🤝 Track Overview

Format: 2 slides maximum (PowerPoint or PDF)
Type: Non-Archival
Audience: Clinicians, industry partners, healthcare administrators
Goal: Match clinical problems with technical solutions

What is the Problem-Pitch Track?

The Problem-Pitch Proposals track is a unique opportunity for clinicians, healthcare administrators, and industry partners to present concrete clinical challenges that need technical solutions. This track connects those with real-world problems to computer vision researchers who can help solve them.

🎯 Bridge the Gap: Problems Meet Solutions

Too often, CV researchers work on problems without understanding real clinical needs, while clinicians struggle with challenges that could be solved with modern AI. This track creates a structured way to bring these groups together, fostering collaborations that lead to real-world impact.

Who Should Submit?

👨‍⚕️

Clinicians

Radiologists, cardiologists, pathologists, and other physicians who encounter daily challenges in medical imaging workflows that could benefit from AI solutions.

🏥

Healthcare Administrators

Hospital IT leaders, department chairs, and healthcare executives seeking to improve imaging operations, workflow efficiency, or patient care through AI.

🏭

Industry Partners

Medical device companies, healthcare software firms, and imaging equipment manufacturers with specific technical challenges in their products or services.

🔬

Clinical Researchers

Translational researchers who understand clinical needs and want to connect with CV experts to develop technical solutions.

⚠️ No CV Expertise Required!

You do NOT need to be a computer vision expert to submit a problem-pitch. In fact, we specifically encourage submissions from those without technical AI backgrounds. Your clinical expertise and problem understanding is what matters. We'll help match you with technical experts.

What Makes a Good Problem-Pitch?

Excellent problem-pitches have these characteristics:

Clearly Defined Clinical Problem

Specific, well-articulated challenge with clear impact on patient care or clinical workflow

Good: "Detecting subtle pneumothorax on chest X-rays in emergency settings where radiologists aren't immediately available"

Too vague: "We need better AI for radiology"

Quantifiable Impact

Clear statement of who benefits and how (patient outcomes, time savings, cost reduction, etc.)

Good: "Could reduce time-to-treatment by 30 minutes for 500 patients/year at our center"

Available Data

Existing or collectible imaging data to work with (even if limited)

Good: "We have 2,000 annotated CTs from 2020-2024"

Also good: "We can prospectively collect 500 studies over 6 months"

Feasible Scope

Problem is concrete enough to make progress within a research collaboration timeframe

Good: "Segment liver lesions in contrast-enhanced CT"

Too broad: "Solve all problems in abdominal imaging"

Path to Deployment

At least preliminary idea of how a solution could be integrated into clinical practice

Good: "Solution would integrate with our Philips PACS system and be used during daily read-out"

The Two-Slide Format

Problem-pitch proposals consist of exactly 2 slides that will be presented during a dedicated collaboration session at the workshop.

Slide 1

The Clinical Problem

Purpose: Clearly define the unmet clinical need and its significance

Required Elements:

  • Problem Statement - What is the clinical challenge? (2-3 sentences)
  • Clinical Context - What setting? What specialty? What patient population?
  • Current Standard of Care - How is this handled today? What are the limitations?
  • Why It Matters - Patient impact, frequency, severity, or workflow burden
  • Success Criteria - How would you measure if a solution works?

Optional but Helpful:

  • Example images showing the challenge
  • Statistics (incidence, cost, time burden, etc.)
  • Why existing solutions fail or are insufficient
Example Slide 1 Content:

Problem: Emergency department chest X-rays need rapid triage for critical findings (pneumothorax, large effusion, widened mediastinum), but overnight radiologists are often unavailable, leading to delayed diagnosis.

Context: Level 1 trauma center ED with ~200 chest X-rays/night, single remote radiologist covering multiple hospitals.

Current Process: ED physicians interpret independently, stat radiologist reads only for critical cases if called. Average 45-minute delay for radiologist interpretation.

Impact: ~15 cases/year where critical findings are missed or delayed. Potential for adverse outcomes in time-sensitive conditions.

Success: AI triage system with 95% sensitivity for critical findings, <5% false positive rate, results in <1 minute.

Slide 2

The Data and The Ask

Purpose: Describe available resources and explicitly state what help you need

Required Elements:

  • Data Description
    • Type of imaging (modality, protocol)
    • Volume (number of studies, patients, time period)
    • Annotations/labels (if any)
    • Data access (institutional, public, shareable?)
  • The Ask - What expertise or technical solution are you seeking?
    • What technical capabilities do you need?
    • What type of collaboration are you proposing?
    • What can you contribute?
    • What are the constraints (timeline, budget, IRB, etc.)?

Optional but Helpful:

  • Sample images from your dataset
  • Previous attempts or related work
  • Available resources (annotation support, compute, funding)
  • Timeline expectations
Example Slide 2 Content:

Available Data:

  • 12,000 frontal chest X-rays (2020-2024)
  • Radiologist reports available for all
  • 1,500 have expert annotations for pneumothorax, effusion, and mediastinal widening
  • Data shareable under IRB-approved research agreement
  • Additional prospective collection possible

What We're Looking For:

  • CV researchers with expertise in medical image classification and object detection
  • Experience with imbalanced datasets and high sensitivity requirements
  • Interest in real-world clinical deployment

We Can Provide:

  • Clinical expertise for annotation and validation
  • IRB support for data sharing
  • Potential funding for promising collaborations
  • Path to clinical pilot study if successful

Contact: Dr. Jane Smith, jsmith@hospital.edu

Submission Format

Technical Requirements

  • Exactly 2 slides (one for problem, one for data/ask)
  • File format: PowerPoint (.pptx) or PDF
  • Slide size: Standard 16:9 widescreen format
  • File size: Maximum 20MB
  • Fonts: Use standard fonts (Arial, Calibri, Times New Roman) for compatibility

Content Guidelines

  • Clarity: Use clear, simple language. Avoid excessive medical jargon.
  • Visuals: Include images, diagrams, or charts to illustrate the problem
  • Brevity: Be concise. Each slide should be readable in 2-3 minutes.
  • Contact Info: Include email on Slide 2 for interested researchers to reach you
  • Anonymization: NOT required for problem-pitches (you want to be contacted!)
⚠️ Data Privacy & Confidentiality

When including example images on your slides:

  • Ensure all patient information is removed (no names, MRNs, dates, identifiers)
  • Verify images are properly de-identified per HIPAA/institutional requirements
  • Get IRB approval if needed for showing example cases
  • Don't include proprietary information you can't share publicly

Review & Selection Process

1

Submission

Submit your 2-slide pitch via portal by November 28, 2025

2

Organizer Review

Workshop organizers evaluate for clarity, impact, and collaboration potential

3

Selection

Best pitches selected for live presentation (limited slots available)

4

Notification

Decision by January 2, 2026

5

Workshop Presentation

Present during dedicated "Collaboration-Building Session" on March 6, 2026

Selection Criteria

Clinical Significance

How important is this problem to patient care?

Clarity

Is the problem well-defined and easy to understand?

Feasibility

Is there sufficient data and is the scope reasonable?

Collaboration Potential

Are resources available to support a collaboration?

Innovation Opportunity

Does this enable novel CV research?

Path to Impact

Is there a realistic pathway to clinical deployment?

Note: Due to time constraints, only a limited number of pitches (~10-15) can be presented live. However, all accepted pitches will be featured on the workshop website and in networking materials.

The Collaboration Session

How It Works

Selected problem-pitches will be presented during a dedicated 1-hour collaboration-building session (11:30-12:30) on the workshop day:

Phase 1: Lightning Presentations (30 minutes)

  • Each selected pitch gets 3 minutes to present their 2 slides
  • No Q&A during presentations (to maximize number of pitches)
  • Audience includes CV researchers, other clinicians, and industry partners

Phase 2: Networking & Matching (30 minutes)

  • Presenters set up at designated stations
  • Interested researchers visit stations for detailed discussions
  • Exchange contact information and discuss collaboration possibilities
  • Organizers facilitate introductions and connections

Phase 3: Extended Networking (During Lunch)

  • Continues during lunch + poster session (12:30-14:00)
  • Informal follow-up discussions
  • Small group conversations

What Happens After?

The workshop provides the initial connection. After the workshop:

  • Interested parties follow up directly via email
  • Organizers can help facilitate introductions if needed
  • Collaborating parties define scope, timelines, and agreements independently
  • We track and celebrate successful collaborations that result!

Example Problem-Pitch Topics

Automated Measurement

"Need automated measurement of ejection fraction from echocardiograms with uncertain image quality. Have 5,000 studies with cardiologist measurements."

Rare Disease Detection

"Detecting rare pediatric bone dysplasias on X-rays. Only 200 positive cases but 10,000 normal controls available."

Workflow Optimization

"Prioritizing worklist for urgent findings in brain MRI. Need to identify stroke, hemorrhage, mass effect within minutes of scan completion."

Cross-Scanner Harmonization

"Quantitative measurements from lung CT vary across 3 different scanner models at our multi-site practice. Need robust features."

Report Generation

"Auto-generate structured reports for routine chest X-rays. Have 50,000 image-report pairs to work with."

Quality Assurance

"Detect suboptimal image quality in mammography before patient leaves. Reduce recall rate due to technical issues."

Tips for a Successful Problem-Pitch

🎯 Be Specific

Don't say "AI for radiology." Say "Detect subtle rib fractures on lateral chest X-rays in trauma patients."

📊 Quantify Impact

Use numbers: "Affects 500 patients/year at our center" or "Costs 30 minutes per case" helps researchers understand importance.

🗂️ Describe Your Data

Be concrete: "2,000 MRIs, 2019-2024, 60% female, age 45-75" is better than "some MRI data."

🤝 Show Collaboration Readiness

Demonstrate you're serious: mention IRB status, institutional support, available resources.

📸 Use Visuals

Include example images, workflow diagrams, or charts. A picture is worth a thousand words.

✉️ Make It Easy to Contact You

Put your email prominently on Slide 2. Consider including LinkedIn or institutional profile.

Benefits of Participating

🤝

Find Expert Collaborators

Connect with leading CV researchers who have the technical skills to solve your problem

💡

Get Technical Perspectives

Learn what's possible with current AI capabilities and what new research directions could help

🎓

Educate the Community

Help CV researchers understand real clinical needs, shaping future research directions

🚀

Accelerate Solutions

Turn a persistent clinical problem into an active research project with real-world impact potential

📄

Publication Opportunities

Successful collaborations often lead to joint publications and conference presentations

💼

Access to Expertise

Tap into a community of experts in foundation models, explainable AI, fairness, and more

Important Dates

Submission Deadline: November 28, 2025 (23:59 AoE)
Notification: January 2, 2026
Collaboration Session: March 6, 2026 (11:30-12:30)

Frequently Asked Questions

Q: Do I need technical AI expertise to submit?

A: No! This track is specifically designed for clinicians and industry partners WITHOUT CV expertise. You bring the problem, we help you find the technical experts.

Q: What if I don't have any labeled data?

A: That's okay! If you have imaging data that could be labeled, mention that you can provide annotation support. Many CV methods work with limited labels.

Q: Can I submit multiple problem-pitches?

A: Yes, but we encourage submitting your highest-priority problem. If selected, you'll only present one pitch due to time constraints.

Q: What if my problem-pitch isn't selected for live presentation?

A: All accepted pitches will be featured on the workshop website with contact information, so researchers can still reach out to you.

Q: Do I need to attend the workshop in person?

A: Highly recommended! The networking session is most effective in person. However, remote presentation options may be available (check with organizers).

Q: Is there a registration fee?

A: Problem-pitch presenters must register for the workshop (standard workshop registration fees apply).

Q: What happens if multiple researchers are interested in my problem?

A: Great! You can discuss with multiple groups and choose the best fit. Some problems might benefit from multiple research approaches.

Q: Can I pitch a problem my institution wants to commercialize?

A: Yes, but be clear about IP and commercialization expectations upfront. Many academic collaborations work within these constraints.

Submission Portal

Submission portal will open in October 2025.

You'll upload your 2-slide PowerPoint or PDF file and provide basic contact information.

Contact p2pcv.wacv@gmail.com with questions or for early feedback on your pitch.

Questions?

For questions about the Problem-Pitch track, contact us at p2pcv.wacv@gmail.com with "Problem-Pitch" in the subject line.

We're especially eager to help clinicians and industry partners prepare strong submissions. Don't hesitate to reach out for guidance!

Have a Clinical Challenge? Present It!

Problem-pitches turn real-world problems into research opportunities.