Abstract:
HTML
Purpose: The response of tumors to chemotherapy and radiotherapy is commonly assessed on radiologic images. Evaluation of response is critical for determining if a particular therapeutic agent is effective against a specific tumor type. Image-based tumor measurement techniques have traditionally used the bi-dimensional measurements on CT images, however other techniques such as unidimensional measurements and volume measurements are increasing in popularity. The purpose of our study was to assess the impact of CT slice thickness on automated tumor measurements performed in the following 4 ways: uni-dimensionally, bi-dimensionally, area and volumetrically.
Methods and Materials: 42 pulmonary metastases in 10 patients were assessed on CT scan with 3 different slice thickness; 3.75 mm, 5 mm and 7.5 mm. These lesions were automatically delineated using a 3-dimentional (3-D) multi-criterion segmentation algorithm. The greatest diameter (uni-dimensional), maximal perpendicular diameters (bi-dimensional), maximal cross-sectional area and volumetric assessment for each of the pulmonary lesions were then obtained. Means and variances were calculated and differences in means and equality of variances of the 3 slice thicknesses for each of the 4 measurements were assessed statistically using Linear Mixed Models and the Levene test for equality of variance respectively.
Results: The lesions ranged from 7.3 mm to 31.9 mm in maximal diameter. Difference in means on logarithm scale for uni-dimensional, bi-dimensional, and area measurements were significantly different between slice thickness 3.75 and 5 mm and 3.75 and 7.5 mm, but not 5.0 and 7.5 mm. There was a statistically significant difference in volumetric measurement as slice thickness decreased from 7.5 to 5.0 mm and from 7.5 to 3.75 mm. The variance was slightly greater at 7.5 mm as compared with 5 mm and 3.75 mm for the uni-dimensional, bi-dimensional, and area measurements with a somewhat larger observed difference between these slice thicknesses for the volumetric measurement. None of the differences, however, were statistically different.
Conclusion: Volumetric measurements are likely overestimated at 7.5 mm slice thickness as compared with 3.75 and 5 mm. Tumor measurements obtained volumetrically change with a reduction in slice thickness from 7.5 to 5 and 3.75 mm. For uni-dimensional, bi-dimensional, and area measurements, there is no benefit decreasing from 7.5 to 5.0 mm.
Questions about this event email: zhaob@mskcc.org
Zhao DSc, B,
Automated Tumor Measurement: Impact of CT Slice Thickness on Measurement Techniques and Results. Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL.
http://archive.rsna.org/2003/3108457.html
