Send DICOM files via HTTPS

Question

I want to send DICOM files via HTTPS to an image recognition model hosted on ECS.

The idea is that clients can send their DICOM files to the model's endpoint and get back the model's prediction.

Is it possible to do so? Is there a better alternative?

Answer

DICOM files are typically quite large (in MBs). A better approach would be to upload it to an Amazon S3 bucket and then have your recognition model read it from S3. A bit of wiring will be require on your client side to sequence the multiple API calls together.

See: https://docs.aws.amazon.com/AmazonS3/latest/userguide/PresignedUrlUploadObject.html

See:
https://docs.aws.amazon.com/AmazonS3/latest/userguide/PresignedUrlUploadObject.html

Answer

As suggested in previous answers I had to use a pre-signed URL to upload large objects. I followed the architecture mentioned in this video.

It is essentially a serverless architecture where a Lambda function (behind and API endpoint) requests for a pre-signed URL, then returns it to the client and the client then uploads his object into an S3 bucket.

Send DICOM files via HTTPS

Relevanter Inhalt