Premium
TU‐B‐207B‐01: Challenges in Radiomics and Big Data
Author(s) -
Drukker K.
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4957462
Subject(s) - radiogenomics , overfitting , big data , radiomics , computer science , sample size determination , pipeline (software) , multiple comparisons problem , data science , statistics , artificial intelligence , data mining , mathematics , artificial neural network , programming language
The general learning objectives are to get a better understanding of • The pitfalls along the radiomics/radiogenomics (imaging genomics) pipeline • The crucial role of statistics in the design and evaluation of radiomics/radiogenomics phenotypes and systems • The special challenges associated with big data1. Introduction a. Basic statistical concepts b. Association, correlation, and causality c. Know when your numbers are significant2. Reproducibility a. Replication studies i. The extent of the problem ii. Curse of dimensionality, overfitting, multiple hypothesis testing iii. Sample size, P‐value variability3. “Big Data” a. “Big data” vs. “medium data” vs. “small data” b. Training and testing data sets c. HarmonizationKD receives royalties from Hologic