Jan Holger Rothe, Presenter: Nothing to Disclose

Friedrich D. Knollmann MD, Abstract Co-Author: Nothing to Disclose

Lukas Lehmkuhl MD, Abstract Co-Author: Nothing to Disclose

Roland Felix MD, Abstract Co-Author: Nothing to Disclose

Tumor volume is an important parameter for clinical decision making.Aim of the study is to determine the most precise method for quantifying the volume of hepatic metastases.

Six different sizes of spherical polyamide phantoms (volume: 0,268 – 14,137cm³) with a representative density to livermetastases were evaluated in a surrounding which simulates realistically the situation of liverparenchyma. CT scans were performed with a protocoll used in clinical routine for tumor staging (GE Lightspeed 16).Four different methods were used to quantify the volume of the phantoms:1. semiautomatic, seed-pixel based method 2. manual slice segmentation, 3. threshold based method and 4. crossectional measurement following to the RECIST-Criterias.The Evaluation was made with different slice thicknesses from 0,625 to 1,25mm.Each measurement was done three times, to determine the repeatability of a method.

The mean relative error quantifying the phantoms with a slice thickness of 0,625 mm is 2,2 % using manual slice segmentation and semiautomatic method respectively, 7,9% using threshold based method and 8,7% following the RECIST-Criterias. With a slice thickness of 1,25mm relative error is 3,1% using the manual slice segmentation, 3,8% using the semiautomatic method, 5,8% using the threshold based method and 15,9% following the RECIST-CriteriasConcerning the repeatability mean deviation with a slice thickness of 0,625 mm is 2,9% using the semiautomatic method, 4,7% using manual slice segmentation, 13,4% using threshold based method and 9,4% following the RECIST-Criterias and with slice thickness of 1,25mm 4,9% using manual slice segmentation, 5,1% using the semiautomatic method,16,5% using the threshold based method and 8,3% following the RECIST-Criterias.

Seed-pixel based method and manual slice segmentation have the potential to quantify the volume of livermetastases very precisely and with a high repeatability. Threedimensional measurement leads to a higher precision and repeatability than unidimensional, crosssectional measurement following the RECIST-Criterias.

Rothe, J, Knollmann, F, Lehmkuhl, L, Felix, R, Volume Measurement of Liver Metastases with 16-slice CT: A Comparison of Different Segmentation Algorithms in a Phantom Study.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4411263.html