Plexar develops effective and efficient algorithms for
solving problems with
image (2D) and signal (1D) data.
The following paragraphs describe some of the
solutions
achieved using Plexar’s algorithm development strength.
Endoscopic Imaging Enhancement
From a clinical perspective, it is good for an endoscope to be thin and to be
flexible. Flexibility is achieved with the use of a fiber optic bundle instead
of a glass rod. But with a fiber optic bundle, the image is broken up into
“optical pixels” that cause a mesh pattern artifact in the images.
Also, with a thin fiber optic bundle there is also loss of visualization
in areas of low light and/or saturation in areas of high light.
Plexar worked closely with it's client to developed a family of
algorithms that provide effective solutions to these endoscope problems.
The algorithms remove the mesh artifact while preserving the image
detail and are “field adaptable” for a variety of fiber bundle configurations.
The algorithms also dynamically adjust to the inherent
light levels, reducing the loss of visualization in regions of
low and high light levels. The algorithms operate in “real time”
and have been shown to be effective in a variety of clinical situations.
Here the goal was to produce images using laser light. The measured projections are,
of course, highly scattered or blurred and the amount of blurring depends on the depth
of the object. Standard reconstruction methods generate blurred and distorted objects.
And, in the presence of noise, standard deblurring methods do not give satisfactory results.
Plexar developed a depth-dependent deblurring algorithm to provide an effective solution.
Our algorithm, adapted form the Van-Cittert iterative constrained deconvolution algorithm,
corrects most of the blurring and distortion and is stable in the presence noise.
Our implementation includes fine-tuning parameters that make it widely applicable.
Because x-ray scintillation crystals suffer from the phenomena known as “carrier trapping,”
an afterglow signal is produced. The afterglow looks like a low pass filter with
multiple time constants. The afterglow effects image quality in two ways.
The first way is that it distorts the image, particularly in regions of low x-ray intensity.
The second degrading effect occurs when trying to correct for the effect by high pass
filtering, resulting in significant amplification of electronic noise. Plexar has developed
an adaptive afterglow correction algorithm that removes the distortion with only a
minimal amplification of noise. That is, because the algorithm is adaptive,
it applies its high pass filter only when necessary and only when the electronic noise
is small relative to the x-ray quantum noise. Plexar’s algorithm is configurable for a
variety of types of scintillation crystals.
For more detail and sample results, click here. (coming soon)
Static Discharge Spike Elimination
In TLD measurement devices, the glow measurement is often corrupted by the presence of
static discharge spikes. Even though the spike duration is very short, its amplitude is
very high; therefore, the presence of even a few spikes can cause the measurement of an
average glow value to be high by a factor of 10!
Plexar has developed an algorithm that effectively eliminates the effect of the
static discharge spikes. With this algorithm, accurate glow values are determined with
or without the presence of spikes. The algorithm’s accuracy is not degraded by the
presence of quantum noise in the glow signal.
And it can be configured for a wide variety of TLD devices.
One of Plexar’s client developed a successful method for assessing the capability of
a job candidate in the context of the strength requirement of the job.
The job candidate undergoes a series of safe, isokinetic tests of major muscle groups
generating a set of isokinetic torque curves. Using the result of a statistical formula
applied to the torque data, the staff technicians determine an objective numerical rating
that reflects the strength capability of the worker. In order to increase throughput,
Plexar developed a rule-based algorithm to automate the rating process. Because the
population of workers does not completely conform to any single set of rules, the algorithm
included special rules that identified a case as one requiring human evaluation.
The rule-based algorithm provided a 2-times increase in throughput.
Plexar’s algorithm implementation includes a “what if” capability that easily uses
historical case data to test possible new rules.
Algorithms by Plexar
