Esophageal Motility Study Pipeline

A streamlined pipeline for processing esophageal manometry reports in .docx format. It automates the extraction of patient data, diagnostic tables, embedded images, and textual findings — all neatly structured for analysis and integration.

---

Features

• Extracts patient info and clinical tables to .csv
• Extracts and renames embedded images from reports
• Removes grid lines from diagrams using OpenCV
• Parses diagnostic text sections into structured .json
• Outputs organized, analysis-ready folders

---

⚙️ How It Works

1. Text & Table Extraction

• Reads .docx tables to extract:
  • Patient details
  • Summary metrics
  • Esophageal & UES motility data
• Saves each as a separate .csv

2. Image Extraction & Naming

• Extracts diagrams embedded in the document
• Names them using:
  • Custom defaults (e.g., Swallow Composite)
  • Patient metadata from CSV

3. Image Preprocessing

• Detects and removes grid lines via:
  • Canny edge detection
  • Hough Transform
  • Inpainting (OpenCV)
• Saves clean diagrams to processed_images/

4. Diagnostic Text Extraction

• Captures key sections like:
  • Chicago Classification Findings
  • Procedure, Indications, Impressions
• Exports to a structured .json

---

Output Structure

.
├── extracted_data/
│   ├── Patient_details.csv
│   ├── Esophageal_Manometry_Summary.csv
│   ├── Lower_Esophageal_Sphincter.csv
│   ├── Esophageal_Motility.csv
│   ├── Upper_Esophageal_Sphincter.csv
│   ├── Pharyngeal_UES_Motility.csv
│   ├── Image_filenames.csv
│   └── chicago_classification_findings.json
├── images/
│   └── *.png         # Original extracted images
├── processed_images/
│   └── *.png         # Grid-line removed versions
└── subj.docx         # Input report

---

🔧 Requirements

Install dependencies with:

pip install -r requirements.txt