100% Local · No cloud upload · Your data never leaves your machine
DermUnbound Research Group

Clinico-Dermoscopic-Pathologic
Correlation

A privacy-first desktop application that automates the correlation between pathology reports and clinical & dermoscopic photographs. Built by a Mohs surgeon for daily surgical practice.

View on GitHub DermUnbound Home
Capabilities

From pathology PDF to correlated case review

PathCorrelate bridges the gap between receiving a pathology result and having a complete, correlated case ready for review. Fully automatic.

Report Processing

Extracts text from multi-page pathology PDFs, captures high-resolution screenshots of each report, and handles multi-sample biopsies with Mohs detection.

EXIF Image Classification

Automatically distinguishes clinical from dermoscopic photographs by analyzing ISO, exposure, brightness, and focal length metadata. No manual sorting needed.

Date-Based Matching

Correlates photographs to reports by matching biopsy dates with image capture dates. Distributes images evenly across multiple samples from the same session.

Document Filtering

Intelligent color histogram analysis detects and filters consent forms, referral letters, and other paper documents from the clinical image gallery.

De-identification

Built-in patient data de-identification. Generates fake names, encrypts IDs, shifts birth dates, and replaces phone numbers for teaching and presentations.

Bilingual Interface

Full Hebrew (RTL) and English (LTR) support with comprehensive string tables. Designed for the Israeli healthcare system with international adaptability.

Workflow

Seven steps to correlated cases

From raw pathology PDFs to a complete three-way correlation ready for clinical review.

1

Download Reports

Download unread pathology reports as PDFs from the health system portal. Place them in the designated input folder.

2

Process PDFs

The processor extracts text, captures high-resolution screenshots, detects multi-sample biopsies, and identifies Mohs reports. All fully automatic.

3

Sync Patient Photos

Point PathCorrelate at your patient photos directory. The system indexes images and reads EXIF metadata for classification.

4

Automatic Classification

Each image is classified as clinical, dermoscopic, or document based on EXIF analysis. Document photos (consent forms, referrals) are filtered out.

5

Date Correlation

Images are matched to reports by comparing biopsy dates with photo capture dates, then distributed across samples in report order.

6

Dashboard Review

An interactive dashboard displays all patients with their samples, diagnoses, and correlated images. Click any sample to view the full correlation with clinical and dermoscopic galleries.

7

Clinical Decision

Review the three-way correlation: clinical appearance, dermoscopic features, and histopathological diagnosis. Then document your clinical decision.

Architecture

Built with proven tools

A Python-powered backend for report processing and image analysis, paired with a single-file HTML frontend for zero-dependency deployment.

Python FastAPI Pillow + pillow_heif NumPy PyInstaller HTML / CSS / JS EXIF Analysis PDF Processing HEIC → JPEG Color Histogram
Backend

FastAPI server with endpoints for report data, image serving, HEIC conversion, thumbnail generation, and correlation APIs. Packaged as a standalone Windows EXE via PyInstaller.

Frontend

Single HTML file (~110 KB) with embedded CSS and JavaScript. Responsive design, full i18n support, gallery system with keyboard navigation, and real-time configuration sync.

Privacy Model

100% local execution. No external API calls, no cloud storage, no telemetry. Patient data never leaves the physician's machine. Suitable for PHI handling.

FAQ

Frequently asked questions

Clinico-dermoscopic-pathologic correlation is the process of comparing a skin lesion's clinical appearance and dermoscopic features with its histopathological diagnosis. This three-way comparison helps dermatologists and Mohs surgeons refine their diagnostic accuracy, identify patterns that predict specific diagnoses, and improve clinical decision-making for future cases.
PathCorrelate analyzes EXIF metadata from each photograph, including ISO sensitivity, exposure time, brightness value, and focal length. Dermoscopic images typically have low ISO (50-200), short exposure times, high brightness values, and specific focal lengths characteristic of dermatoscope attachments. Clinical images have different EXIF profiles. This automated classification eliminates manual sorting of hundreds of photographs per patient.
No. PathCorrelate runs entirely on your local machine. No patient data, images, or reports are ever uploaded to external servers. The application processes pathology PDFs and patient photographs locally, making it suitable for handling protected health information (PHI) without cloud privacy concerns.
PathCorrelate processes Hebrew-language pathology reports in standard PDF format. It extracts text from multi-page reports, captures high-resolution screenshots of individual reports, handles multi-sample biopsies, detects Mohs surgery reports, and supports SNOMED coding. The modular architecture allows adaptation to various health system formats.
PathCorrelate uses date-based matching: it compares the biopsy date in each pathology report with the EXIF capture dates of photographs in the patient's folder. It then classifies matched images as clinical or dermoscopic, filters out document photos (consent forms, referral letters) using color histogram analysis, and distributes images across multiple biopsy samples from the same date.
PathCorrelate was developed by Dr. Yehonatan Kaplan, a dermatologist and Mohs surgeon (ACMS Fellow), as part of the DermUnbound research group, which builds open-source, privacy-first clinical tools for dermatology.
Open Source

Ready to streamline your pathology correlation?

PathCorrelate is free and open-source under CC BY-NC-SA 4.0. Clone the repository, build the EXE, and start correlating.

Get Started on GitHub Kaplan Clinic DermUnbound Home