RECIST vs. Volume Measurement in Medical CT Using Ellipsoids of Known Size
Zachary H. Levine, Bruce R. Borchardt, Nolan J. Brandenburg, Charles W. Clark, Balasubramanian Muralikrishnan, Craig M. Shakarji, Joseph J. Chen, Eliot L. Siegel
(a) Purpose: We wanted to test the extent to which two common methods of determining the sizes of tumors would perform when compared to a statistically significant number of well-characterized reference objects. The size of the objects was chosen to correspond to those used in the early detection of lung cancer. A secondary purpose was to explore the ability of a volumetric method to obtain shape information. (b) Materials and methods: 283 uniaxial ellipsoids with sizes from 4 mm to 11 mm were fabricated and measured with a coordinate measuring machine and also scanned using a medical computerized tomography machine. The ellipsoid volumes were determined by counting voxels over a threshold, and using equivalent volumes from the length given by the Response Evaluation Criteria In Solid Tumors (RECIST) criterion. (c) Results: the volumetric method yields 1 order of magnitude smaller residuals compared to the volume as determined by the coordinate measuring machine. Moreover, the elliptic ratio (the ratio of the unique diameter of the uniaxial ellipsoid to its circular diameter) is given within a few percent. The observed elliptic ratio was closer to unity in the observed images than the physical objects. Such an effect is probably do to smearing with the point spread function of the computerized tomography (CT) machine, as illustrated in a simple model. (d) Conclusions: volumetric methods may reduce the time interval required to detect the growth of cancerous lesions by at least a factor of two compared to RECIST.