The Jodcast July 2025
Jul 4, 2025
Join Dr. Mel Ifran, a Postdoctoral researcher at the University of Cambridge, as she discusses her fascinating journey from radio astronomy to cancer detection. She reveals how machine learning techniques used in studying stars can revolutionize medical imaging. The conversation highlights innovative methods for analyzing tissue samples, bridging the gap between astrophysics and healthcare. Mel also shares insights on the role of branding in scientific research and the collaborative potential of machine learning across fields, making for a thought-provoking and inspiring dialogue.
AI Snips
Chapters
Transcript
Episode notes
From Telescope Pipelines To Tissue Samples
- Mel Ifran described training as a radio astronomer building data reduction pipelines and handling instrument outputs during her PhD in Manchester.
- She emphasized those skills transfer directly to processing microscope images and tissue sample intensities in biomedical work.
Running 64 Single-Dish Experiments
- Mel recounted working remotely from Cape Town on the MeerKAT SKA precursor and running 64 dishes as simultaneous single-dish experiments.
- She explained they averaged signals to get high signal-to-noise blurred maps used for cosmology rather than interferometric resolution.
Blurry Maps Enable Fast Cosmology
- Using many dishes as single instruments produces large-area, high signal-to-noise maps suited to cosmology rather than high-resolution imaging.
- That trade-off enables fast surveys of unresolved neutral hydrogen across cosmic time.