Daniel Pinto Dos Santos MD, Presenter: Nothing to Disclose

Katrin Wunder, Abstract Co-Author: Nothing to Disclose

Roman Kloeckner MD, Abstract Co-Author: Nothing to Disclose

Lars Bornemann MS, Abstract Co-Author: Nothing to Disclose

Christoph Düber MD, Abstract Co-Author: Nothing to Disclose

Peter Mildenberger MD, Abstract Co-Author: Stockholder, GeSIT GmbH

To evaluate the effect of different reconstruction kernel on semi-automated segmentation of liver lesions in MDCT.

60 liver lesions were measured by three independent radiologists with the semi-automated segmentation software Oncology-Prototype Software (Fraunhofer MEVIS, Siemens Healthcare, Germany) using MDCT datasets (Brilliance 64, Philips Healthcare, Netherlands, 3mm slice thickness, 2mm increment) reconstructed with standard, soft and detailed kernel (Philips B, A and D). To assure for objective measurements, only lesions with satisfactory initial segmentation were included, manual correction was not used. Effective diameter and volume were measured for each lesion. The observers set an initial seed for each lesion in the standard kernel dataset. Measurements of soft and detailed kernel datasets were performed by exactly copying the seed's position. Measurements were tested for differences using intra-class correlation coefficients and t-test.

Mean effective lesion diameter was 19.9 ± 9.7 mm using the standard kernel. Comparing the three kernels, no significant differences were found. Mean difference was 1% ± 6% comparing standard to soft kernel, 3% ± 13% for standard to detailed kernel and 2% ± 9% for soft to detailed kernel. Intra-class correlation coefficients were > 0.96 in all cases.

Semi-automated 3D-volumetric analysis allows for reliable measurements of liver lesions regardless of the kernel used for reconstruction of the MDCT dataset.

Evaluation of tumor response using semi-automated segmentation software is reliable even if different kernels are used for reconstruction of MDCT datasets.  

Pinto Dos Santos, D, Wunder, K, Kloeckner, R, Bornemann, L, Düber, C, Mildenberger, P, Effect of Kernels Used for Reconstruction of MDCT Datasets on Semiautomated Segmentation and Volumetry of Liver Lesions.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11034325.html