High resolution DVH for small volume ROI

[cutright]

DVHA v0.9.1 allows for high resolution DVH calculations. In User Settings, you can specify a small volume threshold which will trigger a high resolution calculation. You can also specify how many interpolated segments will be calculated in-between ROI planes. This is all based on built-in features to dicompyler-core.

[cutright]

@Kiragroh I think this is what you were looking for?

[Kiragroh]

Amazing. Thank you.

[cutright]

Let us know how it works? I'm a little unsure of the high resolution interpolation factor. I've only tested on 4mm dose grids, hopefully it doesn't matter.

@bastula Did I implement this correctly?

DVH-Analytics/dvha/db/dicom_parser.py

Lines 457 to 458 in d8e0ab4

	kwargs['interpolation_resolution'] = (self.rt_data['dose'].PixelSpacing[0] / self.dvh_high_resolution_factor,
	self.rt_data['dose'].PixelSpacing[1] / self.dvh_high_resolution_factor)

I allow users to specify a dvh_high_resolution_factor in [4, 8, 16, 32]

[Kiragroh]

Maybe my data is not compatible with these interpolation methods. I tried a few setting-combinations and the dvh calculation time rises but dvh stairs remain. CT-slice distance is 1mm and 1mm doseGrid.
Here a screenshot:

[cutright]

@bastula I wonder if Eclipse is also smoothing the DVH itself?

[Kiragroh]

This is a good thought. Because the dvh looks smooth in Eclipse and no long calculation is required.

[cutright]

Have you tried importing one of your plans using the Eclipse DVH, and then import again using the DVHA calculation? I'd be curious how they look plotted on top of one another.

[Kiragroh]

Eclipse does not export DVH-Infos in DICOM-Format (in RP or RD like in some other TPS). I can export a DVH-csv file. That would represent the calculated DVH in Eclipse and will be smooth if I do it via ESAPI (the standard Eclipse-DVH-Export has a fixed Bin-Width that is mostly not enough for small structures).

[cutright]

Are you sure about that? :)

[cutright]

Maybe you mean the Eclipse API doesn't offer it? I have it through the main GUI.

[Kiragroh]

I was sure till you asked.^^ I tried your suggestion and same result in DVHA. Furthermore, I checked the DICOMs again and could not find DVH-data.
In ESAPI you are free to choose a DVH resolution. In Eclipse15.1-GUI I cannot manipulate the DVH-Bin-Width before export. Or I misremember because I do not use this export anymore because the output is not well structured. It is better for post-processing to buil something with ESAPI in my opinion.

[bastula]

Hi Max, Is it possible to get an anonymized copy of the dataset (no need for CT) so I can look at why this is happening in dicompyler-core?

…

On Sat, Dec 12, 2020 at 9:05 AM Max Grohmann ***@***.***> wrote: I was sure till you asked.^^ I tried your suggestion and same result in DVHA. Furthermore, I checked the DICOMs again and could not find DVH-data. In ESAPI you are free to choose a DVH resolution. In Eclipse15.1-GUI I cannot manipulate the DVH-Bin-Width before export. Or I misremember because I do not use this export anymore because the output is not well structured. It is better for post-processing to buil something with ESAPI in my opinion. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#119 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAO7XV6WCHQEITSK4DCSPDSUOBEDANCNFSM4UWAZVFQ> .

[Kiragroh]

No problem, I send the data via eMail to Dan this morning. @cutright: can you please forward my mail.

Thanks guys for taking interest in my use-case. :)