1. Radiologists can be easily familiar with: a) the Information Theory (IT) by C. Shannon (1916-2001), which is widely used in engine search technology, telecommunication and biology; b) the definition of “information” and main measures of information provided by IT.
2. Radiological diagnosis can be formalized as a flux of information between the disease and the reader through the imaging procedure, in accordance with the Shannon’s Binary Transmission Channel model. The flux of information is the “Mutual Information” (MI).
3. The Area Under the Curve (AUC)-ratio is an MI-based measure that can express the diagnostic performance of an imaging technique globally, using a single index that is independent from the probability of the disease.
1. Basic concepts of statistics and IT
2. Formalization of radiological diagnosis in terms of IT:
- From the 2x2 table to MI
- From MI to the AUC-ratio
3. How the AUC-ratio refines conventional measures of diagnostic accuracy: examples from radiological literature
4. Applicability and usefulness in clinical activity and/or research.
Girometti, R,
Fabris, F,
The Radiological Diagnosis in the Light of Information Theory: A Perspective. Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL.
http://archive.rsna.org/2014/14004866.html
