Hi Martijn,
That is very strange.  I don't know why the database would have an empty table. 
 It is acting like snomed codes were not found for any of your cuis in your 
local umls installation, but that is terribly unlikely.

I ran the gui about two weeks ago and the secondary database tables were 
populated.

Sorry that I can't help, this is unfortunate,
Sean

-----Original Message-----
From: Martijn [mailto:mgkersl...@uvic.ca] 
Sent: Tuesday, March 14, 2017 12:07 PM
To: dev@ctakes.apache.org
Subject: Re: 2016AB UMLS (ctakessnorx)

Hi Sean,

It’s an array, but I only sent you a single item out of that array.

The .xml file lists the snomedct_us table, but the database is empty. I’m sure 
that I selected SNOMEDCT_US in the dictionarygui tool. Do you have any idea why 
the database could be empty? 

Thanks.

- Martijn

On 10/03/2017, 10:33, "Finan, Sean" <sean.fi...@childrens.harvard.edu> wrote:

    Hi Martijn,
    
    The UmlsConcept should be in an array in the IdentifiedAnnotation.  Does 
your array only contain a single concept?
    
    When you use the gui, it should store codes in the database as a unique 
table for every (target) vocabulary that you selected in the left panel.  The 
.xml file that it creates should list all of those types in the 
<conceptFactory> section.  In your .xml you should see the line:
             <property key="snomedct_usTable" value="long"/>
    
    If that line is in the .xml, you can inspect your database directly with a 
hsqldb tool.  I can help you do that if needed.  The db should have a table 
named "snomedct_us".
    
    If all of those are ok then I will need to look at the lookup code as 
something must have broken.
    
    Sean  
    
    -----Original Message-----
    From: Martijn [mailto:mgkersl...@uvic.ca] 
    Sent: Thursday, March 09, 2017 5:35 PM
    To: dev@ctakes.apache.org
    Subject: Re: 2016AB UMLS (ctakessnorx)
    
    Hi Sean,
    
    Thanks! I used the GUI to generate a custom dictionary, but I still get the 
UMLS code and not the SNOMED CT one.
    
    If I print the concept that was detected it returns:
    UmlsConcept
       codingScheme: "custom"
       code: <null>
       oid: "null#custom"
       oui: <null>
       score: 0.0
       disambiguated: false
       cui: "C1281583"
       tui: "T023"
       preferredText: "Entire hand"  
    
    As you can see, the code is <null>. The concept is present in the UMLS 
subset and there is also a SNOMED CT code listed there:
    
    Entire hand [A3421866/SNOMEDCT_US/PT] CUI:C1281583 SCUI:302539009
    
    
    Am I doing something wrong?
    
    - Martijn
    
    On 08/03/2017, 08:12, "Finan, Sean" <sean.fi...@childrens.harvard.edu> 
wrote:
    
        Hi Martijn,
        
        The dictionary creator gui is in sandbox just like the command line 
tool, but it is newer and easier to use.
        
        OntologyConceptUtil is in org.apache.ctakes.core.util.  
        
        Sean
        
        -----Original Message-----
        From: Martijn [mailto:mgkersl...@uvic.ca] 
        Sent: Tuesday, March 07, 2017 5:38 PM
        To: dev@ctakes.apache.org
        Subject: Re: 2016AB UMLS (ctakessnorx)
        
        Hi Sean,
        
        Thanks so much for your quick reply.
        I used the command line directorytool. Is that different than the gui? 
Can that explain the decrease in tagged concepts?
        
        I’m not able to find the OntologyConceptUtil class, may I ask what the 
path for that class is?
        
        - Martijn
        
        On 07/03/2017, 14:24, "Finan, Sean" <sean.fi...@childrens.harvard.edu> 
wrote:
        
            Hi Martijn,
            
            Since you say that you've created your own dictionary I will assume 
that you used the gui in sandbox to do so.  If that isn't the case then let me 
know.
            
            The any dictionary created using the default settings on the gui 
does have snomedct and rxnorm codes in addition to the cuis.  However, umls cui 
is always used as the primary normalization code for ctakes annotations.
            
            To obtain codes for an annotation, check the OntologyConceptUtil in 
ctakes core.  It has methods that will return all associated codes as well as 
one to get all associated codes for a scheme/vocabulary (like snomedct_us, 
etc.).  It can do this for a single annotation, a collection of annotations, 
the entire document, or a section of the document (sentence, paragraph, 
section).  It also has methods that allow you to fetch annotations found in the 
document by codes other than the umls cui.
            
            Sean
            
            
            -----Original Message-----
            From: Martijn [mailto:mgkersl...@uvic.ca] 
            Sent: Tuesday, March 07, 2017 5:13 PM
            To: dev@ctakes.apache.org
            Subject: 2016AB UMLS (ctakessnorx)
            
            Hi,
            
            I've been using cTAKES for a bit now, but I still can't figure out 
how to upgrade the UMLS version to the most recent one.
            If I create my own dictionary, cTAKES only returns UMLS concepts 
and no SNOMED CT ones (I'm interested in those). The amount of concepts 
returned is also way less compared to the 2011 UMLS that's included with cTAKES.
            
            Can someone help me out by providing me a proper 2016 dictionary or 
clear explanation how to implement the newest version of the UMLS (with SNOMED 
CT).
            
            
            Thanks!
            
            - Martijn
            
            
        
        
        
        
        
    
    
    


Reply via email to