Daniele Della Latta PhD, Abstract Co-Author: Nothing to Disclose

Chiara Orsini, Presenter: Nothing to Disclose

Ilaria Bernardeschi, Abstract Co-Author: Nothing to Disclose

Giovanna Letizia Di Girolamo BEng, Abstract Co-Author: Nothing to Disclose

Marta Patronelli, Abstract Co-Author: Nothing to Disclose

Michela Guadagni, Abstract Co-Author: Nothing to Disclose

Vincenzo Positano, Abstract Co-Author: Nothing to Disclose

Angelo Monteleone, Abstract Co-Author: Nothing to Disclose

Dante Chiappino MD, Abstract Co-Author: Nothing to Disclose

The developed method enables an accurate extraction of liver and liver lesions. Its reliability makes it suitable for the evaluation of the different types of lesions and their changes over time.

Computed Tomography is a widely used imaging technique for liver tissue and lesions analysis. The high spatial resolution of CT allows a reliable extraction of liver and liver lesions features (boundaries, size and density). The high variability of these parameters makes the segmentation task especially challenging. The gray-level values of parenchyma and lesions depends on contrast timing, scan parameters and patient conditions. A multi-stage semi-automatic hepatic parenchyma and tumor segmentation method is introduced.

The liver segmentation algorithm identifies the liver area in all analyzed subjects. Liver lesions segmentation algorithm detects 98% of the lesions diagnosed by Physician (320 lesions) and their linear size (1.5-30 mm), with a sensitivity of 0.97 and a specificity of 0.95. Lesions segmentation allows also the extraction of parameters (mean CT number, normalized contrast and CNR) in the different contrast phases. Lesion wash-out curves were estimated to distinguish various types of lesions. Wash-out curves allow to obtain an attenuation profile in the three acquisition phases, so we have a response to therapy not only in volume terms, but also in terms of lesions density variation.

Abdominal contrast CT images were acquired from 30 subjects.The liver segmentation algorithm uses information by portal phase and it is performed into two stages. In the first stage a rough segmentation of liver is obtained by a set of default functions: region growing, label region and morphological operations; the second stage refines the rough segmentation result through operator involvement, who can use a set of optional functions: fill, dilate, erode and erase function. The lesions algorithm provides a pre-processing filtering and then thresholding and morphological operations. As for hypodense lesions is used the portal phase, while for those hyperdense is used the arterial phase. The operator is able to select or erase the region labeled as lesion.

Della Latta, D, Orsini, C, Bernardeschi, I, Di Girolamo, G, Patronelli, M, Guadagni, M, Positano, V, Monteleone, A, Chiappino, D, A New Semi-automatic Algorithm for Segmentation and Characterization of Liver Tissue and Liver Lesions from Abdominal Contrast-enhanced CT Images.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12028533.html