How to Price Your Time as a Physician Annotator
Setting your hourly rate as a doctor who annotates medical data: start from your clinical rate, add freelance costs, adjust for specialty and task, and know when to raise it.
Guides and insights on medical data, AI and research.
Setting your hourly rate as a doctor who annotates medical data: start from your clinical rate, add freelance costs, adjust for specialty and task, and know when to raise it.
Medical AI learns from labels that clinicians add first. What annotation work involves, who can do it, how it is paid and how to get started.
Brain-tumour segmentation is multi-parametric. Which MRI sequences a model needs, and how to harmonise them across scanners.
Tumour findings are rare and imbalanced. How to design a CT cohort that trains a model to catch the cases that matter.
DICOM is the clinical source of truth; NIfTI is the research workhorse. When to use each, and the conversion pitfalls to avoid.
Removing PHI means cleaning both the DICOM header and the pixels. A practical checklist plus the HIPAA/GDPR baseline.
Chest radiography is cheap and ubiquitous — and full of shortcut signals. Here is what to check before you train on a CXR dataset.
What separates a training-ready 12-lead ECG cohort from a noisy one — labels, lead configuration, class balance and compliant sourcing.
How patient data is removed from imaging studies, and what researchers should verify before training on them.
A practical guide to sourcing high-quality, well-labelled 12-lead ECG datasets for machine-learning projects.