eduKateSG · Why Science?
Turn magnetic-resonance signals into trustworthy translational evidence—connecting spin physics, acquisition, safety, anatomy, function, diffusion, spectroscopy, reconstruction, quality control, statistics and clinical questions
A fact-checked guide to the NUS Centre for Translational MR Research, 3T MRI, multimodal imaging, brain analysis and responsible imaging literacy.
NUS TMR states that it performs research for Singapore’s translational-medicine needs and brings imaging and analysis teams together so acquired data can match analysis paradigms.
Its official imaging page describes two 3T MR scanners, multimodal whole-body research and capabilities across neuro, body, cardiac, breast and paediatric imaging; its analysis page lists several brain-analysis pipelines.
Did you know? MRI does not take an ordinary photograph. It uses controlled magnetic fields and radiofrequency signals, then reconstructs measurements into images.
TMR and NUS control current research capabilities and participation. Healthcare professionals control clinical interpretation, and this guide never diagnoses from scans. Source status was checked 11 November 2026.
Chapter index
Chapters 1–9: translation questions, MR signal, spatial encoding, contrast, sequences, safety, coils, resolution, motion
Chapters 10–17: quality control, anatomy, function, connectivity, diffusion, spectroscopy, reproducibility, interpretation and evidence receipts
A translational study links a biological or clinical question to a measurable MR quantity and a decision-relevant outcome. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 1 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Define population, comparator, sequence, endpoint and intended use before scanning. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
An available sequence should not dictate a vague question. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Start with a translational question” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Rewrite “scan the brain” as a testable study question. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Certain nuclei produce measurable signals when placed in a magnetic field and perturbed by radiofrequency energy. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 2 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Use vector diagrams and timing models to connect excitation with signal. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
The simplified classroom model is useful but does not describe every quantum detail. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Understand nuclear magnetic resonance” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Draw equilibrium, excitation and recovery in three frames. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
T1, T2 and related processes influence signal and tissue contrast under a chosen sequence. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 3 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Change one timing parameter in a simulation and record contrast changes. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Image brightness is sequence-dependent and is not a fixed tissue property. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Read relaxation and contrast” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Compare two fictional tissues across two parameter sets. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Gradients change magnetic field by location so frequency and phase carry spatial information. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 4 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Trace how sampling in k-space supports reconstruction. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
A reconstructed pixel combines acquisition and processing choices. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Encode position” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Map a one-dimensional signal to a simple spatial grid. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
| Evidence move | What to record | What it cannot prove alone |
|---|---|---|
| Signal question | Population, variable, unit and time window | A definition does not establish a result |
| Comparison | Baseline, comparator and uncertainty | Association alone does not establish cause |
| Decision boundary | Owner, trigger, review and escalation | A model does not replace accountable judgement |
Sequence families weight anatomy, flow, diffusion, function or chemistry differently. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 5 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Justify each sequence by the target measurement and burden. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
More sequences increase time, motion risk and multiple testing. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Choose sequences for questions” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Build a minimal fictional protocol and defend every item. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Strong static fields, changing gradients, radiofrequency energy, implants and projectiles require controlled screening. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 6 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Follow authorised screening, zoning and trained supervision. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Research enthusiasm never overrides current safety procedures. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Protect MR safety” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Classify fictional objects by questions that require expert screening. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Coils, gradients and peripheral equipment shape sensitivity, coverage and experimental capability. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 7 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Record hardware model, configuration and calibration. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
A different coil can change signal and comparability. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Match coils and hardware” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Design a hardware metadata sheet for a simulated study. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Spatial and temporal resolution trade against signal, coverage and scan duration. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 8 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Optimise the whole protocol against the primary endpoint. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Finer voxels do not automatically produce more reliable conclusions. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Balance resolution, time and signal” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Compare three fictional protocols on a trade-off chart. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
| Evidence move | What to record | What it cannot prove alone |
|---|---|---|
| Acquisition question | Population, variable, unit and time window | A definition does not establish a result |
| Comparison | Baseline, comparator and uncertainty | Association alone does not establish cause |
| Decision boundary | Owner, trigger, review and escalation | A model does not replace accountable judgement |
Head movement, breathing, heartbeat and swallowing can create structured artefacts. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 9 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Prevent, monitor, model and report motion rather than silently discard it. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Correction cannot recover information that was never measured. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Control motion and physiology” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Inspect a simulated time series with a motion spike. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Quality control checks scanner stability, artefacts, coverage, signal and protocol adherence. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 10 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Use phantoms, automated metrics and trained visual review with thresholds. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Passing one metric does not guarantee that data answer the study question. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Run quality control” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Create a before-scan and after-scan checklist. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Morphometry can estimate volume, thickness and shape after segmentation and registration. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 11 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Inspect overlays and validate pipelines across relevant populations. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
A group-average anatomical difference does not diagnose a person. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Measure anatomy” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Find three ways segmentation could fail on a toy image. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Task fMRI relates blood-oxygen-level-dependent signals to timed conditions and models. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 12 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Pre-register contrasts and include behavioural performance. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
BOLD is an indirect haemodynamic signal rather than direct neural firing. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Study task-related function” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Build a timing diagram for a harmless cognitive task. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
| Evidence move | What to record | What it cannot prove alone |
|---|---|---|
| Analysis question | Population, variable, unit and time window | A definition does not establish a result |
| Comparison | Baseline, comparator and uncertainty | Association alone does not establish cause |
| Decision boundary | Owner, trigger, review and escalation | A model does not replace accountable judgement |
Connectivity analyses quantify statistical relationships among time series across regions. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 13 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Control motion and compare alternative parcellations and null models. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Correlation between regions does not prove direct communication or causality. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Analyse functional connectivity” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Calculate a tiny fictional correlation matrix. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Diffusion MRI uses water-motion sensitivity to infer features of tissue organisation. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 14 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Report acquisition directions, model, uncertainty and tractography rules. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Streamlines are model outputs and are not individual axons. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Map white-matter organisation” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Trace how crossing fibres can mislead a simple model. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Spectroscopy separates frequency components associated with measurable metabolites under a protocol. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 15 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Check voxel placement, shimming, water suppression and fit quality. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
A peak assignment may overlap and does not by itself establish disease. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Use MR spectroscopy” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Label quality checks on a simulated spectrum. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
Pipelines include conversion, preprocessing, modelling, statistics and visualisation. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 16 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Version code, containers, atlases and decisions; preserve provenance. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
Running the same software does not remove dataset shift or analytic flexibility. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Make analysis reproducible” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Create a pipeline receipt with inputs and outputs. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
The receipt links question, participant, protocol, hardware, quality, analysis, uncertainty and interpretation. Magnetic resonance research turns signals from nuclei in a magnetic field into carefully reconstructed measurements. At the NUS Centre for Translational MR Research, imaging and analysis teams work together so acquisition choices match the scientific question. Chapter 17 treats an image as processed evidence with a protocol, calibration history and uncertainty—not as a self-explaining photograph.
Archive approvals, versions, exclusions and update triggers. A reproducible MR study records scanner, field strength, coil, sequence, timing parameters, spatial resolution, participant preparation, quality checks and analysis version. It also states whether the objective concerns anatomy, function, connectivity, diffusion, spectroscopy or another measurable property. That receipt lets another team understand what the numbers mean and where variation may enter.
A beautiful image cannot substitute for a complete evidence chain. Keep the interpretation boundary firm. A bright or dark region can reflect tissue, motion, acquisition physics, preprocessing or statistical thresholds. Group differences do not diagnose an individual, and research imaging does not replace a clinician’s assessment. Translation requires validation, suitable comparison data, expert review and evidence that a result improves a meaningful decision.
Explore “Keep an MR evidence receipt” without clinical data. Use a grid, a blurred shape or a simulated signal to compare resolution and noise, then document every transformation. Assemble a one-page receipt for a simulated study. The activity should use invented information and should never invite learners to interpret a real person’s scan.
Did you know? Faster acquisition can reduce motion but may trade away signal, coverage or detail. Researchers therefore optimise a whole protocol rather than chase one attractive image. The same reasoning helps in secondary science: variables interact, instruments have limits, and a fair comparison changes one planned factor at a time.
Possible learning routes include physics, mathematics, computing, biomedical engineering, neuroscience, radiography, medicine and statistics. TMR’s official pages describe current capabilities, while universities and professional bodies set admission or registration rules. The article explains capability building and does not promise entry, employment or a clinical role.
| Evidence move | What to record | What it cannot prove alone |
|---|---|---|
| MR evidence question | Population, variable, unit and time window | A definition does not establish a result |
| Comparison | Baseline, comparator and uncertainty | Association alone does not establish cause |
| Decision boundary | Owner, trigger, review and escalation | A model does not replace accountable judgement |
Contents · Previous chapter · Continue in the Science Learning Hub
Primary sources and verification date
Source status for Why Science? | Centre for Translational MR Research was checked 11 November 2026. Research priorities, programmes, recruitment, facilities and participation routes can change; use the linked official owner for the current edition.
- NUS Centre for Translational MR Research — Official mission, equipment overview and current centre navigation.
- TMR Imaging — Official imaging capabilities and two 3T MR scanners.
- TMR Analysis — Official brain-analysis pipelines and capabilities.
- NUS Medicine research programmes — Official institutional description of TMR.
Continue the evidence journey
Centre for Translational MR Research science becomes useful when learners connect every image to its physics, protocol, quality checks, analysis choices and carefully bounded interpretation. Families considering science tuition or a science tutor can use the same evidence questions: what will the learner practise, what feedback will they receive, and how will progress be shown?