Crime scene investigation - the future | Wim Develter and Bram Bekaert | TEDxLeuven

If you look into the number of crime related television series, you know it is a topic that interests many. In this talk, join Wim Develter and Bram Bekaert in this real-life crime scene investigation, and discover the new technologies that can be used to investigate a homicide.

Wim Develter is a forensic and clinical pathologist linked to Gasthuisberg, the University Hospital of Leuven. His area of expertise focuses on corporate damage.

Bram Bekaert started his career in 1996 as a police officer in Brussels while studying part-time for a BSc in Biotechnology. He moved to the UK to study forensic science and obtained a second bachelor degree and an MSc in Forensic DNA Profiling at the University of Central Lancashire. In 2009, he obtained his PhD in Nutri-epigenomics at the University of Surrey and currently is a forensic geneticist at the Laboratory for Forensic Genetics and Molecular Archaeology at the University Hospitals Leuven.

This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at http://ted.com/tedx

Wim Develter: I'd like
to start with a story.

On a sunny Friday afternoon in June 2008,

Peter Van Heel, a 58-year-old farmer,
was walking on his fields

when suddenly, he discovered a body.

He was totally in panic
and called the police.

And as we are Belgium's largest
forensic department,

we were appointed
for the crime scene investigation.

So, arriving on the crime scene,

I could see the remnants
of a decomposing adult.

And all the clothes were burned.

But there were no signs
of fire accelerants.

And when we took a closer look,

I could see some
fancy jewels on the hands,

so I thought, 'This must be a woman.'

And there were some ropes visible
around her wrists and feet.

And as I turned the head a little bit,

I could see three round
skin perforations in the neck.

So that's where we knew

this victim, probably a woman,
was shot, murdered,

packed, transported, dumped and burned.

So, the obvious questions
everybody asked themselves are,

'What happened?'

'Who is this person?'

and, 'When did it happen?'

To answer these, we transported
the victim to the mortuary,

where we did a total-body CT scan,
a forensic autopsy

and a forensic
anthropological examination -

so, the study of the bones.

So, during routine autopsy,

we tried to use these rods,

and we put them through
the holes of the bullets

in order to reconstruct
the bullets' trajectory.

And it's kind of difficult
and time-consuming.

And even in a fresh body,
it's already very difficult.

So, in this case, with decomposition,

this task becomes nearly impossible.

But when you take a look at the damage

caused by a bullet
when it goes through a head,

it perforates the skin, the skull,

and on its path, you have some scattering
of bony fragments from the skull,

little fragments, metallic fragments
from the bullet, air and blood.

So we were thinking,

'Can't we use our CT data

to reconstruct in a digital way
a bullet trajectory?'

So, I did what every Belgian guy
does with a problem:

I went to the pub.

And together with the forensic engineers
and forensic radiologists,

we got drunk,

drunk of ideas, of course.

And the day after,

we had a huge scientific hangover

as we were realising
the complexity of our idea

because we had to test and validate
our idea on a model,

on an animal model, of course.

So we chose sheep.

Not because they're ugly or stupid,

but because they have
a large brain volume,

and they're easy to get.

So, after two years
of programming and testing,

we succeeded to visualise the damage.

But now, we had to apply this
to human skulls and human heads.

So, this is how it works.

Our engineers created a matrix

in which all the heads are divided

in thousands and thousands
of little cubes.

And each little cube
of the traumatised head -

so, meaning the head that has been shot -

is compared with the same cube
in the same anatomical position

of the normal head,

repeating that process for all the cubes
in the traumatised head

to all the heads in a normal data set.

So, that was a huge effort of time,

and after a while, we succeeded.

And what came out was fantastic,

as you can see in this video.

We succeeded

to create the bullet trajectories.

So, in this victim,

the victim was shot
with three gunshots in the neck.

And only one was directly lethal,

and that's the red one
dissecting the right vertebral artery.

So, this approach has many advantages.

First of all, it's fast.

It's much faster than an autopsy.

It's not invasive, so your pathologists
just don't have to cut anymore.

And it's more accurate.

Like in a ricochet -

that means when a bullet
changes direction in the head -

that can easily be seen
with our software tool.

But that's simply not possible
with these rods.

And last but not least,

even in decomposition like in this case,

if you know the bullet's trajectory,

you will know through
which anatomical structures it will go.

So, this question is answered.

We know how this victim died.

Still two questions to go:

who is it and when did it happen?

So, being burned and decomposing,

visual identification is not possible.

And having no dental treatment,

forensic odontology
is not an option either.

So the only thing remaining
is forensic anthropology.

So, what do those bones tell us?

So, based on the description
and measurements

of the forms and the bones
of the skeleton,

we are able to see
something about stature.

In this case, this victim
was about 172 to 177.

Also about the gender -

well, it appeared that there
was an 80 percent probability

that this person was a man.

So, not a woman with fancy jewels.

And he was probably Caucasian.

And the age - and that's a tough one -

between 23 and 57.

So, that's rough science, isn't it?

We could need some help there.

So that's why we sent out a Mayday

to our friends and colleagues
forensic geneticists.

Bram Bekaert: Right. So, we have
an age range of about 34 years.

That's huge, isn't it?

I bet this description
fits about half the men

present here in this theatre.

So, if you want to be able
to put a name down to this body,

we need to narrow this group down.

And this is where forensic genetics
might help a bit.

Now, the standard procedure
for the identification of a person

is to produce a DNA profile
using a blood sample, for example,

or a muscle tissue

and generate a profile

and compare that to a toothbrush
or a profile from his mum or dad

if you have some idea who that person is,

and get a formal
identification in this way.

But what if you have
absolutely no idea who this victim is?

Well, there's lots of information
in our DNA sequence

that can actually reveal
what we look like.

You know, we can use someone's DNA
just in the same way

as an eyewitness would give
a description of someone.

We can almost generate a DNA photofit.

Now, what can DNA tell us today?

Well, we can tell someone's eye, hair
and skin colour, for example,

with fairly good accuracy.

But we can also tell
where someone is originally from

through a biogeographical ancestry test.

And my colleague Peter Claes
told you last year on this same forum

about this new revolutionary technique
of 3D facial morphology

just by looking at someone's DNA.

Now, this is all helpful
and wonderful, obviously,

but age is a very important factor too

when you're trying to build
an image of someone.

And the classical methods
that provide age ranges today,

such as anthropology, odontology,

are just not good enough.

So, this is where
our second idea came from.

We wanted to develop
a new age-estimation test

basically with far better accuracy.

And we wanted to do so
from a single drop of blood.

So, how did we do this?

Well, there are several
processes in our body

that make sure that we
don't like ageing, right,

that we make room for the next generation.

Some of these processes are, for example,

that our heart size increases,
our heart rate decreases,

our bones tend to shrink in size,
and our muscles weaken.

But what's causing these changes,
you know, during the ageing process

has been the subject
for many, many years now.

And it's only recently come to light

that DNA methylation plays a major role
during the ageing process.

Now, DNA methylation is a chemical process

where small caps are being put on our DNA
in order to activate or inactivate genes,

thereby changing
the characteristics of the cell.

It's a natural process
that occurs in our body.

Now, why DNA methylation changes
during ageing, we're not sure yet,

but what we do know
is that our environment -

and by that I mean what we drink,
what we eat, whether we smoke,

whether we have much stress
to endure, for example -

has an effect on these
DNA methylation patterns.

So, when we get older,

the more of these environmental effects
affect our DNA methylation patterns,

they change,

and that's something we can measure.

Now, in order to develop
a test to estimate age,

we looked at the DNA methylation patterns
in four regions of our DNA

in over 200 individuals with a known age.

So, with that data,
we were able to develop a model

that we can now use
to predict the age of people.

If you can tell from this graph,

we've got the predicted ages
plotted against the actual ages.

You can tell that
they're very, very similar.

So, we can use this data now,
basically this model,

to estimate the age of people
with an unknown age

and this with an accuracy
of less than four years.

That's a five-fold difference compared
to the more traditional techniques,

such as anthropology and odontology.

We're basically calculating
someone's biological age or epigenetic age

to determine his actual
or chronological age.

Now, what can we do with this test,
which we've initially called DNAge?

But I've just recently found out
a couple of days ago

that DNAge is actually a very well-known
anti-ageing, moisturising cream,

so forget the name.

(Laughter)

We can estimate the age
of deceased individuals

just as in our burned case
that Wim presented.

But we can also estimate
the age of living individuals,

people who claim to be older or younger
than their official documents tell you.

Very importantly for forensics,

we can also estimate
the age of bloodstains of people

who left bloodstains at crime scenes,

you know, to help identify the offender.

And intriguingly,
you can also use this test

to prove that a bloodstain was,
for example, left seven years earlier,

prior to when the crime was committed

and prove the suspect innocent
of a crime he's accused of.

Now, because these
DNA methylation patterns

are very, very stable,

you can also use them in cold cases -

cases that have been left on the shelf

for several years,
perhaps 30 years, for example,

and can be solved right now.

So, back to our case.

We confirmed through
regular DNA profile he's male.

We know through our
biogeographical ancestry test

that he's most probably
from Southeast Asian descent.

And we now know
that he's about 36 years old,

give or take two and a half years.

That's a lot more information

that the police can now use
to identify the suspect.

So we have one more issue left:

time of death.

Now, the time of death

has always been sort of the Holy Grail
in forensic medicine

because it provides useful information

to verify alibis
from suspects, from witnesses,

but also the timelines
of the victim itself:

you know, where was the victim
prior to his death, who was he with?

So the more accurate
your time of death estimation is,

the better these timelines will be.

Now, everybody who's ever seen, you know,

CSI series, or series
such as 'Silent Witness' -

and to be honest, I love that show -

you know these three standard parameters
that forensic pathologists use:

pooling of red blood cells
towards the lower parts of the body,

muscle stiffening

and cooling down of the body temperature.

Now, what's most of the times
left out in the series

is that this method
has very large error range.

Estimating the time of death
is very challenging

because it's dependent
on a large number of external variables,

such as the type and the number
of layers of clothing

that a person was wearing,

the temperature of where the location was
where the victim was lying

or whether that person had a fever
due to an overdose of drugs.

Now, in this case,
in our burned-victim case,

Wim estimated the time of death
at about two to three days ago.

And that's pretty vague.

And many, many researchers

have already tried to improve
the time of death estimation.

But we've come up
with a completely novel method

which is complementary to the methods
used by the forensic pathologists.

And we're using information
from our own biological clock.

Now, our biological clock

is a 24-hour pattern of behavioural,
physiological and biological aspects

which is evolutionary conserved.

It's the annoying thing that wakes you up

those five precious minutes
right before your alarm clock goes off.

Why?

Or the annoying thing, you know,

why your wife falls asleep
right next to you

after you just started
that fantastic movie with Mel Gibson.

Anyway, so the biological clock

plays a large number
of important physiological roles.

And two of well-known hormones,
called cortisol or melatonin,

they fluctuate in this 24-hour pattern,
and they're part of this biological clock,

and they peek at different
times of the day.

Melatonin, for example -
it's the sleep inducing hormone -

it peeks during the night,

while cortisol - it's the stress hormone -
it peeks early morning.

Now, when a person dies,

all the biological processes
in his body stop.

And so does the biological clock.

So, when we take a blood sample
from that person,

it provides us with a unique snapshot
of exactly the time that person died.

So, when we measure the concentrations
of cortisol and melatonin,

we can indeed have already an idea
of when a person died:

whether early morning,
midday, evening or night.

But we wanted to improve on that.

We wanted to improve
the accuracy of this test

to within two hours
of the actual time of death.

And we're looking at identifying
more of these rhythmic markers,

such as RNA markers,
which are the products of your DNA,

or metabolites, which are
the products of our metabolism.

And we're calling it
'our forensic molecular clock'.

Now, with our three ideas,

we're now able to determine
the exact cause of death.

We've got a lot more information
to identify our victim,

and we now know the exact time of death.

But we're only beginning to get
a grasp of the ideas, of the information

that's still available in bloodstains
or other biomolecules

that are still available
on crime scenes, for example,

but it's very difficult to predict
what future of forensics will be.

What we do know, however,
is that we're not at the end of our means,

but only at the beginning
of what nobody could've imagined before.

We're basically turning
fiction into reality.

Thank you.

(Applause)