MIT-LCP / mimic-code

MIMIC Code Repository: Code shared by the research community for the MIMIC family of databases
https://mimic.mit.edu
MIT License
2.53k stars 1.51k forks source link

Question: Is the MIMC-CXR data related to MIMIC-III clinical database? #1009

Closed vasudev-sharma closed 2 years ago

vasudev-sharma commented 3 years ago

I've a question in regards to the MIMIC-CXR data. I've been wondering is the MIMIC-III clinical database anyway related to the MIMIC-CXR? If it is, then how?

Actually, I'm interested in knowing whether vital signs, laboratory reports, and other information of the patients of the MIMIC-CXR could be retrieved or not. So, I was just curious is there a relation between MIMIC-CXR and MIMIC-III because MIMIC-III has that information, but MIMIC-CXR doesn't. That's why I opened this issue to better understand if this information could be obtained using MIMIC-III.

If there is no relation between the two datasets, then is it possible to get access to BIDMC EHR which was used to collect the x-ray images of MIMIC-CXR data?

Thanks in advance!!

alistairewj commented 3 years ago

No, MIMIC-III is entirely separate. MIMIC-IV is the database you are looking for: https://physionet.org/content/mimiciv/

The PatientID in the DICOM header of the files is the same as subject_id in MIMIC-IV. There is a convenient CSV of MIMIC-CXR metadata in the MIMIC-CXR-JPG project: https://physionet.org/content/mimic-cxr-jpg/

vasudev-sharma commented 3 years ago

Thank you @alistairewj for your insights. I can see that in the MIMIC-CXR-JPG dataset, there are certain patients ( around 11245 patients) who don't have their respective subject_id in the MIMIC-IV. For instance, in MIMIC-CXR-JPG dataset patient_id's: 12427812, 14029832 , ..... and many more, cannot be mapped back to the samesubject_id's of MIMIC-IV.

May I know for these 11245 patients can I get information from MIMIC-IV?

If these patients cannot be mapped back to MIMIC-IV data, may I know was the source of data collection for MIMIC-IV and MIMIC-CXR different for these patients?

alistairewj commented 3 years ago

Hrm, I don't get the same numbers. I only find ~100 patients who aren't in both datasets (which I'd expect just from things being approximate). I also found those two subjects (12427812, 14029832) in the mimic_core.patients table. What did you use as the source for the MIMIC-CXR patient_id (aka subject_id)?