Dynamic imaging of motion problem

[RT steven]

Often in the hospital setting, radiographers are asked to go up to theatre and perform real time imaging of patients for having tumours removed or drilling away some sort of object that may blocking a vessel. The mobile x-ray machine is an image intensifier (II) which is coupled to a TV which allows a surgeon to dynamically view internal anatomy. Vessel patency can be tested by injecting radiolucent contrast and observing its dynamic flow.

As a radiography student, in this research project, i aim to test the precision at which the image intensifier (II) may resolve moving objects.

My objective is to try and think up some very simple experiments which enable me to test the the machine and i will need to explain the precision of the machine in terms of contrast on the screen (distribution of black an white pixels over an edge), how fast an object can move during image acquisition before it becomes too blury.

There are 2 image acquisition modes i will use: single pulse, sequence pulse. In fact, a single pulse is not instantaneous and requires t time to generate an image on the screen. In sequence pulse, the machine make take 7 images per second.

Some ideas i have thought up

dyamic

simple pendulum experiemnt, using simple physics (and a piece of lead tied to a string), i will perform a single pulse, and time how long it takes for the image to appear (perhaps 1 second), then measure the blur over distance somehow on the image -any ideas? I will also do this with a sequence pulse.

my main questions are.. how am i to quantify these results? Can i graph the sharpness as a function of motion speed? how may i quantify sharpness. how can i determine the speed of the object if i know the period of the pendulum, and the blur distance it has moved for a given time for image aquisition. From this, can i use this as a calibration excercise to find out what is the max speed something can move for a moving object to be imaged correctly.

Basically, each pulse has an integration time, which is an average oer distance of the object moving for when the image is acquired.. how can i work this exposure time out? can i work out what this integration is?

My problems is, i can think up the experiments, but I am unsure how to quantify the results (or the physics calcultions involved) and to present them as statistical data.

My ultimate goal is to image flow of contrast in synthetic tubes (to simulate arteries) and see the effect of flow mixing with water and investigate the profile of the image as a function of tube diameter and other variables. I will however need calibration details of the II (obtained hopefully from the above experiments) in order to make calculations for this fluild dynamics model.

your help would be much appreciated!

steve

 

[imabug]

simple pendulum experiemnt, using simple physics (and a piece of lead tied to a string), i will perform a single pulse, and time how long it takes for the image to appear (perhaps 1 second), then measure the blur over distance somehow on the image -any ideas? I will also do this with a sequence pulse.

what you need is something that moves at a known velocity, which you will then acquire images of. if there are any diagnostic physicists working in the hospital where you are doing your clinical work (I assume you have started that phase of your training), ask them if they have any kind of spinning disk test tool you could borrow. One of your instructors may also know where to find one. A pendulum will work, but you will have to perform additional measurements to determine it's frequency and verify with your calculations. Velocity of the bob at different points of the arc will differ though. That will complicate your image acquisition.

my main questions are.. how am i to quantify these results? Can i graph the sharpness as a function of motion speed? how may i quantify sharpness. how can i determine the speed of the object if i know the period of the pendulum, and the blur distance it has moved for a given time for image aquisition. From this, can i use this as a calibration excercise to find out what is the max speed something can move for a moving object to be imaged correctly.

Measuring the amount of blurring should be sufficient. It will be a somewhat subjective measurement, since you are trying to find the edge of a blurry object (what's that?).

Given the velocity of the moving object, you can also calculate the pulse length of the x-ray pulse.

Measuring the contrast of the moving object relative to the background at different velocities will give you an indication of the maximum velcoity an object can have to be resolvable. The faster the object is moving, the more blurring there will be and consequently the less contrast it will have.

 

[RT steven]

imabug, thankyou for your response!

Basically as far as rotating discs or diagnostic phantoms are concerned (for an II), i don't think they quite exist -i guess otherwise my project would be pointless, nevertheless, i could probably image some toy with a slow rotating propeller and stick a piece of lead on the end of it and let it go around in a circle, -i thought of a clockhand, but then that's too slow.

say i know the revolutions per second on this propeller, is that analogous to the period of the piece of lead? with the period, what calculation can enable me to find the velocity? [i really need direction formula wise]

Having obtained this velocity, and applying a single pulse, measuring the distance moved (by the commet tail blur) should give me the pulse time? am i correct in saying this is analogous to the integration time of the machine? (ie the average image of all the action produced over the pulse time) [ie t = d/s]

i guess you are correct that sharpness is a subjective measurement. I plan to use a line pair phantom (phantom that tests the spatial resolution of an xray unit) and find out the smallest resolvable object size for the II. From this, perhaps use it as a benchmark for comparison of the sharpness to be measured on the II images. Or maybe in a more practicle sense, find out what the average patient motion (internal bowel movement etc) is and set me benchmark there -there is nothing we can do about patient motion -the equivalent of noise in an xray image.

To measure the amount of blurring, is it wise perhaps to take the diamter of the object and make it fraction over the commet tail seen on the image? is that a wise consistent relation to measure? I can probably use these same images to do a contrast profile and record the range of contrast seen relative to the background .. great!

I guess then, contrast is related to blurring is related to speed of the object. Perhaps ill graph a contrast/sharpness vs. speed. In a nutshell, there is a variance in velocity of a pendulum at different points so from a graph here, i could probably find the max velocity for an object to be resolved. i think i need help with these calculations, do you have some formulas that will help me -im thinking energy equations ie mgh/2 = mv^2/2 but i can perform these without a pendulum, and i need to relate my images to the formulas i use. you mentioned something about freuqncies and periods .. is there a way?

finally, in terms of the sequence pulse (which might be wiser to use for measuring different velocities), the pulse time would be different to that of a single pulse, is that correct? it would have to be? ie, a single pulse might be 1 sec, but in succession, a sequence may take 7 images per 3 seconds, therefore each pulse is 3/7s and ill need to tie that into my calculation then.

 

[FredGarvin]

Along those lines, why not use a solid disk and have different targets spaced out on varying radii. That will give you multiple targets on one test item, all with varying velocities, but the same angular velocities. You would then have a way to easily make subjective comparrisons.

 

[RT steven]

thanks fredgarvin and imabug for your suggestions and adivce, I am in the middle of testing right now, hope to have some results soon
steve

 

[RT steven]

well, I've done some measurements, these are not precise, but just some basic simple measurements to get an idea of the dynamics of the II.

METHOD: on a rotating disc (1 rev per 60 seconds), placed a few drawing pins along a radius at varying distances and took successive pulses at 1 frame per second over 7 seconds. I obtained 7 images using a still digital camera positioned at a fixed distance from the screen (for reproducibility) and analysed these.

I superimposed 4 of these successive images using some photo editing software to do some measurements.

http://us.f3.yahoofs.com/users/420f5865z26b08ab1/77ae/__sr_/3c6c.jpg?phk_1DCBa.obN14H

I guess from the individual frames, i can do a pixel profile along the edge of each of the pins and see the amount of blurring per velocity.

Am i correct in saying that each frame taken is an integral of the period that the pin took as it moved between frames, ie 1 second? Or is the integral image actually the marginal distance of blur seen for the pin edge?

I have a question, is teh frame rate directly proportional to the velocity of a moving object. ie, 1 frame per second for an object moving at v gives me x blurring, so 2 frames per second for an object moving at 2v should still give me the same amount of x blurring?

what other information could i possibly yield from this experimental set up?