Kavya Mohan

Advancing TB Screening: Integrating Vision Language Models and Patient Metadata

Submitted May 30, 2024

TB claims over 1.3 million lives annually, with around 30% of cases missed by current screenings and diagnostics. The shortage of radiologists further complicates timely and accurate TB screenings, often relying on subjective interpretations that can lead to missed diagnoses or unnecessary treatments, impacting patient’s health. There is a critical need for accurate detection and differentiation between active and chronic TB for effective treatment to scale TB screening globally.

Challenges in Existing TB Screening Approaches
Outdated Architectures: Utilisation of traditional architectures lacking attention mechanisms hampers the efficacy of TB screening.
Scaling Annotation: The need for extensive manual annotation further complicates the screening process, making it inefficient and prone to delays.
Limited Data Integration: Insufficient integration of diverse datasets, including patient metadata, poses a challenge in achieving comprehensive TB diagnosis and analysis

By harnessing the power of MedSAM for lung field segmentation and Vit architecture models augmented with attention mechanisms, alongside the integration of comprehensive patient metadata, we present a revolutionary approach to TB screening. This approach ensures noise reduction in radiological images while embedding crucial patient data, resulting in enhanced accuracy and differentiation between active and chronic TB cases.

Contextualization of the global TB burden and the imperative for improved screening methods.
Challenges in Existing TB Screening
Exploration of scalability issues, architectural limitations, and the absence of patient metadata integration in current screening practices.
Revolutionising TB Screening with Vision Language Models
Overview of Medsam’s role in lung field segmentation and the utilisation of Vit architecture models with attention mechanisms.
Integration of Patient Metadata
Discussion on the significance of embedding patient metadata like geo-specifics, age, and gender for comprehensive TB screening.
Demonstration of Results
Showcase of tangible outcomes achieved through the proposed approach, highlighting advancements in TB detection and differentiation.
Future Implications and Scalability
Exploration of the scalability and broader implications of integrating Vision Language Models and patient metadata in global TB screening initiatives

Attendees will gain insights into the technical challenges of TB screening and the potential of integrating Vision Language Models and advanced data integration techniques to address them. They will leave with a deeper understanding of the tech-driven solution and its implications for advancing TB diagnosis and analysis.

Kavya Mohan - Data Scientist, 5C Network


{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Hybrid Access Ticket

Hosted by

All about data science and machine learning

Supported by

Gold Sponsor

Atlassian unleashes the potential of every team. Our agile & DevOps, IT service management and work management software helps teams organize, discuss, and compl

Silver Sponsor