[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: [dinosaur] Main goals in future dinosaur paleontology



On 04/13/2017 04:41 PM, Mike Habib wrote:
> Sure, basic CT is an old tech, but things like nano-CT are relatively new. 
> The main novelty, however, is the prevalence of imaging in paleontology. CT 
> imaging may be old, but it hasn't been widely available to paleontologists 
> until (comparatively) recently.
> 
> --MBH
> 
> Sent from my Cybernetic Symbiote
> 
>
generally, i agree, and minconization of hardware has been mentioned in
discussions I've had.

Much of the debate relating to phylogeny involves the application of
static data which is processed by trained observers.  This is where the
field has been at for over twenty years. Field workers find bones in
rock, and highly trained preparers try to separate the rock from
mineralized bone.  Then paleontologists measure, scan and draw the
specimens, dividing observations into what is considered important
differentiation which are then processed through computational
algorithms, not much different from what is described above.  But what
has happened over the last 20 years has been nothing short of
astonishing.  Soft tissue and soft tissue remnants have been found in
Cretaceous era bone beds, and even Jurassic age fossils like the famous
German specimen of Archaeopteryx have soft tissue traces in the
surrounding rock.  Hundreds of feathered dinosaurs have been found in
China, and now that people are looking, such specimens are now being
found the Canadian Rockies, the Gobi Desert and even in the collections
standing in museums worldwide.  Proteins have been found in Tyrannosaur
bone tissue, and several Dinosaur mummies have been found which have
crops and soft tissue.  We have extrapolated the color of feathers from
melanosome in feather impressions.  One thing has become very obvious,
tearing the fossils from the rock is not always a good idea. Within the
rock surrounding bones is a huge amount of soft tissue information.
Scanning technology is expanding, and the use of CT Scans, and
Florescence techniques are literally shedding a whole new life  to
paleontological research.  Brain cases are a now a wide area of study
and the dream is to go from encased rock to 3d graphics printer, without
being destructive to the specimen.

What we need to develop, as time moves on, is the ability to not only
scan rock, but also identify likely fossils and attempt to classify them
by artificial intelligence, and remove much of the human factor. Can
this be done?  Yes, and within a few short years we should have such
technology ubiquitously available.  And with worldwide databases of
fossils, we should be much better and developing phylogeny maps of
species and remove much of the human bias that now plagues the field.

If we CAT scan a rock with a fossil within it, how can we identify the
fossil?  This is an enormously important question which has an ethical
wrinkle.  Most CAT scans are using proprietary graphics formats.
Getting specimens into an international database is going to require
ending the dependency of these private formats and to get them into
formats based on international copyright and patent free formats, that
is copyleft formats.  The dependency on proprietary file formats for
scanning equipment is the biggest impediment of future software
development for the community.  The effort to go from CAT scan to the 3d
modeling program, Blender, is just too difficult and even the legality
of tools is questionable.  Development of artificial intelligence is
going to depend on patent free, and royalty free international standards
which allow for reading and manipulation.
>From a technological perspective, consider the depth of the problem.
Even the simplest graphic recognition problem requires considerable
thought and application of theory.  How do we discover and identify, for
example, a black circle on a white sheet of 4 x 4 inch paper?  Engineers
have been grappling with this problem for decades and its solution is
the cornerstone for a broad swath of computational problems from
biometrics, driverless cars, facial recognition, and fossil evaluation.

The usage of image recognition software for evolutionary biology is one
of the largest areas of advancement and opportunity in the coming
generation of computational research and expertise.   What drives this
is the technological breakthroughs in the scanning technologies, which
are getting reduced in size, cheaper and increasingly available to
museums and research labs.  A few years ago a dinosaur mummy with soft
tissue intact was discovered in the Dakotas.  It was termed “Leonardo”
and was scanned with several tools as described by its website:1
What kind of Xrays were used to look inside the fossil?

To quote their own website:

   The science team used several different types of radiation to
   create the Xrays of the dinosaur mummy.  The first they used
   was a water-cooled Xray tube.  This is similar to what is used to
   take Xrays of people, only it is about 5,000 times more powerful. 
   This kind of power was needed because it had to look through more
   than two feet of rock.  Even this wasn’t powerful enough, so the
   team took Leonardo down to Houston, Texas were NASA made room for
   him in a special lead-lined jet hangar at the Johnson Space Center. 
   While there, the science team used Iridium and Cobalt radiation to
   look inside the fossil.  ...  The yellow container holds a small but
   highly radioactive isotope.  The Cobalt in particular was extremely
   powerful.  The science team had to leave the building and hide
   behind another building to be certain they were safe from the
   radiation.  However, the most powerful radiation used in the study
   of the fossil was used on Leonardo’s arm.  A Linear Accelerator
   fired radiation at the fossil at Stanford University’s SLAC
   (Stanford Linear Accelerator Center) also known as the National
   Accelerator Laboratory.  Yes, Leonardo is important enough of a
   fossil to have these kind of resources used in the research!

   How was the digital image of the fossil created?

   The Dinosaur Mummy was the first dinosaur to be replicated in an
   entirely new way.  Because the skin on the surface of Leonardo was
   so fragile, if the mold necessary to make casts, or replicas, of the
   fossil had been created in the same way it has been done for many
   years, the skin would have been damaged or destroyed.  This made the
   traditional method of pouring silicon rubber on the fossil and
   peeling it off out of the question.  The science team had to come up
   with another way.  The answer was a new technology called White
   Light Scanning.  A grid of light and shadow was projected on the
   fossil and special cameras recorded this grid and the surface of the
   fossil in 3D.  The resolution of the cameras is thinner than a human
   hair, so that the digital information they provided shows every fine
   detail of the original fossil. ...  This information was then used
   by a computer to control a machine called a rapid prototyper that
   was able to make a perfect replica of the dinosaur mummy without any
   possibility of damaging the original fossil.

They needed to ship the entire fossil, which weighs tons, from the
Phillips County Museum, in Montana, all the way to Texas to do scans.
In March of 2017, Don Brickman of the Royal Tyrrell Museum, in
Drumheller Calgary, Alberta told me that they had set up florescence
equipment in their labs to do 3d imaging.  This is a complete
breakthrough in technology and we can expect that in the future that CAT
Scanners will likely become standard equipment in most labs.  With the
discovery of so many soft tissue remains, the days of tearing up fossils
is largely over without first doing analysis of the possible soft tissue
remains inside the surrounding rock.

In 2011, SLAC did a wonderful lecture on the analysis of Archaeopteryx.
The presentation done by Uwe Bergmann walks through several types of
images, showing trace materials.  But this leaves us with several
problems, the first being, what kind of image formats are being used to
image fossils in CAT scanners and other scanning technologies.  How can
we prepare these images for analysis and for the analysis by artificial
intelligence. There are no standards.  In correspondence with Ramon
Nagesan, the PhD resident at the Royal Tyrrell, he explains the
difficulty of currently working with CAT Scanners data.  He wrote,

   "The file type that is output from a CT scanner is a "diacom"(sic),
   these need to be converted to either a jpeg or tif in a diacom(sic)
   converter, and then processed into an image stack in a software
   called "ImageJ". These image stacks are then uploaded to "AMIRA"
   (FEI) for processing into 3D models. I'm curious to see if you'll be
   able to streamline this process."

Dicom is a huge copyright problem promoted as “the internet of medical
information”.  According to its standards page,  “It defines the formats
for medical images that can be exchanged with the data and quality
necessary for clinical use. “3  It is controlled by The Medical Imaging
& Technology Alliance (MITA), a division of the National Electrical
Manufacturers Association (NEMA).   They claim to be the leading
organization and collective voice of medical imaging equipment
conglomerate.4  They claim to represent companies whose sales make up
more than 90 percent of the global market for advanced imaging
technologies.  They are not an open forum for digital standards but are
created as a profit center for a few select and well placed medical
device companies.

There are, nevertheless, some free tools to access some of their
formats.   On the sourceforge network there is Open Dicom Viewer5 which
is licensed under GNU Library or Lesser General Public License version
3.0 (LGPLv3).  It is a Java application (written in Java 1.7).
Another tool is  DCMLinux 6, and it is released under the GPL2.  It
promotes itself as, “a complete PACS system, free of charge. Its core is
an Ubuntu 10.04 system fully updated and it contains the DCM4CHEE as its
PACS server. In the near future it will contain many other addons such
as Weasis, Oviyam, Care2x, etc.”

There is an “OpenSourced” C++ library for dicom with java and python
hooks which looks promising for new developers.  Called Imebra7, and
licensed under the GPL2, and it’s development is current.  “The Imebra
SDK is a multiplatform, open source, C++ library for handling DICOM
files, both raw and compressed.”  It actually has an impressive set of
features and looks like it is worth development.  It’s main class is the
imebra::DataSet class and it is fully documented.  It is compiled with
cmake.  You need to include  imebra/imebra.h, to your source files to
access the library.  It uses uniqueptr and can open files like this
according to the official documentation:

std::unique_ptr<imebra::DataSet>
loadedDataSet(imebra::CodecFactory::load("DicomFile.dcm"));
// Retrieve the first image (index = 0)
std::unique_ptr<imebra::Image>
image(loadedDataSet->getImageApplyModalityTransform(0));

// Get the color space
std::string colorSpace = image->getColorSpace();

// Get the size in pixels
std::uint32_t width = image->getWidth();
std::uint32_t height = image->getHeight();

// let's assume that we already have the image's size in the variables
width and height
// (see previous code snippet)

// Retrieve the data handler
std::unique_ptr<imebra::ReadingDataHandlerNumeric>
dataHandler(image->getReadingDataHandler());

for(std::uint32 scanY(0); scanY != height; ++scanY)
{
    for(std::uint32 scanX(0); scanX != width; ++scanX)
    {
        // For monochrome images
        std::int32_t luminance = dataHandler->getSignedLong(scanY *
width + scanX);

        // For RGB images
        std::int32_t r = dataHandler->getSignedLong((scanY * width +
scanX) * 3);
        std::int32_t g = dataHandler->getSignedLong((scanY * width +
scanX) * 3 + 1);
        std::int32_t b = dataHandler->getSignedLong((scanY * width +
scanX) * 3 + 2);
    }
}

Dicom has a networking layer in the specification, something that PNG
also made room for in its specification.  This is a template for views
an image:
 // The transforms chain will contain all the transform that we want to
// apply to the image before displaying it
imebra::TransformsChain chain;

if(imebra::ColorTransformsFactory::isMonochrome(image->getColorSpace())
{
    // Allocate a VOILUT transform. If the DataSet does not contain any
pre-defined
    //  settings then we will find the optimal ones.
    VOILUT voilutTransform;

    // Retrieve the VOIs (center/width pairs)
    imebra::vois_t vois = loadedDataSet->getVOIs();

    // Retrieve the LUTs
    std::list<std::shared_ptr<imebra::LUT> > luts;
    for(size_t scanLUTs(0); ; ++scanLUTs)
    {
        try
        {

luts.push_back(loadedDataSet->getLUT(imebra::TagId(imebra::tagId_t::VOILUTSequence_0028_3010),
scanLUTs));
        }
        catch(const imebra::MissingDataElementError&)
        {
            break;
        }
    }

    if(!vois.empty())
    {
        voilutTransform.setCenterWidth(vois[0].center, vois[0].width);
    }
    else if(!luts.empty())
    {
        voilutTransform.setLUT(*(luts.front().get()));
    }
    else
    {
        voilutTransform.applyOptimalVOI(image, 0, 0, width, height);
    }

    chain.add(voilutTransform);
}

// If the image is monochromatic then now chain contains the VOILUT
transform

// We create a DrawBitmap that always apply the chain transform before
getting the RGB image
imebra::DrawBitmap draw(chain);

// Ask for the size of the buffer (in bytes)
size_t requestedBufferSize = draw.getBitmap(image,
imebra::drawBitmapType_t::drawBitmapRGBA, 4, 0, 0);

// Now we allocate the buffer and then ask DrawBitmap to fill it
std::string buffer(requestedBufferSize, char(0));
draw.getBitmap(image, imebra::drawBitmapType_t::drawBitmapRGBA, 4,
&(buffer.at(0)), requestedBufferSize);

Likewise there are copyright and patent issues with 3d printers.   A
recent visit to Makerbot, the makers of 3d printers who’s offices are
located  at 1 Metrotech Plaza, in Downtown Brooklyn, proved that they
have a rather large Tyrannosaur 3d model that they printed hung right
next to the receptionist desk.  They have intense interest in getting
scans for printing and they are using, according to the presenter, Drew
Lentz, STL files ( StereoLithography ) and the Wavefront .obj file. 8
The good news about these formats is that they are both are open and
adopted by numerous free and proprietary software solutions.  So
everyone is interested, but these systems are generally not talking to
each other.  CAT Scans are the biggest  problem as their graphics
formats are different from machine to machine and all are proprietary,
even secret formats at times.

Despite this, there is a growing number of CAT Scan and images in
archives which have been created and which are publicly available.  The
University of Texas High-Resolution X-ray Computed Tomography Facility,
has DigiMorph.Org, which is, “part of the National Science Foundation
Digital Libraries Initiative, develops and serves unique 2D and 3D
visualizations of the internal and external structure of living and
extinct vertebrates, and a growing number of 'invertebrates.'  The
Digital Morphology library contains more than a terabyte of imagery of
natural history specimens that are important to education and central to
ongoing cutting-edge research efforts.  The Digital Morphology library
site now serves imagery, optimized for Web delivery, for more than 1000
specimens contributed by almost 300 collaborating researchers from the
world's premiere natural history museums and universities.”  Of course,
copyright problems hinder this library, and it can benefit from a
copyleft license.