework

Fig. 4

The radiomic feature maps (RFM) obtained from single and multiparametric radiomics (mpRad) analysis in a patient with a benign lesion. The straight yellow arrow highlights the lesion location. a Multiparametric MRI parameters used for the mpRad framework. b Single radiomic gray-level co-occurrence matrix (GLCM) entropy features maps from each MRI parameter. c The mpRad RFMs tissue signature co-occurrence matrix (TSCM) and tissue signature complex interaction network (TSCIN) radiomic features

Fig. 5

The predictive accuracy between the single parameter based radiomics features and multiparametric radiomics (mpRad) features using receiver operating characteristic (ROC) curve analysis is demonstrated. a The AUC for IsoSVM was 0.87 and shown on the left and displayed in black. The mpRad feature ROC curves (displayed in red) produced area under the ROC curve (AUC) values that were 9–28% greater than the AUCs obtained for single parameter radiomics (ROC curves displayed in blue). b The AUC curves for the each mpRAD feature are shown in the middle. The AUC values for these features ranged from 0.78 to 0.82. c The single radiomic AUC curves for each feature are shown on the right and ranged from 0.62 to 0.75