| Question 1: In order to characterize tumors, radiomics may take advantage of any/all of the following |
| Reference: | Wu et al., Radiomics and radiogenomics for precision radiotherapy, Journal of Radiation Research, 59, i25-i31, 2018
Aerts et al, Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach, Nature Communications 5:4006, 2014
|
| Choice A: | True |
| Choice B: | False |
| Question 2: Delta-radiomics |
| Reference: | Fave et al, Delta-radiomics features for the prediction of patient outcomes in non-small cell lung cancer, Scientific Reports 7:488, 1-11, 2017 |
| Choice A: | Describes a specific filter used during pre-processing of radiomics features |
| Choice B: | Describes the difference between two radiomics features calculated on the same image (same patient) |
| Choice C: | Describes the use of measured changes in radiomics features through time to characterize tumor phenotype (especially response to treatment) |
| Choice D: | Describes the difference between two radiomics features calculated on different patients |
| Question 3: Examples of avoidable flaws in radiomics experiment design include: |
| Reference: | Aerts, Data Science in Radiology: A Path Forward, Clinical Cancer Research 24(3), 532-534, 2018 |
| Choice A: | Evaluating hundreds of parameters, using only a few image samples |
| Choice B: | Training and validating models on the same data |
| Choice C: | Failure to consider multiple hypothesis testing |
| Choice D: | All of the above |
| Question 4: Radiomics features should be |
| Reference: | Kumar et al. Radiomics: the process and the challenges. Magnetic Resonance Imaging, 2012 Nov, 30(9), 1234 –1248. |
| Choice A: | Highly reproducible with a large dynamic range |
| Choice B: | Highly dependent on imaging protocol |
| Choice C: | Strongly correlated to other radiomics features |
| Choice D: | Strongly correlated to the volume of the tumor |
| Question 5: In a typical radiomics workflow, image segmentation |
| Reference: | Gillies et al. Radiomics: Images are more than pictures, they are data. Radiology, 2016 Feb, 278(2), 563-577. |
| Choice A: | Is not important for calculating features |
| Choice B: | Defines the pixels which will be used to calculate the radiomics features |
| Choice C: | Determines whether the image will be processed before feature calculation |
| Choice D: | Should be done manually by one physician |
| Question 6: The co-occurrence matrix is |
| Reference: | Haralick et al. Textural features for image classification. IEEE Transactions on Systems, Man, and Cybernetics, 1973 Nov, SMC-3 (6), 610-621. |
| Choice A: | A new method for calculating imaging features specifically designed for medical images |
| Choice B: | Based on the assumption that image texture information is contained in the average spatial relationship which the gray tones in the image have to one another |
| Choice C: | A set of features designed to measure the relative shape of the tumor |
| Choice D: | Used to calculate first-order histogram based features |
| Question 7: Discretization (the resampling of image intensity values) |
| Reference: | Leijenaar et al. The effect of SUV discretization in quantitative FDG-PET radiomics: The need for standardized methodology in tumor texture analysis. Scientific Reports, 2015 Aug, 5(11075), 1-10. |
| Choice A: | Reduces an infinite possible number of intensities to a finite set |
| Choice B: | Effectively reduces image noise |
| Choice C: | Is only useful for CT images and does not impact PET images |
| Choice D: | Both A and B |
| Choice E: | Both B and C |
| Question 8: Application of standard statistical approach in radiomics can: |
| Reference: | A comparative study of machine learning methods for time-to-event survival data for radiomics risk modelling. Leger S., et al. Scientific Reports 2017 Oct 16;7(1):13206. doi: 10.1038/s41598-017-13448-3. |
| Choice A: | Increase the power of study |
| Choice B: | Demonstrate the limitation of the study |
| Choice C: | Reduce measurement error |
| Choice D: | Provide benchmark for machine learning techniques |
| Question 9: Which of the following methods were NOT appropriate for right-censored survival data: |
| Reference: | Survival Analysis: Techniques for Censored and Truncated Data, JP Klein & ML Moeschberger, ISBN 978-0-387-21645-4 |
| Choice A: | Two-sample t-test |
| Choice B: | Cox proportional hazard regression |
| Choice C: | Log-rank test |
| Choice D: | Dynamic prediction |
| Question 10: Which of the following method can reduce the risk of model over-fitting: |
| Reference: | Predictive Inference. Geisser S. ISBN 978-0-387-21645-4 |
| Choice A: | Cross-validation |
| Choice B: | Cox proportional hazard regression |
| Choice C: | K-means clustering |
| Choice D: | Dynamic prediction |
