You shouldn't need to go through GDI for this. I do something similar where I download TIFs from SSRS and save them to the repository. I use HttpClient and get the results as a byte array which then gets put into a MemoryStream.

Then you can use DocumentImporter to add the TIF images to a DocumentInfo instance. You can also do OCR as you bring the pages in.

Here's a snippet. You should be able to convert that string straight into a byte array. Let me know if you need more information.

var importer = new DocumentImporter();
importer.Document = doc;
importer.OcrImages = true;
importer.OcrOptions.OptimizationMode = OcrOptimizationMode.Accuracy;

var base64String = "";
var ms = new MemoryStream(Convert.FromBase64String(base64String))
importer.ImportImages(ms);