Influence of image truncation on the calculation of Water Equivalent Diameter

August 29, 2018

Presented at the European Congress of Medical Physics in 2018 (August 2018, Copenhagen, Denmark)

Influence of image truncation on the calculation of Water Equivalent Diameter



Evaluate the effect of image truncation on the calculation of Water Equivalent Diameter (WED) for chest and abdomen Computed Tomography (CT) examinations. Correction factors for WED to rectify for the missing information were investigated.


Two clinical sets of 286 chest and 222 abdomen CT examinations for which the middle axial slice was not truncated, were used. The WED of the middle slice was calculated for all cases (WEDnon-trunc). Subsequently, the canvas size of the axial slice was cropped ten times from 95% to 50% in two ways; one symmetrically around the center and one nonsymmetrically (from bottom side). The truncation percentage (TP: ratio of patient border touching image edges to total patient border) and truncated WED (WEDtrunc) were calculated. Correction factors from the literature [1] were applied to the WEDtrunc, resulting in a corrected WED (WEDcor). The initial WEDnon-trunc was compared to the truncated WEDtrunc and the corrected WEDcor and the percentage difference (D) between them was calculated for the abdomen (Dtrunc,abd;Dcor,abd) and chest images (Dtrunc,chest;Dcor,chest).


For symmetrical truncation, the application of correction factors worked well for both chest and abdomen images for TP values below 50%. For TP between 30% and 50% the difference improved from Dtrunc,abd=(-4.6±2.1)% to Dcor,abd=(2.2±3.5)% and Dtrunc,chest=(-7.2±3.4)% to Dcor,chest=(-0.7±3.9)%. On the other hand, for nonsymmetrical truncation between 30% and 50%, the difference after correction is still high Dcor,abd=(-16±12)% and Dcor,chest=(-17±9)%. For TP below 20%, the gain is small, but still significant (p<0.05) and the difference in WED is between -4.5% to 0% before correction and between 3.7% to 0.6% after correction for both symmetrical and
nonsymmetrical truncation.


For small truncation percentages, below 20%, the WED is only underestimated by a small percentage. For symmetrical truncation up to 50%, the use of correction factors reduces the error in the WED calculation, while for non-symmetrical truncation, a correction factor does not seem to be enough to provide accurate results, especially for truncation percentages above 30%.


[1] Anam C, Haryanto F, Widita R, Ari I and Dougherty G. The Size-Specific Dose Estimate (SSDE) for truncated Computed Tomography images. Radiation Protection Dosimetry 2017; 175:313-320