build-small-hackathon/surgical-tissue-segmentation

<div align="center">

<img src="https://img.shields.io/badge/YOLOv26n--seg-Ultralytics-6366f1?style=for-the-badge&logo=python&logoColor=white"/> <img src="https://img.shields.io/badge/Llama3.18B-MetaAI-8b5cf6?style=for-the-badge&logo=meta&logoColor=white"/> <img src="https://img.shields.io/badge/ModalGPU-T4Inference-22c55e?style=for-the-badge"/> <img src="https://img.shields.io/badge/Gradio-HuggingFaceSpaces-f97316?style=for-the-badge&logo=huggingface&logoColor=white"/> <img src="https://img.shields.io/badge/CholecSeg8k-MICCAI_2020-ef4444?style=for-the-badge"/>

🔬 SurgiSight

Surgical Anatomy AI for Laparoscopic Training

Real-time danger-zone detection + AI anatomy explanations for surgical trainees. Built for Build Small Hackathon 2026 — solo project, fully deployed, end-to-end.

🚀 [Watch Demo](https://www.youtube.com/watch?v=Z-jaj31B-ss) &nbsp;|&nbsp; 📄 [Blog](https://huggingface.co/blog/sugan04/surgical-tissue-segmentation) &nbsp;|&nbsp; 🤗 [HuggingFace Space](https://huggingface.co/spaces/build-small-hackathon/surgical-tissue-segmentation) &nbsp;|&nbsp; [Social media - LinkedIn](https://www.linkedin.com/posts/sugan-subramanian_ai-machinelearning-medicalai-ugcPost-7469109783028076544-TVeL/?utm_source=share&utm_medium=member_desktop&rcm=ACoAACixJ8kBbDBD81FWoNnyJCVWR4Lrg1EcVv0) &nbsp;|&nbsp; 📄 [Agent traces](https://huggingface.co/datasets/sugan04/surgisight-traces) HF ID: sugan04 ---

</div>

The Problem

Every year, bile duct injuries occur in roughly 1 in 300 laparoscopic cholecystectomies (gallbladder removal surgeries). This is the most common serious complication in one of the most frequently performed surgeries in the world (~1.2 million per year in the US alone). Many of these injuries happen because trainees — operating under pressure in a visually complex, blood-filled field — cannot reliably identify critical structures in real time.

Current surgical training relies on:

• —Static textbook diagrams — no relevance to live video
• —Senior surgeon supervision — not always available, and creates cognitive load
• —Experience alone — acquired over years, with real patients

There is no tool that watches the surgical video alongside a trainee and says: "That's the hepatic vein. Don't touch it."

SurgiSight is that tool.

The Solution

SurgiSight is an AI assistant for laparoscopic surgical training that:

1. 1.Segments any laparoscopic cholecystectomy frame using a fine-tuned YOLOv8n instance segmentation model, identifying 13 surgical structures in real time.
2. 2.Flags danger zones automatically — Hepatic Vein, Cystic Duct, and Blood trigger a red alert.
3. 3.Explains the anatomy using Meta Llama 3.1 8B, giving the trainee a 3-sentence teaching note grounded in the detected context.
4. 4.Enables interactive Q&A — the trainee can ask follow-up questions in natural language ("Why is the cystic duct dangerous here?") and get expert-level answers.
5. 5.Exports clinical-grade reports in both PDF and Word format, suitable for case review or portfolio use.
6. 6.Supports multilingual responses (English and French), with text-to-speech for each AI reply.

Everything runs in a single Gradio interface, deployed on Hugging Face Spaces, with GPU inference handled by Modal.

Architecture

┌──────────────────────────────────────────────────────────────────┐
│                     HUGGING FACE SPACES                          │
│                      (Gradio Frontend)                           │
│                                                                  │
│   ┌─────────────┐    ┌──────────────────┐    ┌───────────────┐  │
│   │  Image      │    │  Results Card    │    │  AI Chat      │  │
│   │  Upload     │ →  │  (Detections,    │    │  (Llama 3.1   │  │
│   │  + Conf     │    │   Alert, Brief)  │    │   8B via HF)  │  │
│   │  Slider     │    └──────────────────┘    └───────────────┘  │
│   └──────┬──────┘                                               │
│          │ PIL Image bytes                                        │
└──────────┼───────────────────────────────────────────────────────┘
           │
           ▼ modal.Cls.from_name() remote call
┌──────────────────────────────────────────────────────────────────┐
│                        MODAL (GPU T4)                            │
│                                                                  │
│   ┌─────────────────────────────────────────────────────────┐   │
│   │  SurgiSightDetector.run()                               │   │
│   │  ├── Load YOLOv26n-seg weights (CholecSeg8k fine-tune)   │   │
│   │  ├── Run instance segmentation inference                 │   │
│   │  ├── Draw colour-coded masks on annotated frame          │   │
│   │  └── Return: annotated_bytes + detections list           │   │
│   └─────────────────────────────────────────────────────────┘   │
└──────────────────────────────────────────────────────────────────┘
           │
           ▼ annotated_bytes + [{cls_id, conf}]
┌──────────────────────────────────────────────────────────────────┐
│                 GRADIO APP (back to Spaces)                      │
│  ├── Map cls_id → CLASS_NAMES (13 classes)                       │
│  ├── Check DANGER_CLASSES → build alert string                   │
│  ├── Call Llama 3.1 8B via HF InferenceClient → explanation      │
│  └── Render results HTML + chat panel                            │
└──────────────────────────────────────────────────────────────────┘

Core Concepts Explained

1. Instance Segmentation vs. Object Detection

Most people know object detection — the model draws bounding boxes around objects. Instance segmentation goes further: it draws a pixel-level mask for every detected object, letting you see the exact shape and boundary of each structure rather than just a rectangle around it.

In surgery, this matters enormously. A bounding box around "Hepatic Vein" tells you it's somewhere in the frame. A mask tells you exactly which pixels belong to it — so you know precisely where not to cut.

SurgiSight uses YOLOv8n-seg, the nano (smallest) version of Ultralytics YOLOv8's segmentation model. It was chosen because:

• —Nano = fast inference, deployable on modest GPU
• —The seg variant predicts both bounding boxes and segmentation masks
• —YOLOv8 is the current industry standard for real-time detection tasks

2. Fine-tuning on CholecSeg8k

A generic YOLOv8 model trained on COCO (everyday objects) has never seen laparoscopic video. It would not recognise a Cystic Duct or L-hook electrocautery instrument.

CholecSeg8k (Hong et al., MICCAI 2020) is a publicly available dataset of 8,080 annotated frames from laparoscopic cholecystectomy videos, labelled across 13 classes:

The YOLOv8n-seg model was fine-tuned on this dataset. The resulting model achieves mAP50 = 0.581 on the validation set — competitive for a nano model on a specialised medical segmentation task.

3. What is mAP50?

Mean Average Precision at IoU threshold 0.50 (mAP50) is the standard metric for object detection and segmentation tasks.

• —IoU (Intersection over Union): Measures how much the predicted mask overlaps with the ground truth mask. IoU = 1.0 means perfect overlap; 0.0 means no overlap.
• —Precision at 0.50: A detection is counted as "correct" if its IoU with the ground truth is ≥ 0.50.
• —Average Precision: Area under the precision-recall curve for a single class.
• —mAP50: Mean of AP50 across all classes.

A score of 0.581 means the model is reliably identifying the right structures in the right places more than half the time — meaningful for a 13-class medical task where even expert human annotators disagree on boundaries.

4. Modal for GPU Inference

Hugging Face Spaces runs on CPU by default. Running YOLOv8 inference on CPU is slow (~3-5 seconds per frame). SurgiSight offloads all inference to Modal, a serverless GPU platform.

The SurgiSightDetector class is deployed as a Modal app. When the Gradio frontend calls detector.run.remote(image_bytes, conf), Modal spins up a T4 GPU container, runs the inference, and returns the result — in under 2 seconds. The Spaces app never needs a GPU itself.

This is a key architectural decision: decouple the UI from compute, so the demo stays free to host while getting GPU-grade speed.

5. Retrieval-Augmented Context for the LLM

Rather than asking Llama 3.1 a generic anatomy question, SurgiSight gives it grounded context: the exact list of detected structures and the current safety alert. The system prompt is dynamically constructed per frame:

"You are a surgical anatomy teacher for a junior resident.
Detected in a laparoscopic cholecystectomy frame: Liver, Cystic Duct, Grasper, Hepatic Vein.
Safety status: ⚠ DANGER ZONE: Hepatic Vein, Cystic Duct — Extreme caution required.
Answer concisely in 2-4 sentences."

This means every LLM response is frame-specific, not generic. The AI knows what it's looking at.

6. The Critical View of Safety (CVS)

The Critical View of Safety is a surgical standard — before clipping the cystic duct, the surgeon must confirm:

1. 1.The hepatocystic triangle is cleared of fat and fibrous tissue
2. 2.Two and only two structures enter the gallbladder

SurgiSight's DANGER_CLASSES logic is inspired by CVS: the Cystic Duct and Hepatic Vein are flagged because confusing them is how bile duct injuries happen. The system teaches this principle explicitly in AI responses.

Tech Stack

Project Structure

surgisight/
├── app.py                    # Main Gradio application (this repo)
├── modal_inference.py        # Modal GPU deployment (separate deploy)
├── requirements.txt          # Python dependencies
├── examples/
│   ├── frame_80_endo.png     # CholecSeg8k example frame
│   ├── frame_912_endo.png    # CholecSeg8k example frame
│   ├── frame_2176_endo.png   # CholecSeg8k example frame
│   └── frame_939_endo.png    # CholecSeg8k example frame
└── README.md                 # This file

Setup & Deployment

Prerequisites

pip install gradio modal ultralytics huggingface-hub Pillow \
            reportlab python-docx gtts

Environment Variables

Deploy the Modal GPU Backend

# Authenticate with Modal
modal token new

# Deploy the inference class
modal deploy modal_inference.py

# The app name must match: modal.Cls.from_name("surgisight", "SurgiSightDetector")

Run Locally

python app.py
# Opens at http://localhost:7860

Deploy to Hugging Face Spaces

1. 1.Create a new Space (Gradio SDK)
2. 2.Add secrets: HF_TOKEN, MODAL_TOKEN_ID, MODAL_TOKEN_SECRET
3. 3.Push this repo — Spaces auto-deploys on push

How to Use

1. 1.Upload a surgical frame — drag and drop any laparoscopic cholecystectomy image, or click one of the four provided example frames (sourced from CholecSeg8k).
2. 2.Adjust confidence threshold — the slider (default 0.25) controls how certain the model must be before flagging a structure. Lower = more detections, higher = fewer but more confident.
3. 3.Click "▶ Run Analysis" — the model runs on Modal GPU and returns the segmented image with colour-coded masks within ~2 seconds.
4. 4.Read the Results Panel — the Safety Alert (green ✓ or red ⚠) and the detected structures with confidence bars appear immediately.
5. 5.Ask the AI — the chat panel opens automatically. Use the suggested questions or type your own.
6. 6.Change language — use the dropdown to switch to French; all responses (including previous ones) are re-translated.
7. 7.Export — click ⬇ PDF or ⬇ Word to download a full clinical-style report including both images, detection table, and anatomy notes.

Model Performance

Limitations & Ethics

This is a research prototype. It is explicitly not a medical device and should never be used in real surgical procedures or to guide clinical decisions.

• —The model was trained on 8,080 frames from a limited set of procedures. It may not generalise to all patients, camera angles, or surgical conditions.
• —mAP50 of 0.581 means the model makes mistakes — both false positives (flagging safe tissue) and false negatives (missing danger).
• —AI anatomy explanations are generated by a general-purpose LLM and are not verified by a medical professional.
• —The system is intended purely for educational simulation and training aid purposes.

DISCLAIMER: Research prototype only. Not a medical device. Contains no real patient data. Built for Build Small Hackathon 2026.

References

• —CholecSeg8k: Hong, W.-Y., et al. CholecSeg8k: A Semantic Segmentation Dataset for Laparoscopic Cholecystectomy Based on Cholec80. MICCAI 2020 Workshop. arXiv:2012.12503.
• —YOLOv8: Jocher, G., et al. Ultralytics YOLOv8. Ultralytics, 2023. https://github.com/ultralytics/ultralytics
• —Llama 3.1: Meta AI. The Llama 3 Herd of Models. arXiv:2407.21783, 2024.
• —Critical View of Safety: Strasberg, S. M., et al. An analysis of the problem of biliary injury during laparoscopic cholecystectomy. Journal of the American College of Surgeons, 1995.
• —Modal: https://modal.com — Serverless GPU infrastructure.

Author

Built solo for Build Small Hackathon 2026.

<div align="center"> _{CholecSeg8k · MICCAI 2020 · No patient data · Research prototype · Build Small Hackathon 2026} </div>