How to Use OncoVision

Step-by-step guide to analyzing tissue images

🚀

Quick Start

Create a project → Add a dataset → Upload images → Run analysis → View results. The entire workflow takes under 2 minutes.

👤
STEP 1

Create an Account

Auth Page
  1. Open OncoVision Universal in your browser
  2. Click "Create Account" and enter your name, email, and password
  3. You'll be logged in automatically and taken to the Dashboard
📁
STEP 2

Create a Project

Projects Page
  1. Navigate to Projects from the sidebar
  2. Click "+ New Project" in the top right
  3. Give your project a name (e.g. "Breast Tissue Study") and optional description
  4. Your project is now created — it will hold all related datasets
📂
STEP 3

Add a Dataset

Project Detail Page
  1. Click into your project to open it
  2. Click "+ New Dataset"
  3. Name your dataset (e.g. "Batch 1 — H&E Slides")
  4. The dataset is a container for related images you want to analyze together
⬆️
STEP 4

Upload Images

Upload Page or Dataset Page
  1. Go to Upload in the sidebar, or open your dataset
  2. Drag & drop tissue images (JPEG, PNG, TIFF) into the upload zone
  3. You can upload multiple files at once — batch upload is supported
  4. Supported: histopathology slides, biopsies, H&E stained samples
🏷️
STEP 5

Tag with Taxonomy

Dataset Image List
  1. Open your dataset to see the image table
  2. Use the Organ dropdown to tag each image (e.g. Breast, Lung, Colon, Kidney)
  3. Use the Modality dropdown to specify the imaging type (e.g. H&E Pathology, IHC, Fluorescence)
  4. Tags save instantly — no save button needed
  5. This metadata travels through the analysis pipeline and enriches your results
🧪
STEP 6

Run Analysis

Dataset Image List
  1. Open your dataset to see all uploaded images
  2. Click "🧪 Run Analysis" next to any image
  3. The AI pipeline processes the image in under a second using the GPU
  4. Status changes from Pending → Processing → ✓ Analyzed
📊
STEP 7

View Results

Results Page
  1. Click "📊 View Results" on any analyzed image
  2. Review the Tissue Classification Probabilities (7-class breakdown)
  3. Check the Confidence & Uncertainty gauges
  4. Examine the Grad-CAM Heatmap to see which tissue regions the AI focused on
  5. Read the AI-generated recommendation text
🔄
STEP 8

Compare & Iterate

Dashboard & Dataset Pages
  1. Use the Dashboard to see aggregate stats and recent runs
  2. Navigate between results using "← Back to Dataset"
  3. Run analysis on more images to build a comprehensive dataset
  4. The model improves with each analysis — confidence increases over time

💡 Tips

Best Image Quality

Use high-resolution scans (20x–40x magnification). H&E stained slides produce the best results.

Organize by Study

Create separate projects for each research study, and datasets for each batch or patient group.

Check the Heatmap

Always review the Grad-CAM heatmap — it shows whether the AI focused on relevant tissue structures.

Tag Your Images

Always select Organ and Modality for each image. This metadata enriches analysis and enables taxonomy-driven filtering of results.

QC Status Explained

QC Status runs automated quality checks on uploads: blur detection, artifact detection, stain quality, and out-of-distribution scoring. Images flagged for review may need re-scanning.

Model in Training

The AI model is continuously improving. Low confidence scores are expected — every analysis helps refine the model. Always verify results with a pathologist.