Robust Volume Calculations of Tumors of Various Sizes
Adele P. Peskin, Karen Kafadar, A.M. Santos, Gillian Haemer
Many advances in medicine today require the accurate reading of computerized tomographic (CT) images of the body. Tumors in the lung, for example, are classified according to their detected growth, i.e. change in volume, over a period of time. CT data are collected as sets of three-dimensional grid points. Tumors are often so small that a large proportion of the pixels that represent the tumors lie near the tumor surfaces. If an edge of a tumor lies between two pixel locations, radiologists must determine which of those pixels should be included in a measurement of the tumor size, determinations which can have large effects on estimated tumor volumes. Current techniques to measure these partial volumes, or 3-D voxels in the grid that are only partially filled, in this case by a tumor in a scan of the lung, vary widely in resulting tumor volume measurements. We present a statistical method that leads to accurate estimates of these volumes.
2009 World Congress in Computer Science Computer Engineering and Applied Computing
, Kafadar, K.
, Santos, A.
and Haemer, G.
Robust Volume Calculations of Tumors of Various Sizes, 2009 World Congress in Computer Science Computer Engineering and Applied Computing, Las Vegas, NV, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=901891
(Accessed June 6, 2023)