CT-US multimodal segmentation

In collaboration with national and international research institutions and with Datamind, an Italian company expert in algorithms, we have developed an automatic segmentation method that acts both in single mode and in multi-modality [Ermacora et al., UMB 41(10), 2646-2662, 2015].

The algorithm, developed within the project co-financed by the POR FESR 2007-2013: “Advanced multimodal segmentation station in Radiotherapy” (2010-2012), was trained on co-registered CT-3DUS images of the prostate.

We developed and implemented a speed of sound aberration correction algorithm of 3DUS images co-registered with CT images and acquired using curvilinear ultrasound transducers [Fontanarosa, Pesente et al., PMB 58, 1341-1360, 2013; Fontanarosa et al., PMB 38 (5), 2665-2673, 2011].

The algorithms have been implemented in iTA Data.

We have patented a system that automatically selects the best segmentation algorithm or algorithms and parameters to be used in a given context: the algorithm uses the best imaging modality (s) available for a specific case and selects only the information that allows you to obtain the best possible contouring (EP2863360 – Multimodality Image Segmentation of Volumetric Data Sets).