MIMIC IV Echo Data

[JBarsotti]

Description

You all are doing wonderful work with MIMIC! My collaborators are cardiologists and would be interested in getting their hands on MIMIC-IV-ECG. I know it comes out in August but also know that it is available to the consortium members. I was just wondering whether there is a process to become a consortium member and whether it is worth it.

Thanks!

[alistairewj]

You can reach out to [email protected] for more detail and those responsible for the consortium will get back to you I'm sure.

[JBarsotti]

Thank you! I am working on a presentation using your MIMIC data right now. :)

[emiramaro]

Hi, my colleagues and I are currently looking at the MIMIC-IV echocardiogram dataset. We are trying to use the echocardiogram data and linking it to the MIMIC-IV Note. It is mentioned that the note will be made available on the MIMIC-IV Note for the studies that have note_id within the ‘study_list.csv’. However, upon examination on the MIMIC-IV Note ‘radiology.csv’ dataset, we did not find any linked report to the note_id within the echocardiogram dataset. We did find some echocardiogram descriptions within the ‘discharge.csv’ dataset in a form of discharge summaries.

Thus, is the cardiology report for the dataset has not been made available within the MIMIC-IV note? Or is the one within ‘discharge.csv’ would be the echo report?

Thank you!

[briangow]

@emiramaro , the cardiologist notes for a given Echo have not been released yet. We are working on getting those released, hopefully it'll be sooner than later.

[emiramaro]

Thank you for the response, Brian.

[WangMengxiao319]

@briangow, thanks for your elaborate work! I am also looking forward to the cardiologist notes relevant to Echo.

[ankushjindal5]

@briangow Hey Brian, wondering when the notes would be released, I believe the MIMIC echo data has been released for quite a long time, it would be very helpful if those scans are complemented by their reports. Truly, It would be a game changer for Medical AI community. Can you provide a timeline for the release ?

[briangow]

@ankushjindal5 , we are actively working on this but don't have an estimated release date at this point.

[Rapuris]

Hello, I am working with the MIMIC-IV Echocardiogram dataset. However, I am unable to find out how exactly the different views within the dataset are represented. I could not find metadata within the dicom files suggesting what probe position/angle view number _XXXX corresponds to. Moreover, do the view numbers between different patients correspond to the same probe position/angle (i.e., do all views _0001 represent the same probe position across all patients? Any help would be greatly appreciated!

[Rapuris]

Following back up. Any advice would be greatly appreciated!

[briangow]

@Rapuris , sorry for the late reply. The Echo metadata doesn't contain information regarding the probe position/angle for a given view. The view numbers also do not necessarily correspond to a particular probe position/angle.

We are hoping to release a derived dataset at some point which uses ML to classify the view information.

[Rapuris]

@briangow Thank you for the update! As of now, is there any way to distinguish between the videos/frames in the study (as to type of ultrasound imaging, view, etc.)? I assume not, but I want to confirm. I'm looking forward to the derived dataset as well. Any estimation of when that is coming out?