Cardiac MRI Z-Scores (Children) Experimental

Patient

SexThe ventricular and aortic area equations differ for boys and girls.
BSA, Du Bois m²—
BSA, Mosteller m²—

Ventricles (van der Ven 2020)

Contour as the reference did. Short-axis bSSFP; papillary muscles and major trabeculae counted as myocardium, out of the blood pool; LV mass at end-diastole, RV mass at end-systole; Du Bois BSA. A report contoured another way will not compare (how contouring moves normal values). Enter absolute values, not mL/m². No z-score below BSA 0.40 m² or under 0.6 years.

Aortic diameters (Kaiser 2008)

Ungated contrast MRA. The reference images sum every cardiac phase; each diameter is the shortest one through the vessel centre on a true cross-section; Mosteller BSA. A gated cine or end-systolic echo measurement will not compare directly. Not below BSA 0.5 m².

Aortic areas, distensibility and PWV (Voges 2012)

Gated cine at maximal distension. 3 T gradient-echo cine, cross-sectional area at maximal distension; distensibility uses the maximal and minimal areas and the arm-cuff pressure taken during the scan; referenced by age and sex. Few young children: 2 girls and 2 boys under 5 years, none under 2.3, so no z-score below 2.3 years.

Pediatric CMR z-scores: ventricles (van der Ven 2020), aortic diameters (Kaiser 2008), aortic areas, distensibility and pulse wave velocity (Voges 2012)

Model, equations and sources

Ventricles: van der Ven 2020

Published (Supplementary material 4), for Z = −2, −1, 0, +1, +2, boys and girls:
  value (mL or g) = (a × BSA + b)^p
BSA (Du Bois) = 0.007184 × height[cm]^0.725 × weight[kg]^0.425

Our inversion [derived]: piecewise-linear between adjacent printed lines in value^(1/p) space
  r_k = a_k × BSA + b_k   (the printed line for Z_k, k = −2 … +2)
  Z = Z_k + (value^(1/p) − r_k) / (r_(k+1) − r_k),   r_k ≤ value^(1/p) ≤ r_(k+1)
  Below −2 and above +2 the outermost pair (−2, −1 or +1, +2) is extended

The paper prints the lines; it does not print a z-score formula. The inversion above is ours [derived]: it returns every printed line exactly (a value on the +2 line reads z +2.00), so the flag and the number agree. Within 0.005 of ±2 the z is shown to three decimals; the flag always follows the printed line. Between lines it interpolates in value^(1/p), where the lines are straight in BSA. In all 16 tables a and b also change almost linearly with Z (within 0.007 of a least-squares line, the size of two-decimal rounding), so the interpolation bends little. Expected value and the −2 to +2 range in the results are the printed Z = 0, −2 and +2 lines. Beyond them the z-score extends the outermost pair and is flagged. Stroke volume is EDV − ESV and gets its z from the published stroke-volume equations; ejection fraction is computed and has no z (the paper gives no equation for it). Enter absolute values: the z-scores are for absolute volumes and masses against BSA, as published; indexed values (mL/m², g/m², Du Bois) are shown for reference only.

Population. 141 healthy children (73 girls, 68 boys), median age 12.7 years (0.6 to 18.5), re-analysed from three European centres; all Caucasian; 1.5 T short-axis bSSFP. Only 12 were under 6 years (8 girls; 4 boys aged 0 to 3). The paper prints no BSA range; its age and sex groups have BSA interquartile ranges from 0.40 to 1.87 m² (Table 2), and few of its children were 3 to 6 years old (our reading of Table 2 and the Limitations). Small children (the site’s rules, our choice [derived]): no z-score below Du Bois BSA 0.40 m² or under 0.6 years; from 0.40 to 0.65 m², or under 6 years, the z-scores probably read low, because the expected (Z = 0) lines sit above every median of the reference’s under-6 children in Table 3 (see Verification), so they should not be used alone to call a ventricle normal-sized or small; from 0.65 to 1.0 m² a caution. The corrigendum (doi:10.1093/ehjci/jeaa320) concerns the percentile curves in Supplementary materials 1 and 2, not these equations.

Aortic diameters: Kaiser 2008

predicted diameter (mm) = a + b × BSA^0.5       (Table 2)
z = (measured − predicted) / SD of residuals
BSA (Mosteller) = √(height[cm] × weight[kg] / 3600)

The z formula is printed in the paper. Diameters are the shortest diameter through the vessel centre on a true cross-section of a 3D contrast-enhanced MRA without ECG gating, so the image sums all cardiac phases; the authors note that echo and angiographic measurements are usually end-systolic. 53 children, 2 to 20 years, BSA 0.52 to 1.9 m²; the authors do not recommend the equations below BSA 0.5 m², so no z-score is shown there. Site names follow the paper’s Methods and Figure 1. One equation for boys and girls. The abstract prints the diaphragm level as (−1.27 + 9.89 × BSA^0.5) mm; Table 2 prints +1.27, which this page uses: at the cohort’s median BSA (1.05 m²) +1.27 gives 11.4 mm, beside the measured median of 11.7 mm (Table 1), where −1.27 would give 8.9 mm.

Aortic areas, distensibility and PWV: Voges 2012

z = ((X / M)^L − 1) / (L × S)        L, M, S for the age and sex (Tables 6 to 10)
value at z = M × (1 + L × S × z)^(1/L)
distensibility (10−3 mmHg−1) = (Amax − Amin) / (Amin × (Pmax − Pmin))

The LMS tables give L, M and S at each whole year from 0 to 30, by sex. Between whole years this page interpolates L, M and S linearly: our choice [derived]; the paper prints whole years only. 71 heart-healthy volunteers aged 2.3 to 28.3 years (2 girls and 2 boys under 5 years); 3 T gradient-echo cine with retrospective gating; areas at maximal distension; pressure by arm cuff during the scan. The rows for ages 0 to 2 lie below the youngest subject: they extend the fitted curves and are not observed values, so the page gives no z-score below 2.3 years (the site’s rule). No result above 30 years, where the tables end; ventricles and diameters are still computed.

Coefficients

van der Ven 2020, Supplementary material 4: (a × BSA + b)^p at Z = −2, −1, 0, +1, +2
MeasureSexpa at Z −2 to +2b at Z −2 to +2
Kaiser 2008, Table 2
Sitea (mm)b (mm/m)SD (mm)

Voges 2012 Tables 6 to 10 (155 rows of L, M and S) are in /cross-sectional/cmr-z/calc.js, copied from the paper and checked cell by cell by the page’s tests.

Verification

Not yet verified; no official calculator was compared. Checked so far: every coefficient against the published tables (16 van der Ven tables of five rows, Kaiser Table 2, Voges Tables 6 to 10), that every printed van der Ven row returns its exact z, and simulated cases computed directly from the published lines and tables. Against the van der Ven Table 3 medians at each group’s median BSA (Table 2), the Z = 0 line is within 10% in 45 of 48 comparisons for the all-ages, 6–12 and 12–18 year groups, and every one of those medians falls within ±1 z (medians are not means, so this is agreement, not a test). Under 6 years the expected (Z = 0) lines sit above the medians. Table 2’s under-6 BSA cell looks swapped between girls (printed 0.43) and boys (0.55): Du Bois from each group’s own median height and weight gives 0.54 and 0.43; Table 3 divided by Table 4 implies about 0.50 and 0.46; and the printed interquartile ranges match the other sex’s height and weight (our reading; the paper does not say). With the cell swapped back, the Z = 0 line is 4 to 25% above every girls’ median and 6 to 41% above every boys’ median; none is below. Part of this is structural: the positive intercepts make the indexed Z = 0 line (mL/m²) turn upward again as BSA falls below its lowest point, at 0.31 to 0.75 m² depending on the measure, while Table 4’s indexed medians fall in the youngest children. Status stays experimental until an independent review and the site owner’s approval.

References

Z-scores are computed in this browser from the published equations; nothing you enter in the calculator is stored or sent.

Decision aid only. A z-score depends on how and where the structure was measured; compare like with like.