Bruising software

According to UK academics, in some criminal cases forensic experts have made significant errors when they have used photographs to determine the time that bruises were inflicted on victims of crime.

Prof Peter Vanezis from the William Harvey Research Institute said this month that although forensic scientists are often asked to give their expert opinion as to when bruising had occurred on a victim based on photographic images, they can’t accurately determine when the bruise occurred.

The news comes as the result of a study that Prof Vanezis and his team recently undertook when they asked forensic experts to look at 132 photographic images of bruises, ranging from 0 to 209 hours in age.

The bruises were produced on the upper arms of 11 subjects by a suction pump. Daily sequential photographs were taken until they were no longer visible to the naked eye, and then the 15 forensic experts were asked to estimate the age of the bruises and also place them in chronological order.

Margaret Pilling, a medical student who worked with Prof Vanezis on the study, said that while the forensic experts were most accurately able to pinpoint the time in very fresh bruises – between 0 and 12 hours – there were still a number of significant misjudgements.

What’s more, she said that the median difference between the estimated age of the bruises and the real age was 26 hours – a considerable disparity that led the research team to its conclusion that the forensic experts’ estimates of bruise age from photographs were unreliable.

While none of this forensic science might initially seem to have much do with engineering, my dad always told me that behind every problem lies an opportunity just waiting to be exploited, and clearly, there exists a burning need to automate this entire bruise identification system.

Now I’m not sure how much of an engineering challenge that might be, but if it were up to me, I’d first go about capturing those images that the researchers have produced with an image scanner and then write some artificial neural network and possibly some fuzzy classifier software to more accurately determine when the bruises were inflicted.

After all, it’s an approach that has already been tried and tested to examine bruises on apples and potatoes, so why not the human skin?

Whether my solution might be viable or not, the news made me realise that there must be an enormous number of similar such problems out there where researchers could possibly make use of engineering skills to solve their problems, but – either though lack of knowledge or expertise – are unaware of who to contact to help.

And although there are plenty of wonderful government schemes to transfer technology out of the universities to help industry, perhaps what we need is a simpler approach.

Why not just develop a simple bulletin board where problems such as the one discovered by Prof Vanezis might be posted so that entrepreneurial folks in industry could pick them up and develop solutions?