Hiatal Hernia Size
Computer Vision–Based Assessment of Hiatal Hernia Size from Surgical and Endoscopic Videos
Hiatal hernia is a common condition in which part of the stomach herniates through the esophageal hiatus of the diaphragm. Accurate assessment of hiatal hernia size is clinically important for surgical planning, selection of repair technique, and evaluation of post-operative outcomes. However, intra-operative estimation of hernia size is often subjective and varies between surgeons.
The aim of this project is to develop a computer vision system that analyzes laparoscopic or endoscopic video to estimate the size of a hiatal hernia. Using real surgical video data, students will design, implement, and evaluate vision-based methods that detect relevant anatomical landmarks (e.g., diaphragmatic crura, esophagus, stomach) and compute quantitative measures related to hernia size.
Mentor Details:
Prof. Yoav Mintz
Mentor Details:
Requirments:
Students will aim to:
Analyze surgical/endoscopic video data and identify key anatomical landmarks
Design a computer vision pipeline for hernia detection and size estimation
Apply deep learning–based models (e.g., CNNs, Transformers) for segmentation or detection
Incorporate temporal information to stabilize measurements across frames
Evaluate accuracy and consistency of size estimates against clinical references
Problem Statement
Surgical and endoscopic videos of hiatal hernia present several challenges for computer vision systems:
Large inter-patient anatomical variability
Deformation of anatomy due to insufflation and traction
Variable camera angle, zoom, and orientation
Partial occlusion by instruments and sutures
Lack of absolute scale in monocular video
Estimating hernia size requires not only identifying anatomical structures, but also inferring spatial relationships and dimensions under changing visual conditions. The problem is to design a vision-based system that can robustly estimate hernia size or severity from video frames or sequences.
Project Objectives
Students will aim to:
Analyze surgical/endoscopic video data and identify key anatomical landmarks
Design a computer vision pipeline for hernia detection and size estimation
Apply deep learning–based models (e.g., CNNs, Transformers) for segmentation or detection
Incorporate temporal information to stabilize measurements across frames
Evaluate accuracy and consistency of size estimates against clinical references
Technical Scope
The project may include one or more of the following components:
Detection and segmentation of diaphragmatic hiatus, esophagus, and stomach
Landmark detection of crura edges and hernia boundaries
Geometric measurement estimation (area, diameter, or relative size)
Temporal smoothing or tracking across video frames
Weakly supervised learning using operative reports or surgeon-provided estimates
Required Knowledge and Prerequisites
Core Requirements
Familiarity with fundamental computer vision concepts
Experience with convolutional neural networks (CNNs)
Basic knowledge of deep learning frameworks (e.g., PyTorch, TensorFlow)
Ability to work with image and video datasets
Recommended Background
Semantic segmentation and object detection
Geometric computer vision concepts
Video-based modeling and tracking
Evaluation metrics for regression and estimation tasks
No prior surgical knowledge is required; relevant anatomical and procedural context will be provided.
Project Difficulty and Expected Level
Vision complexity: High (anatomical variability, occlusion, lack of scale)
Modeling complexity: Moderate to high (segmentation + measurement)
Domain knowledge: Low
This project is well-suited for:
Teams of 2–4 students
Expected Outcomes
A working computer vision prototype for hiatal hernia size estimation
Quantitative evaluation against reference measurements or expert estimates
Analysis of variability and failure modes
Well-documented codebase and a technical report