What is it about?
We developed an automated system to assess haemoglobin electrophoresis images used in thalassaemia testing. The dataset contained 524 patient images from 103 strips, and seven deep-learning models were compared after automatic lane extraction. InceptionV3 performed best with 95.8% accuracy; the smaller MobileNetV2 produced similar accuracy with lower processing time.
Featured Image
Photo by Cassi Josh on Unsplash
Why is it important?
Electrophoresis images can be difficult to interpret where specialist haematologists are scarce. A reliable automated tool could support rapid screening, reduce workload and make decision support more accessible, although clinical validation across larger and more diverse datasets is still needed.
Perspectives
From my perspective, this work matters because Electrophoresis images can be difficult to interpret where specialist haematologists are scarce. I hope the findings support further research and informed, evidence-based decisions.
Yasar Yousafzai
Khyber Medical University
Read the Original
This page is a summary of: Deep Learning Assisted Automated Assessment of Thalassaemia from Haemoglobin Electrophoresis Images, Diagnostics, October 2022, MDPI AG,
DOI: 10.3390/diagnostics12102405.
You can read the full text:
Contributors
The following have contributed to this page
