We want to perform segmentation and volume calculation of cerebral infarcts from CT and/or MRI. Does anyone know of available tools for such segmentation?
There are many publications on this topic, but few have released their code. By contrast, the FSL group will very shortly be releasing a new module dedicated to automated segmentation and measurement of cerebral infarcts. The module is called BIANCA and the published article describing it can be downloaded here: https://www.ncbi.nlm.nih.gov/pubmed/27402600
An alternative that is not as nice and fully automatic as the one described by Jerome, but it would probably work well for you and available now.
3D Slicer's new "Segment editor" module contains tools for manual and semi-automatic segmentation of structures. For example, you could try to use "Auto-complete" effect with "grow from seeds" method - it allows you to create a full 3D segmentation from just a few strokes inside and outside the infarcted area. You can then compute volumes using "Segment statistics" module. 3D Slicer is free and open-source, make sure you use a recent nightly version - available for download at www.slicer.org.
We are not looking for the possibility to segment WMH, but cerebral infarcts. Many of the tools for segmentation of WMH applies prior knowledge of WMH location which is a disadvantage for us since cerebral infarcts does not have a common location like WMH (correct me if I am wrong here).
Maybe the 3D Slicer is the way to go for semiautomatic segmentation by a radiologist or something like that.
Thanks again and please correct me if I have not completely understood your suggestions :-)
Built-int tools in the 3D Slicer's segment editor are not specific to any particular anatomy (they just work based on assumptions such as contrast difference, smooth shape, etc).
Let me know if you have any questions about how to use the module for your images, I can give a quick intro using screen sharing.
Another very powerful semi-automatic tool is itk-SNAP (http://www.itksnap.org/), which is even able to segment based on multi-modality images simultaneously (T1, axial T2, coronal T2 and DTI, for example). We used it to segment neonatal infarction succesfully in the article linked below, which contains a brief explanation of the methodology we followed.
Article Three-Dimensional Map of Neonatal Arterial Ischemic Stroke D...
Yes, ITK-Snap used to be the best free open-source manual/semi-automatic segment editor for years. It's very polished and fast.
I'm one of the developers of the Segment Editor in 3D Slicer, so I may be biased, but I think we managed to develop a segment editor that rivals ITK-Snap and for certain tasks it can even replace expensive commercial software, such as Mimics. Highlights:
segments may overlap with each other: for example, if you segment a lesion in an organ, you can still keep your organ segment intact, the tumor won't be subtracted from the organ's segment; you can import DICOM RT structure sets without information loss
edit segments in 3D views: remove parts, split segments, etc.
add/subtract/intersect segments (very useful for 3D printing, you can combine anatomy with parts designed in CAD software)
apply joint smoothing to segments that respects interfaces between segments (segments that touch each other will not be pulled apart by the smoothing)
use any segment as mask for editing other segments
create complete 3D segmentation from partial segmentation: all you need to paint a few strokes in a few slices and grow complete segments from that; or segment a couple of slices and interpolate between
We've created a small tutorial demonstrating how to create an needle insertion training phantom by combining CT scan of a patient with CAD designed parts:
Slicer's segment editor is quite new, not very polished yet, but we are constantly improving usability and add new features. We would love to hear feedback from the community and improve our software based on that.