Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Neurocrit Care. Author manuscript; available in PMC 2010 August 1.
Published in final edited form as:
PMCID: PMC2818373

Comparison of Hematoma Shape and Volume Estimates in Warfarin Versus Non-Warfarin-Related Intracerebral Hemorrhage

Kevin N. Sheth
Division of Neurology, University of Maryland Medical Center, Baltimore, MD, USA
Tracy A. Cushing
Department of Emergency Medicine, Mt. Auburn Hospital, Boston, MA, USA
Lauren Wendell
Department of Neurology, Massachusetts General Hospital, Boston, MA, USA
Michael H. Lev and Javier M. Romero
Department of Radiology, Massachusetts General Hospital, Boston, MA, USA
Kristin Schwab
Department of Neurology, Massachusetts General Hospital, Boston, MA, USA
Eric E. Smith
The Calgary Stroke Program, Department of Clinical Neurosciences, University of Calgary, Calgary, AB, Canada
Steven M. Greenberg
Department of Neurology, Massachusetts General Hospital, Boston, MA, USA
Jonathan Rosand
Department of Neurology, Massachusetts General Hospital, Boston, MA, USA



Hematoma volume is a major determinant of outcome in patients with intracerebral hemorrhage (ICH). Accurate volume measurements are critical for predicting outcome and are thought to be more difficult in patients with oral anticoagulation-related ICH (OAT-ICH) due to a higher frequency of irregular shape. We examined hematoma shape and methods of volume assessment in patients with OAT-ICH.


We performed a case–control analysis of a prospectively identified cohort of consecutive patients with ICH. We retrospectively reviewed 50 consecutive patients with OAT-ICH and 50 location-matched non-OAT-ICH controls. Two independent readers analyzed CT scans for hematoma shape and volume using both ABC/2 and ABC/3 methods. Readers were blinded to all clinical variables including warfarin status. Gold-standard ICH volumes were determined using validated computer-assisted planimetry.


Within this cohort, median INR in patients with OAT-ICH was 3.2. Initial ICH volume was not significantly different between non-OAT-ICH and OAT-ICH (35 ± 38 cc vs. 53 ± 56 cc, P = 0.4). ICH shape did not differ by anticoagulation status (round shape in 10% of OAT-ICH vs. 16% of non-OAT-ICH, P = 0.5). The ABC/3 calculation underestimated median volume by 9 (3–28) cc, while the ABC/2 calculation did so by 4 (0.8–12) cc.


Hematoma shape was not statistically significantly different in patients with OAT-ICH. Among bedside approaches, the standard ABC/2 method offers reasonable approximation of hematoma volume in OAT-ICH and non-OAT-ICH.

Keywords: Cerebral hemorrhage, Tomography, X-ray computed, Warfarin


Hematoma volume is the most potent predictor of mortality and poor functional outcome in patients with intracerebral hemorrhage (ICH) [1, 2]. The ability to accurately calculate this volume at the bedside is critical for both medical care and clinical research. First, there is widespread variability in the use of DNR orders, which can significantly impact clinical decision making and patient outcome [3, 4]. Use of clinical decision scores, such as the ICH score or FUNC score, to help accurately guide such decision making, requires accurate hematoma volume estimation [2, 5]. In addition, many clinical trials include ICH volume as a criterion for inclusion [6, 7]. The most commonly used method of performing this bedside calculation is the ABC/2 technique, based on the formula for calculating ellipsoid volume [8]. This method is widely used and reliable.

Hematoma shape has been the subject of recent interest as a potential predictor of ICH growth [9]. One initial shape classification for ICH included three categories: regular, irregular, and separated [10], and these investigators went on to suggest that irregular shape may independently predict ICH growth [11]. These and other descriptions of ICH shape have been mostly qualitative [12]. It has been suggested that in oral anticoagulation-related ICH (OAT-ICH) patients, adjusting the formula to ABC/3 may lead to a more accurate measurement, in theory by accounting for more irregularly shaped hemorrhages [13]. However, this study did not compare OAT-ICH with non-OAT-ICH shapes or volumes. It is possible that this adjustment may avoid potential underestimation of ICH volume in OAT-ICH by ABC/2 [14].

In order to address the question of how best to perform this bedside assessment, we examined a cohort of patients with OAT-ICH against a matched set of patients with non-OAT-ICH. We tested two hypotheses: (1) oral anticoagulation-related hematomas were more likely to be irregularly shaped than non-warfarin hematomas and (2) ABC/3 provides a more accurate estimation of hematoma volume than ABC/2 for warfarin-related hematomas.


Study Subjects and Design

This was a matched case–control study of a prospectively collected cohort of consecutive patients with ICH presenting to a single tertiary care hospital [1517]. The study included the first 50 patients with OAT-ICH presenting to our hospital from October 1998 to June 2004, for whom records were available and CT scans available for reading, were identified. For each OAT-ICH, the next patient with ICH unrelated to warfarin but in a similar location [lobar vs. deep] was chosen as a control. This dataset was mostly identical to that used in a previously published analysis of perihematomal edema [18]. The study was approved by the Institutional Review Board.

Study subjects were greater than 18 years of age and presented within 24 h of symptom onset. Patients with ICH secondary to trauma, brain tumor, hemorrhagic conversion of ischemic stroke, or vascular malformation were excluded as patients with primary intraventricular and infratentorial hemorrhage. Patient demographics and characteristics have been previously described [1517]. Symptom onset was defined as the time when the subject or bystander noticed the sudden onset of neurological symptoms. When this time was not available, the patient's last seen well time was used as the symptom onset time. ICH was defined as OAT-ICH if warfarin was registered as a routine medication. Laboratory results including PT and INR were collected prospectively. In addition, we assessed subscores of hematoma volume based on two widely used prognostic scoring systems, ICH and FUNC [1, 2]. For both OAT-ICH and non-OAT-ICH, we evaluated the number of patients whose score would have been changed by ABC/2 versus the ABC/3 calculation.

CT Determination of Hematoma Volume and Shape

Hematoma volume and shape were determined from baseline CT scans obtained in the emergency room. Images were electronically transferred in DICOM format to a dedicated workstation for image analysis. ALICE software (Parexel Corporation, Waltham, MA) was used to manually trace the area of hemorrhage on each 5 mm tomographic slice. Planimetric volume measurements were used as described previously [16]. Readers were blinded to the clinical status of the patients. Hemorrhages in the subcortical white matter were classified as lobar, and those ICH that touched the thalamus or basal ganglia were classified as deep. Patients with cerebellar or brainstem ICH were excluded.

In order to approximate bedside volume determinations, four readers (KS, JG, TC, LW) independently measured both ABC/2 and ABC/3 volumes while blinded to other clinical data and radiographic images. The ABC/2 calculation was made as follows: A—maximal hematoma diameter, B—diameter of hematoma 90° to A, C—number of slices in vertical plane with hematoma multiplied by slice thickness of 0.5 cm. Any slice with less than 25% of overall hematoma volume was not included in C [8]. These values (ABC) were multiplied and then divided by 2 and 3, respectively, for each cohort. For each hemorrhage, the shape was categorized as round or irregular as previously described [10, 13] The interrater reliability was 95%, with a kappa of 0.80. Following the approach of Fujii et al., we compared the accuracy of the two formulas using the dichotomized categorization, regular and irregular [11].

Statistical Analysis

Continuous variables are reported as medians, 25th and 75th percentiles. Dichotomous variables were analyzed with Fisher's exact, and categorical and continuous variables were analyzed with the Kruskal–Wallis test. Unpaired t-tests were used to compare differences in accuracy between ABC/2 and ABC/3 calculations. All analyses were performed with Stata software (Stata Corp, College Station, Tex.).


From October 1998 to June 2004, 234 cases of spontaneous cases of ICH were presented to our hospital. Of these, there were 50 individuals whose ICH was related to warfarin; the median INR was 3.2 (2.4–4.4). In our study, only 3/50 patients in the OAT-ICH group had an INR < 1.5. They were included in the OAT-ICH group because they can still be considered coagulopathic and ultimately represented a small proportion of the 50 OAT patients. ICH volumes and shapes were analyzed in all 50 patients and in 50 location-matched controls. Table 1 shows the baseline demographics of these patients.

Table 1
Demographic features and baseline variables

Hematoma Size and Volume Measurements

Table 2 demonstrates volume analysis by formula and planimetric measurement. There was no significant difference in hematoma shape between the warfarin and nonwarfarin hematomas. In addition, there were no significant differences in hematoma volume between those with OAT-ICH and those with non-OAT-ICH.

Table 2
Accuracy of ABC/2 and ABC/3

We next compared the accuracy of ABC/2 and ABC/3 for different types of hematomas (Table 3). The ABC/3 method was not more accurate for any specific subset.

Table 3
Discrepancy between the ABC/2 and ABC/3 measurements

We evaluated whether any patient would be misclassified by a bedside system versus computerized planimetry. In two widely used ICH prognostic scoring systems, the ICH score [1] and FUNC score [2], a portion of the patient's score is determined by the presenting ICH volume. Thus, patients with a volume of <30 cc receive 1 point for the ICH score (0 points FUNC score), those with 30–60 cc receive 2 points (2 points FUNC score), and those with >60 cc receive 4 points using the FUNC score. We evaluated whether any patient would be misclassified by a bedside system. Specifically, we assessed how often the mismeasurement of ICH volume using either ABC/2 or ABC/3 would result in a change in either the ICH or FUNC score for the patient. No patient was misclassified using the ABC/2 method, while 17 patients with OAT-ICH would have been put into a lower 30 cc category, leading to an overly optimistic prediction.


Overall, we found no significant differences in shapes of hemorrhages between patients with OAT-ICH and non-OAT-ICH-related hemorrhages. In addition, we found that the ABC/2 method is more accurate than ABC/3 for OAT-ICH. These findings suggest that a single consistent volumetric calculation can be made irrespective of the underlying anticoagulation status.

We found that hematoma shape did not differ between OAT and non-OAT-ICH. While it may be that our sample size was too small to detect such a difference, this is the largest cohort so far used to address this question. Prior reports have suggested that disparate mechanisms of hemorrhage formation in patients taking warfarin may lead to more irregular hemorrhage shapes [19]. While the pathophysiology of OAT-ICH and non-OAT-ICH may differ, our results suggest that hematoma shape may not be a manifestation of this difference. It may be that final shape is more influenced by non-anticoagulation-related factors such as trajectory of the initial arterial rupture, varying locations of brain atrophy, physical architecture of the brain, and amount of ventricular blood. In this circumstance, it would appear likely that hematoma location would significantly influence final shape, given the different architecture of different cerebral locations. As location was used to select this cohort, we could not test this hypothesis here, and we have found no published data reporting an independent association between ICH location and shape. It is of note that one report has raised the possibility of an association between hematoma volume and shape, suggesting that larger ICH may be more irregular and that more irregular hematomas are more likely to expand [9]; however, any association with location is not yet clear.

Hematoma volume is a potent predictor of clinical outcome; therefore, the accuracy of the estimated volume is critical when providing prognostic information [1, 15, 16]. Due to its strong association with poor clinical outcome, hematoma volume has also been used as an endpoint in clinical trials [6, 20]. In the absence of a readily available quantitative computerized measuring program, it is imperative that the bedside formula clinicians and investigators utilize is accurate and precise. While the ABC/3 calculation was more accurate in a subset of one previous study [13], we were unable to validate this finding in any subset of our cohort.

It is notable that we found bedside volume measurement to underestimate volume in relation to computerized planimetry. In previous studies, ABC/2 volume has been demonstrated to be accurate for non-warfarin-related hemorrhages and OAT-ICH that were roundly shaped [8, 13, 21]. With regard to OAT-ICH with irregular shapes or involving multiple foci, reports have been inconsistent in concluding whether or not the ABC/2 method over- or under-estimated the actual hematoma volume [13, 14]. However, one such report included only four patients with OAT-ICH and a non-ellipsoid hemorrhage [14]. Another found ABC/2 to overestimate hematoma volume in contrast to our analysis in which ABC/2 and ABC/3 consistently underestimated hematoma volume [13]. However, the degree of overestimation was small; only 6% in the round group and 15% in the irregular group. Our cohort includes a wider range of hematoma volumes, and this may partially explain the disparity. In addition, our method of applying ABC/2 excluded those slices in variable “C” that contained at least 25% of overall hematoma volume, as was done in the initial study [8]. Inclusion of these slices may account for differences in volume calculations.

Ultimately, the issue of accuracy for a hematoma volume formula relates to prognosis. Neurologists often use bedside calculations of volume and validated scales to deliver prognostic information. Importantly, and most relevant for clinicians, we find that neither bedside calculation results in an overly pessimistic prediction, which might lead to inappropriately premature withdrawal of medical care.

There are limitations to this study. This was a nested case–control study performed within a prospectively ascertained cohort of ICH patients. While all CT studies were obtained on admission and early in clinical presentation, images were obtained at varying time points. Thus, if shape does change over time, we are unable to assess whether an individual hematoma that is one shape at a given time point on an imaging study is another shape earlier or later. It is also possible that the results of this study may not be generalizable to patients with other causes of hemorrhage, such as ICH related to thrombolysis or vascular malformations.


The study presented here is the first we know of to systematically compare hematoma shape in OAT-ICH with non-OAT-ICH. Because shape does not significantly differ in these two populations, similar methods may be used to calculate hematoma volume.

However, for maximally accurate measurements, efforts should be made to develop the ability to perform bedside computer generated planimetric analyses.


Funding was provided by The National Institute of Neurological Disorders and Stroke (NIH R01 NS059727 and K23NS059774), American Heart Association Grant-in-Aid #0755984T, and the Deane Institute for Integrative Study of Atrial Fibrillation and Stroke.


Conflicts of Interest Disclosure Dr. Goldstein has received consulting fees from CSL Behring and Genentech.


1. Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster G. Volume of intracerebral hemorrhage. A powerful and easy-to-use predictor of 30-day mortality. Stroke. 1993;24(7):987–93. [PubMed]
2. Rost NS, Smith EE, Chang Y, Snider RW, Chanderraj R, Schwab K, et al. Prediction of functional outcome in patients with primary intracerebral hemorrhage: the FUNC score. Stroke. 2008;39(8):2304–9. [PubMed]
3. Becker KJ, Baxter AB, Cohen WA, Bybee HM, Tirschwell DL, Newell DW, et al. Withdrawal of support in intracerebral hemorrhage may lead to self-fulfilling prophecies. Neurology. 2001;56(6):766–72. [PubMed]
4. Hemphill JC, III, Newman J, Zhao S, Johnston SC. Hospital usage of early do-not-resuscitate orders and outcome after intracerebral hemorrhage. Stroke. 2004;35(5):1130–4. [PubMed]
5. Hemphill JC, III, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke. 2001;32(4):891–7. [PubMed]
6. Mayer SA, Brun NC, Begtrup K, Broderick J, Davis S, Diringer MN, et al. Efficacy and safety of recombinant activated factor VII for acute intracerebral hemorrhage. N Engl J Med. 2008;358(20):2127–37. [PubMed]
7. Qureshi AI. Antihypertensive treatment of acute cerebral hemorrhage (ATACH): rationale and design. Neurocrit Care. 2007;6(1):56–66. [PubMed]
8. Kothari RU, Brott T, Broderick JP, Barsan WG, Sauerbeck LR, Zuccarello M, et al. The ABCs of measuring intracerebral hemorrhage volumes. Stroke. 1996;27(8):1304–5. [PubMed]
9. Barras CD, Tress BM, Christensen S, MacGregor L, Collins M, Desmond PM, et al. Density and shape as CT predictors of intracerebral hemorrhage growth. Stroke. 2009;40(4):1325–31. [PubMed]
10. Fujii Y, Tanaka R, Takeuchi S, Koike T, Minakawa T, Sasaki O. Hematoma enlargement in spontaneous intracerebral hemorrhage. J Neurosurg. 1994;80(1):51–7. [PubMed]
11. Fujii Y, Takeuchi S, Sasaki O, Minakawa T, Tanaka R. Multivariate analysis of predictors of hematoma enlargement in spontaneous intracerebral hemorrhage. Stroke. 1998;29(6):1160–6. [PubMed]
12. Zhan RY, Tong Y, Shen JF, Lang E, Preul C, Hempelmann RG, et al. Study of clinical features of amyloid angiopathy hemorrhage and hypertensive intracerebral hemorrhage. J Zhejiang Univ Sci. 2004;5(10):1262–9. [PMC free article] [PubMed]
13. Huttner HB, Steiner T, Hartmann M, Kohrmann M, Juettler E, Mueller S, et al. Comparison of ABC/2 estimation technique to computer-assisted planimetric analysis in warfarin-related intracerebral parenchymal hemorrhage. Stroke. 2006;37(2):404–8. [PubMed]
14. Freeman WD, Barrett KM, Bestic JM, Meschia JF, Broderick DF, Brott TG. Computer-assisted volumetric analysis compared with ABC/2 method for assessing warfarin-related intracranial hemorrhage volumes. Neurocrit Care. 2008;9(3):307–12. [PubMed]
15. Rosand J, Eckman MH, Knudsen KA, Singer DE, Greenberg SM. The effect of warfarin and intensity of anticoagulation on outcome of intracerebral hemorrhage. Arch Intern Med. 2004;164(8):880–4. [PubMed]
16. Flibotte JJ, Hagan N, O'Donnell J, Greenberg SM, Rosand J. Warfarin, hematoma expansion, and outcome of intracerebral hemorrhage. Neurology. 2004;63(6):1059–64. [PubMed]
17. O'Donnell HC, Rosand J, Knudsen KA, Furie KL, Segal AZ, Chiu RI, et al. Apolipoprotein E genotype and the risk of recurrent lobar intracerebral hemorrhage. N Engl J Med. 2000;342(4):240–5. [PubMed]
18. Levine JM, Snider R, Finkelstein D, Gurol ME, Chanderraj R, Smith EE, et al. Early edema in warfarin-related intracerebral hemorrhage. Neurocrit Care. 2007;7(1):58–63. [PubMed]
19. Hart RG. What causes intracerebral hemorrhage during warfarin therapy? Neurology. 2000;55(7):907–8. [PubMed]
20. Mayer SA, Brun NC, Begtrup K, Broderick J, Davis S, Diringer MN, et al. Recombinant activated factor VII for acute intracerebral hemorrhage. N Engl J Med. 2005;352(8):777–85. [PubMed]
21. Gebel JM, Sila CA, Sloan MA, Granger CB, Weisenberger JP, Green CL, et al. Comparison of the ABC/2 estimation technique to computer-assisted volumetric analysis of intraparenchymal and subdural hematomas complicating the GUSTO-1 trial. Stroke. 1998;29(9):1799–801. [PubMed]