Software Development for Clinical Use in Pathology

Introduction

• Why pathology residents should care about software: LIS limitations, workflow inefficiencies, unmet needs.
• Custom software can improve diagnostics, communication, and safety.
• Real-world impact: frozen section alerts, digital image management, urgent case triage.

Why Build Custom Software?

• Tailored to your lab's needs and workflow — especially important for subspecialty services.
• Enables integration with external systems (EHRs, image viewers, AI pipelines).
• Supports innovation and regulatory adaptation (e.g., NGS reporting, frozen sections).
• From Gershkovich & Sinard: Functionality gaps arise when needed tech exists but isn't in your LIS.

Challenges to Be Aware Of

• Development effort requires domain understanding.
• Validation and long-term support must be considered.
• Institutional inertia: "we're not in the software business" is often a myth.
• From Sinard & Gershkovich: Even one or two developers with domain knowledge can be effective.

Software Lifecycle & DevSecOps in Pathology

• 1. Requirement Gathering – Start with specific use cases from the clinical workflow.
• 2. Functional Specification – What should the system do? Create screen diagrams, button behaviors, exceptions.
• 3. Technical Specification – Database, architecture, and performance planning.
• 4. Implementation and Testing – Use version control, containers (Docker), staging environments.
• 5. Security and Compliance – HIPAA, access logs, secure APIs.
• 6. Deployment & Feedback Loop – Track usage and iterate.

DevSecOps Philosophy

• Secure-by-design.
• Rapid, adaptive iterations.
• Infrastructure-as-code and automation.
• Key point: LIS vendors often cannot move fast enough due to outdated architecture.

Image Analysis & AI

• QPath – Use Groovy scripts to quantify cell populations, analyze TILs, or export tiles for deep learning.
• ImageJ/Fiji – Open-source, powerful for smaller ROI and pixel-level analysis.
• HistomicsTK – Feature extraction and visualization for whole slide images.

Digital Pathology Infrastructure

• Digital Slide Archive (DSA) – WSI management with REST APIs.
• OpenSeadragon – Lightweight JS-based WSI viewer you can embed in web tools.
• Open Source Repositories – GitHub, SourceForge, MONAI.

Clinical Applications

• Alerting systems for STAT/RUSH cases.
• Specimen tracking and intraoperative diagnosis tools.
• NGS variant interpretation modules.

Spotting Gaps in Workflow

• What is frustrating in your daily workflow? What leads to delays, errors, or repetitive effort?
• From lecture material: If your LIS doesn't notify you of STAT slides without a resident, that's a gap.

Solutions

• 7 Options: Ignore, patch manually, switch LIS, request vendor fix, outsource, build yourself, or integrate OSS.
• Residents can help design or prototype solutions with Jupyter Notebooks, low-code tools.
• Learn to write or read functional specs to guide developers.

Conclusion and Call to Action

• Pathology informatics is a core discipline – not optional.
• Custom software development is powerful, cost-effective, and sustainable.
• You can participate by reporting gaps, helping design use cases, or even coding solutions.
• Every meaningful tool starts with someone who understood the problem deeply.

Q&A and Demo Time

• Live demo of QPath on an H&E slide.
• Jupyter notebook using DSA API to pull case info.
• Discussion: What bugs you the most in your LIS today — and what would you build to fix it?

Workflow Augmentation Example