How this instrument works
Intracerebral hemorrhage on CT looks roughly like a squashed ellipsoid, and the ABC/2 method treats it as exactly that. A is the largest diameter of the hemorrhage on the CT slice where it looks biggest; B is the diameter measured perpendicular to A on that same slice; C is how many slices show any hemorrhage at all, multiplied by the slice thickness, giving a rough third dimension. Multiply the three together and divide by two — the division by two comes from folding the standard ellipsoid volume constant, four-thirds pi over six, together with the approximations built into how B and C are measured, so the arithmetic simplifies to a single, memorable division.
Rosemary Kothari, Thomas Brott, Joseph Broderick and colleagues at the University of Cincinnati described the method in 1996, timing it against a slower, more rigorous computer-based planimetric measurement in 118 patients. The bedside ABC/2 estimate took a mean of 38 seconds per scan and correlated with the planimetric gold standard at R-squared of 0.96, with near-perfect agreement between different observers and between the same observer measuring twice. That speed and reliability, measured against a harder method rather than assumed, is why ABC/2 spread so quickly into stroke units and emergency departments.
The version most people learn and this instrument implements is the simple one: count every slice that shows any hemorrhage at all, full weight, no matter how small the bleed looks on that particular slice. Kothari's original description was actually more granular — a slice counted fully only if its hemorrhage area was at least 75% of the largest slice's area, half-weight between 25% and 75%, and not counted at all below 25%. A 2017 comparison by Khan and colleagues found the simplified, unweighted version performs comparably to the more granular original at the clinically important 30 cm³ threshold, which is presumably why the simple version became the one actually taught and used.
- Enter A, the largest hemorrhage diameter in centimetres, measured on the CT slice where the bleed looks biggest.
- Enter B, the diameter perpendicular to A, measured on that same slice.
- Enter C, the number of CT slices showing any hemorrhage multiplied by the slice thickness in centimetres.
- Read the estimated hemorrhage volume in cubic centimetres from the working block.
Worked example — A 4 cm, B 3 cm, C 5 cm
A hemorrhage measures 4 cm across its widest point and 3 cm perpendicular to that, on the CT slice where it looks largest. It shows up on slices spanning 5 cm of hemorrhage-positive thickness. Multiply the three: 4 × 3 × 5 = 60, divide by 2, and the estimated volume is 30.0 cm³ — landing exactly on the 30 cm³ threshold this site's own ICH score calculator uses to award a point toward its mortality prediction.
A smaller bleed, 2 cm by 2 cm across 2 cm of slices, comes to 2 × 2 × 2 ÷ 2 = 4.0 cm³ — well under any severity threshold. A larger one, 6 cm by 5 cm across 7 cm, comes to 6 × 5 × 7 ÷ 2 = 105.0 cm³ — well above it, and large enough on its own to weigh heavily in most bedside prognostic assessments.
Questions
Why divide by two instead of using the full ellipsoid formula?
Because the full ellipsoid volume formula is four-thirds pi times the product of three radii, and ABC/2 works in diameters instead of radii and folds the geometric constant together with the rough, slice-counting way C gets measured. The result is that the awkward constant collapses to a plain division by two, which is precisely why the method spread — the arithmetic is fast enough to do at the bedside without a calculator, let alone software.
How accurate is ABC/2 compared to computer-based volume measurement?
In Kothari and colleagues' original 1996 validation against 118 patients, ABC/2 correlated with formal planimetric measurement at R-squared of 0.96, with an average measurement time of 38 seconds and excellent agreement both between different raters and for the same rater measuring twice. It is an estimate, not a precise volumetric reconstruction, but it is a well-validated one.
Does this calculator use the original, weighted version of C?
No — it uses the simple, unweighted version: every slice showing any hemorrhage counts fully toward C, regardless of how large the bleed looks on that slice. Kothari's original 1996 description actually proposed weighting slices by how much of the largest slice's area they covered — full weight above 75%, half weight between 25% and 75%, zero below 25% — but the simple, full-weight-per-slice version is what became standard bedside practice and what most current calculators, including this one, implement.
Does the simplified version give a meaningfully different answer?
Rarely, according to a 2017 comparison by Khan and colleagues, which found the simplified and weighted versions performed comparably against a computer-assisted planimetric standard, particularly around the clinically important 30 cm³ cutoff. The simplified version showed less bias overall against the gold standard in that comparison, which is one reason it has not been displaced by the more granular original.
Why does the 30 cm³ mark matter specifically?
It functions as a decision threshold in several bedside prognostic tools, including this site's own ICH score, where a supratentorial hemorrhage volume of 30 cm³ or more contributes a point toward a composite score that predicts 30-day mortality. A volume estimate that lands right at that line, as in the worked example above, is exactly the situation where knowing whether ABC/2 tends to run a little high or low on borderline bleeds matters most.
What are the main sources of error in ABC/2?
Irregularly shaped hemorrhages are the biggest one — the method assumes something close to an ellipsoid, and bleeds that are markedly irregular, multilobulated, or mixed with intraventricular extension depart from that assumption in ways a three-measurement estimate cannot capture. Slice thickness on the scan also matters for C: thicker slices give a coarser, less precise estimate of that third dimension than thinner ones do.
References
- Kothari RU, Brott T, Broderick JP et al., 1996, Stroke — original ABC/2 method (PubMed)
- Khan M et al., 2017, J Neuroimaging — simplified vs weighted comparison (PubMed)
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.