Q.1 Which imaging modality uses X-rays to create images of the inside of the body?
CT
MRI
Ultrasound
X-ray
Explanation - X-rays are electromagnetic radiation that can pass through the body; different tissues absorb X-rays differently, creating contrast in the image.
Correct answer is: X-ray
Q.2 What does MRI stand for?
Magnetic Resonance Imaging
Mechanical Resonance Imaging
Magnetic Radiation Imaging
Mechanical Radiation Imaging
Explanation - MRI uses the interaction of magnetic fields and radio waves with protons in the body to generate images.
Correct answer is: Magnetic Resonance Imaging
Q.3 Which of these is a non‑ionizing imaging technique?
CT
X‑ray
Ultrasound
PET
Explanation - Ultrasound uses high‑frequency sound waves, which are not ionizing radiation.
Correct answer is: Ultrasound
Q.4 In a CT scan, what is the main source of the image?
Sound waves
Radioactive tracer
X‑ray beams
Magnetic fields
Explanation - CT images are generated from X‑ray projections taken around the patient.
Correct answer is: X‑ray beams
Q.5 Which imaging device uses sound waves?
CT
X‑ray
Ultrasound
MRI
Explanation - Ultrasound imaging sends and receives acoustic waves to form an image.
Correct answer is: Ultrasound
Q.6 What is the purpose of a contrast agent?
To make images darker
To increase sound intensity
To highlight specific tissues
To reduce radiation dose
Explanation - Contrast agents increase the difference in signal between tissues, improving visibility.
Correct answer is: To highlight specific tissues
Q.7 What is the name of the process that reconstructs an image from many X‑ray projections?
Filtering
Reconstruction
Amplification
Transmission
Explanation - Reconstruction algorithms assemble the raw projection data into a 2‑D or 3‑D image.
Correct answer is: Reconstruction
Q.8 Which modality is commonly used to detect bone fractures?
Ultrasound
MRI
CT
PET
Explanation - CT provides high‑contrast bone images, making fractures easy to see.
Correct answer is: CT
Q.9 Which imaging technique is safest for pregnancy?
X‑ray
CT
Ultrasound
PET
Explanation - Ultrasound does not involve ionizing radiation, making it safe during pregnancy.
Correct answer is: Ultrasound
Q.10 In a CT scan, the patient lies on a table that moves through a ring of X‑ray tubes. This statement is:
True
False
Only in MRI
Only in Ultrasound
Explanation - CT scanners use a rotating gantry with X‑ray source and detector while the patient table travels.
Correct answer is: True
Q.11 What is the purpose of filtering in image reconstruction?
To increase brightness
To remove noise
To sharpen edges
To reduce image size
Explanation - Filters, such as the Ram‑Lak filter, reduce low‑frequency components that contribute to blurring.
Correct answer is: To remove noise
Q.12 Which part of the body is best imaged by MRI?
Bones
Muscles
Brain
Lungs
Explanation - MRI provides excellent soft‑tissue contrast, especially for brain imaging.
Correct answer is: Brain
Q.13 What type of radiation does PET use?
X‑rays
Gamma rays
Ultrasound
Magnetic fields
Explanation - PET detects gamma photons produced by positron‑electron annihilation events.
Correct answer is: Gamma rays
Q.14 Which imaging modality uses a radioactive tracer to show how organs are working?
CT
PET
Ultrasound
X‑ray
Explanation - PET imaging visualizes metabolic activity using radiotracers.
Correct answer is: PET
Q.15 Which image processing technique helps to reduce grainy appearance in images?
Sharpening
Blurring
Noise reduction
Cropping
Explanation - Noise‑reduction algorithms filter out random fluctuations while preserving image detail.
Correct answer is: Noise reduction
Q.16 The term "resolution" in imaging refers to what?
Size of the image file
Level of detail that can be seen
The speed of scanning
The amount of radiation used
Explanation - Resolution measures the smallest distinguishable structure in an image.
Correct answer is: Level of detail that can be seen
Q.17 Which imaging method is most useful for viewing soft tissues?
X‑ray
CT
Ultrasound
MRI
Explanation - MRI provides superior soft‑tissue contrast without ionizing radiation.
Correct answer is: MRI
Q.18 Which imaging technique uses a rotating table to acquire data from many angles?
Ultrasound
PET
MRI
CT
Explanation - CT scanners rotate the X‑ray source and detector around the patient while the table moves.
Correct answer is: CT
Q.19 Explain the role of the Fourier slice theorem in CT image reconstruction.
It states that the 1D Fourier transform of a projection equals the 2D Fourier transform of the image
It provides a direct back projection
It eliminates noise
It is used in PET only
Explanation - The theorem links each 1D projection to a slice of the 2D Fourier domain, enabling filtered back projection.
Correct answer is: It states that the 1D Fourier transform of a projection equals the 2D Fourier transform of the image
Q.20 Which iterative reconstruction technique solves the reconstruction problem by minimizing the L2 norm?
Simultaneous algebraic reconstruction technique (SART)
Simultaneous iterative reconstruction technique (SIRT)
Maximum a posteriori (MAP)
Least‑squares reconstruction
Explanation - Least‑squares methods minimize the sum of squared differences between measured and predicted data.
Correct answer is: Least‑squares reconstruction
Q.21 What is the purpose of the "Hounsfield unit" scale?
To quantify signal intensity in MRI
To provide a standardized attenuation measurement in CT
To measure diffusion coefficients
To calibrate ultrasound depth
Explanation - Hounsfield units (HU) map X‑ray attenuation to a scale where water is 0 HU and air is −1000 HU.
Correct answer is: To provide a standardized attenuation measurement in CT
Q.22 In MRI, which factor primarily determines the signal‑to‑noise ratio (SNR)?
Magnetic field strength
Echo time
Repetition time
Field of view
Explanation - Higher magnetic fields increase proton polarization, thus increasing SNR.
Correct answer is: Magnetic field strength
Q.23 Which type of regularization is commonly used to promote sparsity in compressed sensing MRI?
L2 regularization
Total variation
L1 regularization
Ridge regression
Explanation - L1 regularization encourages sparse solutions, which is the core of compressed sensing.
Correct answer is: L1 regularization
Q.24 Which reconstruction method is used in PET to handle incomplete data?
Filtered back projection
Maximum likelihood expectation maximization (MLEM)
Conjugate gradient
Fourier transform
Explanation - MLEM iteratively refines the image to fit the measured PET data, handling incomplete or noisy data.
Correct answer is: Maximum likelihood expectation maximization (MLEM)
Q.25 What is the main difference between deterministic and stochastic image reconstruction methods?
Deterministic uses fixed models; stochastic uses random processes
Deterministic uses iterative methods; stochastic uses Fourier transform
Deterministic increases noise; stochastic reduces it
Deterministic uses fixed models; stochastic uses fixed models
Explanation - Deterministic algorithms have a fixed mapping; stochastic ones incorporate random noise models or priors.
Correct answer is: Deterministic uses fixed models; stochastic uses random processes
Q.26 Which MRI sequence provides the best contrast for detecting acute ischemic stroke?
T1‑weighted
T2‑weighted
Diffusion‑weighted imaging (DWI)
Proton density
Explanation - DWI highlights restricted water diffusion in infarcted tissue, making it the most sensitive for acute stroke.
Correct answer is: Diffusion‑weighted imaging (DWI)
Q.27 What does the term "k‑space" refer to in MRI?
The space of spatial frequencies
The space of time points
The space of echo times
The physical location of the scanner
Explanation - k‑space is the Fourier domain where raw MRI data are sampled before image reconstruction.
Correct answer is: The space of spatial frequencies
Q.28 Which of the following is NOT a typical artifact in PET imaging?
Random coincidences
Scatter
Motion
Beam hardening
Explanation - Beam hardening is an X‑ray CT artifact, not seen in PET imaging.
Correct answer is: Beam hardening
Q.29 In MRI, what is the benefit of using a parallel imaging method like GRAPPA over traditional SENSE?
It requires fewer calibration lines
It improves noise amplification
It requires no additional hardware
It is only applicable to CT
Explanation - GRAPPA uses autocalibrated data, reducing the need for separate calibration scans.
Correct answer is: It requires fewer calibration lines
Q.30 Which of the following best describes the advantage of a 3D convolutional neural network (CNN) in volumetric CT reconstruction?
It processes each slice independently, ignoring inter‑slice correlations
It captures spatial context across slices, improving continuity and reducing artifacts
It reduces computational load by using 2D kernels
It eliminates the need for calibration
Explanation - 3D CNNs learn volumetric features, leading to smoother reconstructions and fewer slice‑by‑slice inconsistencies.
Correct answer is: It captures spatial context across slices, improving continuity and reducing artifacts
Q.31 In the context of iterative reconstruction, what is the "data fidelity term"?
It penalizes deviation from measured data
It introduces noise
It smooths the image
It scales the radiation dose
Explanation - The fidelity term ensures the reconstructed image remains consistent with the acquired data.
Correct answer is: It penalizes deviation from measured data
Q.32 Explain the role of the "regularization parameter" in penalized reconstruction methods and the trade‑off it controls.
It controls between data fidelity and prior penalty
It controls noise
It controls dose
None of the above
Explanation - A larger parameter increases regularization (smoothing) at the expense of fidelity, and vice versa.
Correct answer is: It controls between data fidelity and prior penalty
Q.33 Which advanced reconstruction technique uses a stochastic gradient descent approach for PET image reconstruction?
Ordered subsets expectation maximization (OSEM)
Simultaneous algebraic reconstruction technique (SART)
Gradient descent with L1 regularization (GD‑L1)
Back projection
Explanation - GD‑L1 employs stochastic gradient descent to minimize an objective function with an L1 penalty, suitable for sparse PET data.
Correct answer is: Gradient descent with L1 regularization (GD‑L1)
Q.34 Which of the following best summarizes the impact of iterative model‑based reconstruction (IMBR) on diagnostic accuracy in low‑dose CT?
It reduces image quality and diagnostic confidence
It improves lesion detectability without increasing noise
It eliminates the need for contrast agents
It only speeds up reconstruction time
Explanation - IMBR incorporates physics and statistical models, yielding high‑quality images at reduced dose.
Correct answer is: It improves lesion detectability without increasing noise
Q.35 Which statement best reflects the potential of deep learning‑based reconstruction in replacing traditional iterative methods?
It guarantees perfect reconstruction with no artifacts
It offers faster reconstruction and can learn complex priors, but risks overfitting and lack of interpretability
It eliminates the need for raw data acquisition
It always reduces radiation dose
Explanation - Deep learning models can be trained to approximate reconstruction, but they may be opaque and require large data.
Correct answer is: It offers faster reconstruction and can learn complex priors, but risks overfitting and lack of interpretability
Q.36 Which of the following is a key factor limiting the clinical adoption of deep learning reconstruction methods in medical imaging?
Interpretability
Generalizability across scanners
Computational cost
All of the above
Explanation - Clinical deployment demands explainability, robustness, and efficient computation.
Correct answer is: All of the above
Q.37 Which of the following statements best captures the ethical considerations of AI‑driven reconstruction algorithms?
They have no ethical issues if they improve quality
They raise concerns about bias, accountability, transparency, and patient data privacy
They are always transparent and explainable
They eliminate the need for radiologists
Explanation - AI systems can inherit biases, lack explainability, and process sensitive data.
Correct answer is: They raise concerns about bias, accountability, transparency, and patient data privacy
Q.38 Which statement correctly describes transfer learning in medical imaging reconstruction?
Training a new model from scratch for every device
Reusing a pre‑trained model and fine‑tuning on device‑specific data, reducing training time
Ignoring device‑specific characteristics
Only applicable to MRI, not CT
Explanation - Transfer learning leverages knowledge from related tasks, accelerating adaptation to new hardware.
Correct answer is: Reusing a pre‑trained model and fine‑tuning on device‑specific data, reducing training time
Q.39 Which of the following best explains the trade‑off between spatial resolution and noise in iterative reconstruction?
Higher resolution always reduces noise
Higher resolution increases noise
Lower resolution increases noise
No trade‑off
Explanation - Refining resolution captures fine detail but also amplifies statistical noise.
Correct answer is: Higher resolution increases noise
Q.40 Which statement best explains the benefit of joint sparsity regularization in multi‑contrast MRI?
It treats each contrast independently, increasing computational cost
It leverages shared sparsity patterns across contrasts, improving reconstruction quality
It eliminates the need for any regularization
It only works for CT images
Explanation - Joint sparsity imposes a common sparsity structure across different contrast images, leading to better results.
Correct answer is: It leverages shared sparsity patterns across contrasts, improving reconstruction quality
Q.41 Which of the following best summarizes the future direction of medical image reconstruction?
Exclusive reliance on physics‑based models
Exclusive reliance on data‑driven models
Hybrid approaches combining physics‑based priors with deep learning to balance interpretability, performance, and generalizability
Stopping development
Explanation - Hybrid methods aim to exploit the strengths of both physics modeling and learned priors.
Correct answer is: Hybrid approaches combining physics‑based priors with deep learning to balance interpretability, performance, and generalizability