A New Practical Intracerebral Hematoma Volume Calculation Method - - PowerPoint PPT Presentation

A New Practical Intracerebral Hematoma Volume Calculation Method and Comparison to simple ABC/2

F.Cuce¹, G. Tulum, O. Dandin, T. Ergin,

• O. Karadas, O. Osman

¹Gulhane Education and Training Hospital, Department of Radiology, Ankara, Turkey

Introduction

• Intracerebral

Hemorrhage Scoring (ICH) is used to predict morbidity and mortality in the cerebral hematoma.

• Surgical indication in the

MISTIE-III protocol.

• The accurate and timely calculation of brain

hematoma volume is important.

• The most common and practical formula

frequently used is simple ABC / 2 (sABC / 2).

• The planimetric method is the gold standard,

but in practice, it is not yet available.

• Our study is about the use of ABC / 2, and

developing a more effective formula.

Subjects and methods

• We reviewed the records of 128 patients from

January to April 2017.

• We only enrolled acute and spontaneous

parenchymal and subdural hematomas in our

• study. (49 patients with 58 hematomas were

enrolled in the study).

• We obtained a new formula as V=0.34ABC+10.
• Our method, sABC/2 formula were compared

with planimetric method (using ManSeg 2.6 software) as a gold standart

Computed Tomography Technique:

• All examinations were performed on 64-slice

CT scanner.

• The scanned objects were also reconstructed

in 3.0 mm thick transverse.

• Matrix size is 512 x 512.

Hematoma volume calculation

• AxBxC/ 2
• Choose the slice where the

hematoma is the tickest.

• A:maximal

anteroposterior diameter

• B:hematoma thickness
• C:number of slices in which

the hematoma is visualized multiplied by the slice thickness

• Determinations of hematoma volume and

shape were performed by two radiologists.

• The hematoma volumes were both calculated

with sABC/2 and the proposed formula by the calculator, and also with the gold standard technique using ManSeg 2.6 software.

Statistical Analysis:

• Agreement between planimetry and baABC was

evaluated using Bland–Altman plots. Sensitivity, specificity, and AUC were calculated. Root mean square errors of the methods were obtained.

• Concordance between planimetry and volumes
• btained by other estimation methods were

assessed using t-test.

• Bland–Altman plots were generated for sABC/2

and baABC methods in comparison to the planimetric method using both original and log transformed units.

Results

• All methods were

concordant with the planimetric method.

• Variability in the

concordance between planimetry and the two methods appeared to increase as volumes increased.

• V=0.34ABC+10

• baABC method

performs better than sABC/2.

• For parenchymal cases,

both methods approximately have close error values,

• But in case of subdural

hematoma, baABC is 15.1 ml better than sABC/2.

sABC/2 baABC RMSE of all ICH (mL) 33.99 25.08 RMSE of parenchymal ICH (mL) 28.61 28.20 RMSE of subdural ICH (mL) 37.56 22.43

• Volume errors of

cerebral hematoma for all cases.

• Sensitivity and specificity values were
• btained for differentiating a cerebral

hematoma volume of > 30 mL.

• sABC/2 had a sensitivity of 97 % at a

specificity of 96 % and AUC: 0.99,

• baABC had a sensitivity of 94 % at a specificity
• f 92 % and AUC: 0.99.

• Concordance between

planimetry, sABC/2 and baABC was all high and all are about 0.90.

Method Cerebral hematoma 95% CI sABC/2 0.90 (.84- .94) baABC 0.90 (.84- .94)

Discussion

• Interobserver correlation of sABC/2
• The high coherence degree (ICC:.96) of sABC/2 amongst the readers from

different departments (Khan et al).

• The planimetric method correlation was lower for various readers with

little experience in comparison to an experienced single reader (Haitham et al) .

• The correlation of sABC/2 and planimetric method were found to be high

amongst the readers as 0.91-1.0 between two radiologists (Divani et al).

• In our study, a high degree of correlation between two radiologists was
• btained (0.98) .
• Also the degree of correlation between planimetric method, the proposed

formula and sABC/2 were calculated as equal (0.90).

• Parenchymal hematoma errorr rates for sABC/2
• The error rate was 10 % in < 20 ml, 37 % in > 40ml

(Haithamın et al).

• The error rate of was 9.9 % in < 20 ml, 37.1 % in > 40 ml

(Wang et al).

• This error rate, which increases parallel to the volume,

generally tends to be overestimated.

• In our study, planimetry to sABC/2 and baABC ratios were all a

bit smaller than 1( 0.98 and 0.96).

• Parenchymal hematoma shape
• Erroneous results

with regular shape hematoma are by 3.33 ml (9.76 %) and in irregular shaped hematoma 7.19 ml (18.37 %) (Xu X et al).

• The over-measured volume by sABC/2 is 32.1 % in an

irregularly shaped hematoma (Huttner et al).

• In our study, the planimetric correlation of our new formula

and sABC/2 in irregularly shaped hematoma (for 10 cases) is by 63 %.

• Subdural hematoma
• The correlation with planimetric method was high [average

volume is 91.0 ml with the sABC/2 and 82.4 ml with the planimetric method] (Gebel et al).

• Sucu et al. found that the correlation of all the five different

sABC/2 types which they used at different measurement points was high.

• In our study, baABC is 15.1 ml better than sABC/2 in subdural

hematomas.

Thank you…