MORE | Spring 2022

Fine-Tuning Pre-Trained Models to Determine Model Superiority for Enhanced Medical Image Diagnosis

Health icon, disabled. A red heart with a cardiac rhythm running through it.

The biggest challenge in computer-aided diagnosis remains the slow nature of the data annotation process. No large-scale evaluation has performed a comparison on the many pre-trained models in existence. In this project, selected pre-trained models are benchmarked to compare, evaluate and determine their superiority in medical image disease detection. The potential of this research project extends to identifying and giving good recommendations for segmentation and disease detection tasks resulting in an overall enhancement and efficiency in algorithms implementation for medical image diagnosis.

Student researcher

Daniella Asare

Daniella Asare

Biomedical engineering

Hometown: Accra, Greater Accra Region, Ghana

Graduation date: Spring 2022