2016 AAPM Annual Meeting
Back to session list

Session Title: Challenges in Radiomics and Big Data
Question 1: A high correlation coefficient implies good agreement..
Reference:JM Bland, DG Altman, Lancet i: p307-10, 1986
Choice A:True
Choice B:False
Question 2: Which of the statements is true?
Reference:*N Altman, M Krzywinski, Nature Methods, 12, p899-900, 2015
Choice A:Correlation is independent of sample size and noise.
Choice B:When the number of features is small with respect to the sample size, large but spurious correlations frequently occur
Choice C:When there is a large number of observations, small and substantively unimportant correlations may appear statistically significant
Choice D:All of the above.
Question 3: In a replication study of the same size as the original study...
Reference:Lazzeroni et at, Molecular Psychiatry, 19, 1336-40, 2014
Choice A:The performance of the originally identified predictive markers is expected to be same.
Choice B:The obtained p-values are expected to be the same.
Choice C:All of the above.
Choice D:None of the above.
Question 4: Two studies identified important genes associated with breast cancer survival; the first group identified 76 genes, and the second one identified 70. How many genes did they have in common?
Reference:Ein-Dor et at, PNAS, 103, 5923-28, 2006
Choice A:70
Choice B:30
Choice C:3
Choice D:None
Question 5: “Big data” will automatically lead to a better understanding of the fundamental questions of basic and applied science (incl. radiomics).
Reference:RJ Gillies, PE Kinahan, and H Hricak, Radiology, 278, 563-77, 2016
Choice A:True.
Choice B:False.
Question 6: In a retrospective systematic review of studies on the use of PET or CT texture features in assessment of prognosis/clinical outcome, how many of the 9 studies’ results remained statistically significant after proper statistical re-analysis?
Reference:A Chalkidou, MJ O’Doherty, PK Marsden, PLoS ONE 10(5): e0124165, 2015
Choice A:All 9
Choice B:6
Choice C:3
Choice D:None
Back to session list