RT Journal Article
SR Electronic
T1 Measurement of the Body Surface Area from a Computed Tomography of a PET/CT
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 1736
OP 1736
VO 59
IS supplement 1
A1 Decazes, Pierre
A1 Vera, Pierre
A1 Daou, Doumit
A1 Kerrou, Khaldoun
A1 Blanchet, Benoit
A1 Gardin, Isabelle
A1 Modzelewski, Romain
YR 2018
UL http://jnm.snmjournals.org/content/59/supplement_1/1736.abstract
AB 1736Objectives: The body surface area (BSA) is used to calculate the dose of chemotherapy. It is generally estimated from mathematical formulae (BSAMath) using height, weight and, sometimes, gender. But a measurement of the BSA could be preferable than an estimation by mathematical formulae which have an alarming ambiguity (Redlarski G et al., Sci Rep, 2016). Our main objective was to evaluate a 3D surface reconstruction method measuring the BSA on a whole-body computed tomography (BSACT_WB) coupled with positron-emission tomography (PET/CT) and to compare these measurements with those calculated by multiple BSAMath formulae. The second objective was to evaluate the measurements performed on a CT with a limited field of acquisition from 20 cm below the ischium to the vertex (BSACT_V20), more frequent in clinical routine when a PET/CT is performed. Methods: First, the body voxels were automatically isolated using pre-treatment CT data. The body voxels were smoothed to avoid a stair step effect with the surface reconstruction algorithm. Then, BSACT_WB measurement was based on a marching cube algorithm to create a 3D body surface from small triangles. Finally, the BSACT_WB value was obtained by adding the surface of the triangles. BSACT_V20 was measured on truncated CT from vertex to 20 cm below the ischium by a similar Methods: For BSACT_V20 to be an estimate of whole BSA, an extrapolation of the BSA outside the truncated field of acquisition was made using the mean ratio of the BSA from vertex to 20 cm below the ischium divided by whole BSA, this ratio being calculated on the whole population. BSACT_WB was evaluated on a humanoid phantom with a mould as the reference standard. BSACT_WB, BSACT_V20, and 25 BSAMathformulae values were also evaluated on CT of PET/CT from 117 patients (59 women and 58 men). The results were compared using the paired t-test, intra-class coefficient correlation (ICC), Bland-Altman plots, and an evaluation of the outliers (beyond +/- 7%). Results: On the humanoid phantom, BSACT_WB had a near null difference of 0.03% compared to the mould technique, validating the method. Concerning the paired t-test, only three BSAMath formulae and BSACT_V20 had a mean insignificantly different from the mean of BSACT_WB. All BSAMath had good agreement with BSACT_WB (minimal ICC = 0.88 [95% CI: 0.83;0.91]; maximal ICC = 0.96 [95% CI: 0.94;0.97]) with variable mean differences on the Bland-Altman plots (maximal mean difference -12.3% [95% CI: -22.2%;-2.4%], minimal mean difference -0.2% [95% CI: -7%;+6.6%]). The agreement of BSACT_V20 with BSACT_WBwas slightly better compared to the best BSAMath formulae (ICC 0.97 [95% CI: 0.96;0.98]) with the smallest mean difference (+0% [95% CI: -5%;+5%]). Only BSACT_V20 had no outlier beyond +/- 7%. The whole process lasted less than 3 minutes. Conclusions: We have validated an accurate, fast and automatic measurement of BSA from whole-body PET/CT. An accurate estimation of BSA is also possible from a CT with a limited field of acquisition, more frequently used in clinical routine than a whole-body PET/CT.