METHODS

A Cary 100 Bio UV-visible spectrophotometer fitted with a Cary fibre optic couple and Cary fibre optic reflectance probe (Varian Australia Pty Ltd, Frenchs Forest, New South Wales, Australia) was used to obtain the spectra (purchase price AU$ 22 000 in October 2000). Zero and baseline corrections were used. The baseline was collected using a labsphere reflectance standard (Varian Australia Pty Ltd). The probe was placed on the standard with an interposing layer of acetate with the baseline/zero mode selected. The optimal height of the probe within the boss was determined by finding the point of maximal light throughput using the Cary software. Then, the 100% reflectance baseline was recorded. To set the 0% baseline, the boss was placed on a closed black canister in which a hole permits light from the probe to enter the canister without reflection.

After baseline collection, the acetate target was placed on to the area of the skin to be examined. The boss containing the probe was then placed on to the acetate target, taking care to match up the perimeter of the probe with the perimeter photocopied on to the acetate (fig 1). Once the probe was in place, the absorption spectrum of the bruise was obtained over the range 780–380 nm. A data interval of 1 nm was selected, with a signal averaging time of 0.2 seconds for each data point (resulting in a scan time of 70 seconds). Double beam mode was used with 2.6 Abs of rear beam attenuation and the slit width was set to 3.0 nm. It was necessary to darken the room while obtaining the spectra to prevent extraneous light travelling through the acetate being captured by the probe. It was also important to prevent the probe, or fibre optic cable, from moving during the capture of the spectrum because any movement can create background noise, compromising the reproducibility of the system.

A series of measurements was taken from the same site of a fresh bruise by leaving the acetate in place, but removing and reapplying the probe. Next, a series of areas was sampled within the same bruise. Finally, after three days, the bruise was scanned again. All scans were performed by the same operator. The first derivative of each scan was calculated using the spectral calculator of the CaryWin UV software (supplied by Varian Australia Pty Ltd with the Cary 100 Bio UV-visible spectrophotometer). A filter size of 9 and an interval of 3 were found to be optimal.

Fifty scans of bruises from live human volunteers were acquired (human research ethics committee approval reference HS/TG HREC2002/5/4.6(1432)). All the bruises had been acquired in normal daily activities and only bruises on the arms or legs were studied. The age of the bruise was ascertained by asking the volunteer and only bruises for which the volunteer could accurately pinpoint the time of injury were used. Colorimetry values were calculated using the Cary Win UV color application, 85-101684-00 version 2.00(15) supplied by Star-tek (Victoria, Australia). CIEL*a*b values were calculated based over a scan range of 780–380 nm using a 1 nm data interval. Correction was performed using the zero and baselines of the scans. Results were based on a D65 illuminant with an observer angle of 2°. Colour matching or colour correction were not used. Using the software package, a negative value indicated more yellowness.

Statistical analysis was performed using Statview version 4.5 (Abacus Concepts Inc, Berkeley, California, USA).

RESULTS

The series of scans taken by repeated applications of the probe to the same area of the bruise showed excellent reproducibility (fig 2). The expected absorption peaks of haemoglobin at 576 nm, 543 nm, and in the range 414–422 nm were present.²⁶ The absorption range is just over 1 Abs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Scans of fresh bruise (less than six hours old): five consecutive scans of the same area.

Scans taken from different areas in the fresh bruise (fig 3) show a greater degree of variability. However, when the first derivatives of these scans are produced (fig 4), it can be seen that the variability of the baseline is removed and that the curves are roughly confluent over the range 510–470 nm. On the first derivative scans, the magnitude at 480 nm ranges from 0.0 to −0.00128 Abs/nm. At 490 nm, the magnitude ranges from +0.00050 to −0.00021 Abs/nm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Scans of fresh bruise (less than six hours old): four different areas.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

First derivatives from scans in fig 3.

Spectra of the bruises three days later again showed variation between areas sampled (fig 5). Absorption peaks caused by haemoglobin are still present, but are reduced in intensity. There is a new slope over the region 510–470 nm. There is no perceptible absorbance peak in the range 660–620 nm. The first derivatives again reduce the differences between the scans (fig 6). In the first derivative scans, the absorption at 480 nm ranges from −0.00193 to −0.00237 Abs/nm. At 490 nm, the absorption ranges from −0.00151 to −0.00205 Abs/nm. The first derivatives show no evidence of absorption as a result of biliverdin because there are no peaks in the range 660–620 nm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Scans from old bruise (three days): four different areas.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

First derivatives from scans in fig 5.

The 50 bruise scans were acquired from 27 bruises taken from 25 volunteers (two volunteers presented on two separate occasions with bruises). The age range of the subjects was from 22 to 63 years (mean, 37.5; median, 36). The sites of the bruises were: upper arm, eight; lower arm, four; hand, five; upper leg, two; lower leg, eight. Two bruises were scanned four times, four were scanned three times, nine were scanned twice, and 12 were scanned on one occasion only. The “b” colorimetry value of 50 bruises ranging from 1 to 378 hours old (median, 95) varied from −3.73 to −0.77 (median, −1.3). The first derivative values at 490 nm ranged from −0.06 to −0.0002 Abs/nm (median, −0.001), and the first derivative values at 480 nm ranged from −0.07 to −0.0002 Abs/nm (median, −0.002). A correlation analysis revealed that an increasing negative first derivative value at 490 nm corresponded to a more negative (yellow) “b” value (coefficient of correlation²  =  0.78; fig 7). The correlation of the first derivative value at 480 nm was not so good (coefficient of correlation²  =  0.72).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Correlation between colorimetry “b” values and first derivatives at 490 nm in 50 bruises.

DISCUSSION

Reflectance spectrophotometry has previously been used to study the pigments and colour of living human skin.²⁷ For example, this method has been used to estimate the melanin density in white individuals to determine an individual’s susceptibility to epidermal tumours.²⁸ The technique has also been used in the assessment of wound healing.²⁹ In addition, it has previously been used to study bruises (DO Carson MSc thesis, University of Dundee, UK, 1998); however, in that study, although reflectance spectroscopy was sufficiently sensitive to identify the presence of haemoglobin in a bruised area, it was unable to identify haemoglobin degradation products. Our present study shows that reproducible readings can be obtained from one point within a bruise (fig 2). However, as would be expected, there is variation between different areas within a bruise (fig 3). This variability makes comparison between bruises difficult and could mask the presence of degradation products of haemoglobin. This problem can be overcome by obtaining first derivatives of the scans.

The first derivative method calculates the slope of the scan. The first derivative of a rapidly rising line is a large positive value. For a gentle downward sloping line it is a small negative value. The first derivative of a flat line is zero. Therefore, for the region of the spectrum of haemoglobin from 510 to 470 nm, where the spectrum is almost level (fig 3), the first derivative is around zero (fig 4). Bilirubin has a broad absorption peak at 460 nm.^14,30 Thus, superimposition of the descending side of the bilirubin curve on the flat region of the haemoglobin spectrum results in a negative first derivative value in the range 510–470 nm. Previously published work using the first derivative method to measure bilirubin in the presence of haemoglobin used the first derivative value at 490 nm and at 480 nm.³¹ We found that the measurement of the first derivative value at 490 nm shows more change than the value at 480 nm so that 490 nm is better for discriminating between fresh bruises and three day old bruises (fig 6).

The first derivative method also has the advantage of reducing the difference between scans where there is a difference in baseline offsets. This is seen in fig 4, where there is near confluence of the lines in the range 510–470 nm, and where all the original curves (fig 3) are flat, but have different baseline offsets. There is less confluence in the same region in fig 6, which is attributed to heterogeneity of bilirubin distribution within the bruise.

“We found that the measurement of the first derivative value at 490 nm shows more change than the value at 480 nm so that 490 nm is better for discriminating between fresh bruises and three day old bruises”

We had previously determined, using a Kodak colour separation guide (cat 152 7654), that there was an excellent, but negative, correlation between Adobe Photoshop (version 6.0 for Macintosh, Adobe Systems Ltd) CIEL*a*b “b” values derived from image analysis of a photograph and the colorimetry “b” values obtained by directly scanning the colour reference card (coefficient of correlation²  =  0.92; results not shown). The correlation between the first derivative values at 490 and 480 nm are consistent with the suggestion that greater amounts of bilirubin correspond to increasing yellow colour in a bruise.^12,32,33 The coefficient of correlation was better for the 490 nm values than for the 480 nm values, suggesting that the 490 nm first derivative value would have a greater potential for use in the determination of the age of a bruise.

Haemoglobin is metabolised by haemoxygenase to biliverdin.^34–36 Biliverdin has an absorption peak in the range 660–620 nm.³⁷ It would be expected that it would not be possible to detect the presence of biliverdin, because it has been shown that it is rapidly metabolised to bilirubin by biliverdin reductase.^34–36 There is no discernable peak in the range 660–620 nm in the three day old bruises in the native (fig 5) or in the first derivative scans (fig 6), which is consistent with biliverdin being short lived.

It was found that the acetate could allow surrounding light to reach the probe, by acting as a light channel. Although this could be avoided by using a blackened acetate sheet, reduction of ambient light was found to be sufficient to avoid this problem. The fibre optic probe is extremely sensitive to movement, with any movement producing noise on the scan. Therefore, it is essential that the probe is held steady. It must be noted that the probe records only surface refection and will only capture the results of biological processes near the surface of the skin.

With the above methodology, using the spectrophotometer and the supplied software to derive the first derivatives, it is possible to produce scans that can differentiate between new and old bruises. Furthermore, the results are consistent with the suggestion that the yellow colour in an old bruise correlates with the presence of bilirubin.^12,33 However, at least some of the yellow colour may also be attributable to the local formation of haemosiderin. The local formation of haemosiderin within a bruise is well documented.^9,10,38 Haemosiderin has been described as having a brown colour in the tissue, but by using the Perl’s staining method, it appears blue.³⁹ Haemosiderin is formed when the iron that is released by the degradation of haemoglobin is complexed with protein (mostly ferritin). Its atomic structure (determined by x ray diffraction pattern analysis and electron microscopy) is indistinguishable from ferritin, and it appears that haemosiderin comprises an aggregate of ferritin particles that have undergone partial proteolysis.^11,40,41 The ultraviolet to visible absorption spectra of ferritin and haemosiderin show increasing absorption from long to short wavelengths, with no specific peaks.^42,43 Any superimposition of absorption as a result of haemosiderin over the spectra of haemoglobin and bilirubin would increase the negative value of the slope from 470 to 510 nm (where the absorption by haemoglobin is flat and the absorption of bilirubin is declining¹⁴). Therefore, the presence of haemosiderin would increase the negative value of the first derivative at 480 or 490 nm, and it is not possible to distinguish absorption as a result of haemosiderin from that of bilirubin. Thus, the yellow colour in a bruise is attributed to the formation of local breakdown products of haemoglobin, with the spectroscopic measurement of absorption at 480 or 490 nm being predominantly attributed to bilirubin (because it shows the greater rate of change over that range).

The ability simultaneously to demonstrate the presence of haemoglobin and measure its degradation, revealing a correlation between local haemoglobin degradation and the formation of yellow colour in bruises in living human subjects by a non-invasive method, has not been described previously. Studies are being performed to test the applicability of this technique to the ageing of bruises. Specifically, a large study is in progress to acquire a sufficient number of volunteers to investigate issues such as: whether body location exerts an effect, the effect of skin colour, whether all bruises with a bilirubin spectrum appear yellow, calculation of sensitivity and specificity data, whether naked eye detection is better than spectrophotometry, and the effect of the size of the bruise or other variables. Work is also in progress to assess the application of this method to bruises in cadavers and to compare the spectrophotometry results with histology. It is also intended to investigate the effects of lividity and decomposition.

Our study has shown that, as would be predicted, biliverdin does not accumulate,^34–36 and that the yellow colour that develops in a bruise can be attributed to the local degradation of haemoglobin. Given that it takes time for macrophages to accumulate^9,38 and metabolise haemoglobin,¹¹ this would account for the previously reported delay before a yellow colour is apparent in a bruise.⁴