[evangress]

Summary

Adds a set of classes for medical radiology students. They follow the project's conventions: keyword-only arguments, docstrings that state units, and comments that explain the why.

Class	Covers
Radiation	Inverse square law, attenuation (half-/tenth-value layers, shielding thickness), photon wavelength, radioactive and effective half-life, equivalent/effective dose (ICRP 103 weighting factors)
Radiography	mAs, 15% kVp rule, mAs correction for a new SID, grid conversion, heat units, magnification, geometric unsharpness, detector Nyquist limit, deviation index
CT	Hounsfield units, DICOM rescale, window width/level with presets (vectorized over a whole slice), rough HU tissue lookup, DLP and effective-dose estimate
MRI	Larmor frequency, T1 recovery, T2 decay, spin-echo signal, T1/T2/PD weighting from TR/TE
Ultrasound	Wavelength, echo depth, impedance and reflection, attenuation, axial resolution, PRF depth limit, Doppler shift and velocity
Dicom	pydicom read, technical header summary (no patient fields), pixels in HU, pixel spacing, basic de-identification, PNG export

Also included:

• radiology_demo.py: one worked example per modality, plus a sample CT slice shown in abdomen, lung, and bone windows.
• pydicom dependency: its bundled anonymous sample images are used by the demo and tests.
• gui.py: the console starts with an instance of each new class.
• README: a radiology section, a table of medical-imaging packages (SimpleITK, nibabel, scikit-image, highdicom, pynetdicom, MONAI, TotalSegmentator, pyradiomics), and learning resources.

For reviewers

• Some reference values differ between textbooks: the heat-unit factor for high-frequency generators, CT window presets, MRI T1/T2 times at 1.5 T, and DLP-to-effective-dose coefficients. Comments note the sources and spread. Please have an instructor or medical physicist look these over.
• The README and docstrings mark these as teaching calculations only. Dicom.anonymize removes common identifiers, which is not full de-identification under DICOM PS3.15 or HIPAA.

Test plan

• uv run pytest: 109 passed (48 new in tests/test_radiology.py, checked against textbook answers)
• uv run radiology_demo.py runs and writes radiology.png

🤖 Generated with Claude Code