In this paper we propose a pipeline to integrate breast diffusion and perfusion MRI for diagnosis, surgical planning and follow-up. Dynamic contrast enhanced (DCE) and diffusion weighted (DWI) MRI provide complementary information on the tissue structure and properties: while DCE-MRI allows the characterization of the lesion angiogenesis, DWI techniques can probe the apparent diffusion coefficient (ADC) and therefore assess the nature and cellularity of the lesions. Here we propose a two-step process for the integration of these modalities. First, dissimilarity-based clustering is performed on DCE-MRI to identify the different tumoral subregions. These are then mapped onto the DWI images following inter-modal registration. The probability density functions (PDFs) of the so-identified subregions in the ADC map are extracted and compared through non-parametric testing. Results show that subregions corresponding to different clusters hold statistically different PDFs, indicating a degree of consistency in the information obtained from the two modalities while providing a posterior validation of the registration method. This enables the efficient integration of the information brought by DCE and DWI, respectively, while taking advantage of their complementarity.
Multimodal MRI-based tissue classification in breast ductal carcinoma / C.A. Mendez, F.P. Ferrarese, P. Summers, G. Petralia, M. Bellomi, G. Menegaz. ((Intervento presentato al 9. convegno IEEE International Symposium on Biomedical Imaging-ISBI : From Nano to Macro tenutosi a Barcelona nel 2012.
Multimodal MRI-based tissue classification in breast ductal carcinoma
G. Petralia;M. Bellomi;
2012
Abstract
In this paper we propose a pipeline to integrate breast diffusion and perfusion MRI for diagnosis, surgical planning and follow-up. Dynamic contrast enhanced (DCE) and diffusion weighted (DWI) MRI provide complementary information on the tissue structure and properties: while DCE-MRI allows the characterization of the lesion angiogenesis, DWI techniques can probe the apparent diffusion coefficient (ADC) and therefore assess the nature and cellularity of the lesions. Here we propose a two-step process for the integration of these modalities. First, dissimilarity-based clustering is performed on DCE-MRI to identify the different tumoral subregions. These are then mapped onto the DWI images following inter-modal registration. The probability density functions (PDFs) of the so-identified subregions in the ADC map are extracted and compared through non-parametric testing. Results show that subregions corresponding to different clusters hold statistically different PDFs, indicating a degree of consistency in the information obtained from the two modalities while providing a posterior validation of the registration method. This enables the efficient integration of the information brought by DCE and DWI, respectively, while taking advantage of their complementarity.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.