Deploying Medical Imaging Apps on the Cloud

At PYCAD, we constantly work on deploying medical imaging applications on the cloud.

Here are 2 ways that we’ve used to do deployment:

1 - Google Cloud Run:

This is a service from Google Cloud Platform (GCP) where you can easily deploy containerized applications with few steps and without managing much of the infrastructure.

Pros:

• Easy to use.
• You pay only when your app is being used.
• Automatically scales when your app has more traffic.

Cons:

• Does not support GPU compute.

2 - Google Compute Engine:

This is another service from GCP which allows you to rent a virtual machine (VM) with the configuration that you like. You can choose the number of CPUs, GPUs and how much RAM you want.

Pros:

• You have much more control over the infrastructure of your app.
• You can use GPUs.

Cons:

• Takes more steps to configure than Cloud Run.
• You will be paying for the VM whether your app is being used or not.

As you can see, both approaches are on GCP. This is because we really like this cloud provider. Though they have less services than AWS, but they have very thorough documentation and many tutorials.

I should also mention that these are not the only 2 ways to deploy web apps on GCP. Also, we didn’t delve deep into the best way to deploy a medical imaging app if it contains deep learning models inside. Maybe I'll do this in the future.

​

My Brother Released an Open Source Medical Imaging Library

​

My Brother Mohammed has released a free open source library for medical imaging tasks. It's a library that allows you to do many things such as:

• Image File Conversion: Between NIFTI and DICOM formats.
• Visualization: Upcoming feature to visualize medical imaging data.
• Data Preprocessing: Including data splitting and annotation conversion.
• Dataset Creation: Configurations for YOLOv8.

Check out the library github repo.

Show him some love by adding a star (⭐) to that repo!

Moreover, you can easily figure out how to use the library through a chatbot interface where you can ask questions about what the library can do and also how to do specific things with it, such as converting formats or visualization.

You can test the chatbot live here.

​

Tweet of the Day

​

The Imaging Wire @TheImagingWire AI for medical imaging is evolving rapidly. We got an update from @Rick_Abramson of @annalise_ai on major topics in AI. Check it out at https://lnkd.in/ezxGya4W #radiology #medicalimaging #ImagingAI #AI 3:44 PM • Nov 1, 2023 3 Retweets 10 Likes

​

​