🩹

Medical Imaging & Clinical Data Lab

Vertex AI Workbench with Healthcare API

Overview

Healthcare AI has unique requirements: HIPAA compliance, specialized data formats (DICOM for imaging, FHIR for clinical data), domain-specific tools, and regulatory considerations. The Medical Imaging Lab provides all of this in a pre-configured GCP environment. You'll work with de-identified medical imaging datasets — chest X-rays, brain MRIs, liver CT scans — using the MONAI framework purpose-built for medical image analysis. Clinical NLP exercises use MedCAT for extracting medical concepts from doctor's notes. A FHIR-compliant test server lets you practice building integrations with electronic health record systems.

What You'll Do in This Lab

  • Load, view, and preprocess DICOM medical images using Cloud Healthcare API
  • Train disease detection models on chest X-ray datasets with MONAI
  • Build medical image segmentation models (tumor, organ delineation) with U-Net
  • Extract medical entities from clinical notes using MedCAT NER
  • Query and write to FHIR resources — patients, observations, conditions
  • Build explainable AI for clinical models using GradCAM and SHAP

Lab Workflow

1

Launch

Start the lab. A T4 GPU instance with MONAI, a Cloud Healthcare API DICOM store, and a FHIR test server are provisioned.

2

Access Data

Browse de-identified imaging datasets using the OHIF Viewer. Query DICOM metadata with DICOMweb APIs via Cloud Healthcare API.

3

Preprocess

Convert DICOM to training-ready tensors using MONAI transforms — resize, normalize, augment, and create data loaders.

4

Train

Train classification or segmentation models on the T4 GPU. Use MONAI's medical-specific architectures and loss functions.

5

Explain

Generate GradCAM heatmaps showing which regions of the image the model focuses on. Critical for clinical AI trust.

6

Integrate

Write model predictions back to the FHIR server as Observation resources. Build an end-to-end clinical AI workflow.

Hardware & Environment

Machine Type n1-standard-8 (8 vCPU, 30 GB RAM)
GPU 1x NVIDIA T4 (16 GB VRAM)
DICOM Store Cloud Healthcare API with CheXpert, RSNA, and custom de-identified datasets
FHIR Server Cloud Healthcare API FHIR R4 store with synthetic patient data
Imaging Viewer OHIF Viewer (web-based DICOM viewer)
Session Length 2-4 hour sessions

Pre-installed Tools

Cloud Healthcare API (DICOM/FHIR) MONAI (medical imaging framework) spaCy + MedCAT (clinical NLP) 3D Slicer OHIF Viewer

Frequently asked questions about this lab

What is the Medical Imaging & Clinical Data Lab? +
HIPAA-aligned environment for healthcare AI development. Pre-loaded with de-identified medical imaging datasets (X-ray, CT, MRI), clinical NLP tools, and FHIR-compliant test servers.
Which courses use this lab? +
This lab is included in: AI for Healthcare & Life Sciences.
What hardware does this lab run on? +
Vertex AI Workbench with Healthcare API. Machine Type: n1-standard-8 (8 vCPU, 30 GB RAM); GPU: 1x NVIDIA T4 (16 GB VRAM); DICOM Store: Cloud Healthcare API with CheXpert, RSNA, and custom de-identified datasets; FHIR Server: Cloud Healthcare API FHIR R4 store with synthetic patient data.
What software comes pre-installed? +
Comes pre-loaded with Cloud Healthcare API (DICOM/FHIR), MONAI (medical imaging framework), spaCy + MedCAT (clinical NLP), 3D Slicer, OHIF Viewer. No local installs or dependency setup required — open your browser and start working.
Can I bring my own datasets and code into this lab? +
Yes. Datasets can be uploaded directly or synced from Google Cloud Storage. Notebooks and source files have built-in Git integration so you can push work to your own GitHub or GitLab repos.
Do I need to enroll in a course to use this lab? +
Yes. Lab environments are provisioned per-student as part of an AI Labs course enrollment. Browse the courses linked above to find programs that include this lab.

Related labs

Other AI Labs environments students typically use alongside this one.

Ready to Try This Lab?

Enroll in a course that uses this lab, or visit our Houston center for a hands-on demo.