| Re: I need to bulid a digital phantom urgently, Thanks for help! [message #84030] |
Mon, 22 April 2013 23:23 |
huiqiang.liu.37
Messages: 19
Registered: April 2013
|
Junior Member |
|
|
Dick,
Thank you for the prompt reply.
Yes, i agree with you that the 75% packing fraction is very tight with varying size spheres, just like i konwn that someone achieved this model by using the forecasting software Crystal Ball.
yes, i want to create 3D volume array of values with three types of values: background(simulating lung tissue),sphere shell(alveoli,about size of 8-12um),hollow sphere interior(air-filled). of course , if the model does work, it will have very Strong Commonability for us.
Thanks for your help.
Liu
在 2013年4月23日星期二UTC+9上午6时14分38秒，Dick Jackson写道：
> huiqiang.liu.37@gmail.com wrote:
>
>> This model is like as follows: a segment of lung tissue was modeled as an
>
>> array of randomly positioned hollowed spheres (simulating alveoli). A
>
>> 1*1*11.6 mm3 volume was created to match the thickest lung region (11.6 mm),
>
>> with simulated alveoli given a 75% packing fraction and a Gaussian
>
>> distribution of diameters. A mean diameter of 60 μm with a standard deviation
>
>> of 10 μm was used to match the known size distribution of 38–80 μm .
>
>>
>
>> Thank you so much. Liu
>
>
>
> Liu,
>
>
>
> In researching this a bit, it looks like 75% packing density is not possible
>
> with identical spheres...:
>
> http://en.wikipedia.org/wiki/Random_close_pack#For_spheres
>
>
>
> ... but with the variety of sizes as you described, it may be possible to
>
> approach that. 75% is in fact very, very tight.
>
>
>
> In any case, my first idea, to place spheres into a volume randomly (without
>
> overlap) until the volume is full enough, is clearly not going to work. I think
>
> there's no chance of getting close to this optimal packing by random placement.
>
> Perhaps someone out there has done this kind of thing before? Perhaps starting
>
> with a random set of spheres, and an optimization algorithm to have them push
>
> apart until they no longer overlap? I am reminded of something seen in data
>
> visualizations called force-directed graph drawing:
>
> http://en.wikipedia.org/wiki/Force-directed_graph_drawing
>
>
>
> These other issues (which I wrote about first) are comparatively minor!:
>
>
>
> In order to help you, I think we need more a little more information. I guess
>
> you're looking to create a 3-D volume array of values with possibly three types
>
> of values:
>
> - background
>
> - sphere shell, and
>
> - hollow sphere interior
>
> If the spheres are hollow, we also need to know how thick the shell is (or the
>
> diameter of the interior sphere), whether a constant or perhaps a fraction of a
>
> given sphere's diameter.
>
> Also necessary is a scale for the array, that is the physical size represented
>
> by each 3-D array element, or voxel. I'll assume the voxels are cubes.
>
>
>
> --
>
>
>
> Cheers,
>
> -Dick
>
>
>
> Dick Jackson Software Consulting
>
> Victoria, BC, Canada
>
> www.d-jackson.com
|
|
|
|
| Re: I need to bulid a digital phantom urgently, Thanks for help! [message #84032 is a reply to message #84030] |
Mon, 22 April 2013 14:14  |
Dick Jackson
Messages: 347
Registered: August 1998
|
Senior Member |
|
|
huiqiang.liu.37@gmail.com wrote:
> This model is like as follows: a segment of lung tissue was modeled as an
> array of randomly positioned hollowed spheres (simulating alveoli). A
> 1*1*11.6 mm3 volume was created to match the thickest lung region (11.6 mm),
> with simulated alveoli given a 75% packing fraction and a Gaussian
> distribution of diameters. A mean diameter of 60 μm with a standard deviation
> of 10 μm was used to match the known size distribution of 38–80 μm .
>
> Thank you so much. Liu
Liu,
In researching this a bit, it looks like 75% packing density is not possible
with identical spheres...:
http://en.wikipedia.org/wiki/Random_close_pack#For_spheres
... but with the variety of sizes as you described, it may be possible to
approach that. 75% is in fact very, very tight.
In any case, my first idea, to place spheres into a volume randomly (without
overlap) until the volume is full enough, is clearly not going to work. I think
there's no chance of getting close to this optimal packing by random placement.
Perhaps someone out there has done this kind of thing before? Perhaps starting
with a random set of spheres, and an optimization algorithm to have them push
apart until they no longer overlap? I am reminded of something seen in data
visualizations called force-directed graph drawing:
http://en.wikipedia.org/wiki/Force-directed_graph_drawing
These other issues (which I wrote about first) are comparatively minor!:
In order to help you, I think we need more a little more information. I guess
you're looking to create a 3-D volume array of values with possibly three types
of values:
- background
- sphere shell, and
- hollow sphere interior
If the spheres are hollow, we also need to know how thick the shell is (or the
diameter of the interior sphere), whether a constant or perhaps a fraction of a
given sphere's diameter.
Also necessary is a scale for the array, that is the physical size represented
by each 3-D array element, or voxel. I'll assume the voxels are cubes.
--
Cheers,
-Dick
Dick Jackson Software Consulting
Victoria, BC, Canada
www.d-jackson.com
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
