Ilias Gatos, Abstract Co-Author: Nothing to Disclose

Stavros Tsantis PhD, BEng, Presenter: Nothing to Disclose

Ioannis Theotokas, Abstract Co-Author: Nothing to Disclose

Pavlos Zoumpoulis MD, PhD, Abstract Co-Author: Nothing to Disclose

Aikaterini Skouroliakou, Abstract Co-Author: Nothing to Disclose

George Kagadis PhD, Abstract Co-Author: Nothing to Disclose

Contrast-enhanced ultrasonography (CEUS) of focal liver lesions (FLLs) has been accepted in clinical practice as a reliable lesion characterization tool during all vascular phases (arterial, portal and venous). Due to breathing, lesions may move, change size and shape, or even exit from the US scan plane in successive frames. CEUS video assessment is currently performed either by means of visual inspection or by built-in software packets that provide low quality time-intensity curves with no quantitative assessment. 

Twelve video clips of 12 patients with FLLs disease have been subjected to quantification into the proposed automatic lesion detection algorithm. The lesion detection procedure employed frame by frame, involves a modified active contour algorithm with constraints derived from Continuous Wavelet Transform employing the ‘Mexican Hat’ wavelet filter. The values and positions of local minima and maxima are utilized within the active contour model minimizing the edge function and stopping the evolving curve with a minimum number of iterations. Compared against manual segmentation by an expert physician, the lesion detected had an average overlap value of 0.90 ± 0.06 for all CEUS videos included in the study. Following the lesion detection procedure time-intensity curves are computed and plotted from lesion and reference areas. In addition, several parameters are computed such as Area Under the Curve, Mean Transit Time, Perfusion Index, Rise time and Time to Peak to strengthen the evaluation procedure.

A new type of software has been proposed to quantify FLLs enhancement during CEUS. After lesion detection throughout the video sequence, the degree and phase of lesion enhancement relative to the parenchyma is computed and visualized in order to evaluate the character of the lesion, as most malignant lesions tend to be hypoechoic whereas the majority of solid benign are iso- or hyper-enhancing.

Quantified Time Intensity curves analysis of contrast enhancement in FLLs could add strength to subjective interpretation, differential diagnosis and the evaluation of tumor response on antiangiogenic treatment.

http://abstract.rsna.org/uploads/2014/14010654/14010654_55gj.jpg

Gatos, I, Tsantis, S, Theotokas, I, Zoumpoulis, P, Skouroliakou, A, Kagadis, G, Automatic Focal Liver Lesions Quantification on Contrast Enhanced Ultrasound Videos.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14010654.html