{"slug": "building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai", "title": "🩺 Building RadScan AI: Autonomous Multimodal Radiology Triage on GCP Cloud Run & Vertex AI", "summary": "RadScan AI, a multimodal radiology triage system built for the All Things Agentic Hackathon, combines a 2.5D Volumetric CNN-BiGRU neural network with Vertex AI Gemini models on Google Cloud Run to detect 12 pathology targets and draft structured reports in under 3 seconds. The system processes 24 parallel depth slices across three planes, using Grad-CAM heatmaps for lesion localization, and scales to zero instances when idle to minimize costs.", "body_md": "Disclaimer:This article was created for the purpose of entering the All Things Agentic Hackathon hosted by Google Cloud and Devpost.\n\nRadiologists worldwide face extreme burnout, evaluating 50+ volumetric MRI scans daily. Each scan contains 24 to 40 high-resolution DICOM slices. Manually inspecting every plane, checking ligament continuity across depth slices, and dictating repetitive clinical reports takes **10 to 15 minutes per study**.\n\nWe built **RadScan AI** to eliminate this workflow bottleneck.\n\nBy combining **GCP Cloud Run (L4 GPU scale-to-zero microservices)**, a **2.5D Volumetric CNN-BiGRU Neural Network** (trained on 819,100 DICOMs / 530 GB data), and **Vertex AI Gemini 3.5 Flash / 1.5 Pro**, RadScan AI acts as an autonomous radiology co-pilot that performs 12-target pathology detection, pinpoints lesion coordinates with Grad-CAM visual heatmaps, and drafts structured DICOM reports in under **3 seconds**—saving radiologists ~6 minutes per scan.\n\nRadScan AI is architected as two decoupled, serverless microservices on **Google Cloud Platform**:\n\n```\nflowchart TD\n    subgraph Client [\"Next.js 14 Medical Workspace\"]\n        A[1-Click Sample Buttons / DICOM Upload] --> B[Multi-Planar Slice Slider]\n        B --> C[Grad-CAM Heatmap Opacity Layer]\n        B --> D[Gemini Clinical Report Generator]\n    end\n\n    subgraph Backend [\"FastAPI Microservice (GCP Cloud Run L4 GPU)\"]\n        E[POST /api/v1/predict] --> F[2.5D Volumetric CNN-BiGRU Engine]\n        F --> G[Grad-CAM Heatmap Synthesizer]\n        E --> H[POST /api/v1/report]\n        H --> I[Google ADK / Vertex AI Gemini 3.5 SDK]\n    end\n\n    Client --> Backend\n```\n\n`google-cloud-aiplatform`\n\nand `google-genai`\n\nSDK in native JSON mode (`response_mime_type=\"application/json\"`\n\n).Single 2D MRI slices often mimic tears due to volume averaging artifacts. RadScan AI processes 24 parallel depth slices across Sagittal, Coronal, and Axial planes simultaneously:\n\nFor a 3D MRI volume stack S = {s1, s2, ..., s24}, spatial CNN features f_t = CNN(s_t) are fed into a Bidirectional GRU to track ligament continuity across consecutive depth slices:\n\nThe explainability weights alpha_k^c for target class c at feature map A^k are computed via backpropagated gradients across feature channels:\n\nDeploying to **GCP Cloud Run** ensures our backend scales down to 0 instances when idle, keeping cloud costs at virtually $0/month while serving fast, sub-second inference on demand:\n\n```\n# Build & Deploy Backend Microservice to Cloud Run\ngcloud run deploy radscan-ai-backend \\\n  --image gcr.io/YOUR_GCP_PROJECT_ID/radscan-ai-backend:v1 \\\n  --platform managed \\\n  --region us-central1 \\\n  --memory 2Gi \\\n  --cpu 2 \\\n  --allow-unauthenticated\n```\n\nBuilding RadScan AI for the **All Things Agentic Hackathon** proved that combining high-performance computer vision with structured Gemini 3.5 LLM agents creates production-grade medical automation.\n\n*#AllThingsAgenticHackathon #GoogleCloud #VertexAI #Gemini #FastAPI #NextJS #AIHealthcare*", "url": "https://wpnews.pro/news/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai", "canonical_source": "https://dev.to/sahil_5273/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-vertex-ai-42d0", "published_at": "2026-08-25 19:17:41+00:00", "updated_at": "2026-08-25 19:44:05.472875+00:00", "lang": "en", "topics": ["artificial-intelligence", "computer-vision", "large-language-models", "ai-infrastructure", "ai-products"], "entities": ["Google Cloud", "Vertex AI", "Gemini", "RadScan AI", "GCP Cloud Run", "FastAPI", "Next.js", "All Things Agentic Hackathon"], "alternates": {"html": "https://wpnews.pro/news/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai", "markdown": "https://wpnews.pro/news/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai.md", "text": "https://wpnews.pro/news/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai.txt", "jsonld": "https://wpnews.pro/news/building-radscan-ai-autonomous-multimodal-radiology-triage-on-gcp-cloud-run-ai.jsonld"}}