Top of page

Student/Staff Portal
Global Site Navigation

School of Science

Local Section Navigation
You are here: Main Content

Related Contents

Multimodal AI approaches to Cervical and Ovarian Cancer Diagnosis

Project Outline

Cervical and ovarian cancers are responsible for high levels of early female mortality throughout the world. It is crucial to have early and accurate screening and diagnosis to improve survival rates and cost of cancer treatment. Traditional diagnostic methods are limited in their effectiveness. Cytology screening and ultrasound imaging are often limited by inter-observer variability, resource constraints, and the need for specialized expertise.

Recent advances in artificial intelligence (AI), particularly in deep learning and natural language processing may offer solutions to these challenges. For example, Large Language Models (LLMs) have been found to provide increased capabilities in understanding and generating human-like text. Other approaches such and transformer-based vision models have provided improved performance in image classification tasks. This project will explore the Integrating these technologies into a unified multimodal framework to provide tools to enhance diagnostic accuracy, automate clinical reporting, and support decision-making in gynecologic oncology.

This research project will explore novel domain-specific LLMs which will be fine-tuned on gynecologic oncology reports, combined with advanced vision transformers like EVA-02, represents a cutting-edge methodology.

The research will also focus on improving clinical decision-making, addressing a critical gap in current gynecological oncology clinical AI systems. The research will contribute to development of benchmark datasets and simulation of real-world triage scenarios to further enhances the practical relevance and translational potential of the research.

Essential Skills

  • Strong programming skills in Python or other languages
  • Machine learning and deep learning knowledge (CNNs, transformers, GNNs
  • Good oral and report writing skills.
  • Able to work in team environment
  • Good project management skills

Desired Skills

  • Experience with large language models (LLMs) and multimodal data integration
  • High-performance computing experience
  • Experience in Medical science an advantage

Project Area: Computer Science

Supervisor(s): Dr Leisa Armstrong and Dr T. M. Sazzad

Project Level: Masters / PhD

Funding: Applicant should apply for ECU HDR or RTP Scholarship

Start date: Any

Skip to top of page