Aims Bruise photographs serve as a clinical record and may facilitate forensic analysis if physical child abuse is suspected. Currently, only conventional (unfiltered visible light) imaging is employed; however, alternative imaging modalities may provide additional information. Observer variation in interpretation of colour and age of bruises, using conventional imaging techniques, is considerable. Findings from two pilot studies will be presented: (1) Within- and between-observer variation when measuring bruises on printed and digital conventional images; (2) Image modality preference of paediatricians when assessing bruise photographs, and between-observer agreement therein.
Methods Study (1) 45 observers recorded the greatest lengths of six bruises on two occasions. All completed manual and electronic measurements, using a paper tape-measure for printed images and ImageJ software on a lap-top for digital images. Study (2) Nine Child Protection paediatricians independently ranked five image modalities (conventional colour, conventional grey-scale, cross-Polarised, ultraviolet, and infrared), for four bruises, by completing a questionnaire.
Results Study (1) Widespread observer variation existed for both manual and electronic measurements; within-observer SD ranged from 2.1 to 4.7, between-observer SD from 2.1 to 5.2. Observers were found to be prone to rounding bias (digit preference to nearest 5 mm) upon manual assessment. Paired t tests showed mean bruise sizes were smaller for manual measurements than for electronic, this difference was significant 10 of 12 times.
Study (2) Paediatricians almost unanimously preferred cross-Polarised for all four bruises when assessing boundary, shape, colour, size, and absence of light reflectance. Conventional colour and grey-scale were typically ranked second and third. Ultraviolet and infrared were consistently ranked in the last two positions. Between-observer agreement on preference was high, with coefficients of concordance between 0.76 and 0.96. Image modality combinations, chosen to give the most complete picture of the bruise, predominantly consisted of cross-Polarised and conventional (colour and grey-scale).
Conclusions In order to increase consistency and reliability of recordings, a standardised automated method for bruise assessment is sought. Cross-polarised imaging is not currently requested from medical photographers within the context of bruising, but would appear to have the potential to better assess size and boundary. More extensive research to define the optimal imaging protocol is vital.