Magnetic resonance console simulator

HOW STUDENTS LEARN

Understand, operate, practice and correct

Four materials that complete each other. Each exam report points to what to study again.

1 · UNDERSTAND

Radiology review and MRI physics

Why each parameter changes the image, with demonstrations where the student moves the value and sees the effect.
2 · OPERATE

Operator manual

The console workflow step by step, with a video guide for the first session.
3 · PRACTICE

Console

The whole exam, with a patient who moves, tires of breath-holds and squeezes the alarm bulb, and with emergencies.
4 · CORRECT

Exam report

Twenty-three criteria, each finding with its cause and explanation, and each student's history on the dashboard.

INTERACTIVE DEMONSTRATIONS

Physics the student can move

The review and the physics are not running text: each concept has a demonstration with controls, and the result appears right away.

Review demonstration: recovery and decay curves with TR and TE — **TR and TE deciding the contrast**Drag TR and TE on the recovery and decay curves and watch the resulting image change from T1 to T2.

Demonstration: magnetization rotating around the field, with the Larmor frequency — **Magnetization rotates around the field**Switch 1.5 T to 3.0 T and the Larmor frequency follows.

Interactive tree: what you want to highlight in the image — **From clinical goal to sequence**The student starts from what they want to highlight in the image and reaches the sequence family, with the reason for each step.

THE EXAM

The exam, in seven steps

The steps appear in the top bar of the console and can be opened in any order, but the system interlocks follow the sequence: without completed positioning there is no acquisition, and without acquisition there are no series to review or send. The protocol comes before positioning because it determines the coil — on a real console the order arrives through the worklist with the procedure, and the technologist checks protocol and coil before entering the room.

01

Registration

Data, checklist, coils

02

Protocol

Region and parameters

03

Positioning

Laser, landmark, isocenter

04

Exam

Prescription and acquisition

05

Viewer

Series review

06

Post-processing

MPR, MIP, subtraction

07

Film/PACS

Sending and closing

FIDELITY

Compared with a commercial console, aspect by aspect

The criterion is not having the same button. It is that the decision the operator makes here is the one they would make on the scanner, and that the mistake costs the same.

5 consistent 7 simplified 1 gap 3 go further than a real console of 16 aspects compared

The patient responds

A console does not model anxiety or comfort because the patient is on the other side of the glass. Here, not speaking over the intercom has a measurable price in the image, and the emergency call bell is pressed when anxiety crosses a threshold.

The incidents happen

A technologist may go through a whole career without seeing a quench. Here it happens, with the correct procedure required step by step, and the system is left without field afterwards.

The mistake is explained

The final report does not only say what went wrong: it says what would have been right and the physical reason. A console shows the artifact; it does not teach why it appeared.

See the full comparison, line by line

What the simulator keeps track of

Anatomy

Twelve body regions as three-dimensional volumes, with T1, T2 and proton density per tissue.

Patient

Motion, anxiety, comfort and breath change during the exam — and show on the room monitor.

Safety

Checklist before entering, SAR and dB/dt recomputed at every change, and refusal past the limit.

Parameters

TR, TE, matrix, FOV, turbo, acceleration: every change alters time, SAR and the image you get.

Scenarios

Fifteen cases, from a routine head to an acute stroke, each with its own patient and indication.

Emergencies

Five situations with a procedure to follow: ferromagnetic object, alarm bulb, burning, quench and stop.

Instructor panel

A second screen, where the instructor follows the patient signals and triggers incidents.

Report

Twenty-three criteria when the exam is closed, with the mark and what was left undone.

Each of these is explained in detail in the operator manual.

Level of fidelity

The simulator is faithful in operation, in the planning calculations and in contrast: the sequence of decisions, the safety interlocks, the numbers that appear before you press start and the brightness of each tissue in the image come from formulas with the same proportionalities as the real scanner. What it does not do is reconstruct an exam from the physical signal: k-space is filled from a synthetic image of the slice, not from encoded spins, and the anatomy comes from labeled atlases and procedural models, not from the patient being examined. Below, what stays at each level.

CALCULATED

Reacts to each parameter

Image contrast, from the signal equations, based on T1, T2, T2*, proton density and diffusion of 25 tissues
Geometry: the slice is sampled from a volume, so angling the grid tilts the plane and produces a true oblique slice — the localizer, being an image already acquired, stays still
Acquisition time, per sequence and for the whole protocol
SAR, with the 2.0 and 4.0 W/kg limits of IEC 60601-2-33
Signal-to-noise, voxel size and the effect of field strength
TI that nulls fat or CSF, recalculated for the field in use
Prescription: centering, angulation and coverage of the slice group
Aliasing, susceptibility and motion, each from its own cause
Saturation bands and shim volume, with effect and with cost
Gadolinium dose by weight and block by GFR
Cryostat: helium and magnet temperature rise and fall along a curve, in the quench and in the alarm triggered by the instructor, with recovery

MODELED WITH SIMPLIFICATION

Plausible behavior, not physical

dB/dt by a heuristic of bandwidth, thickness and diffusion — without a nerve stimulation model
Patient: four variables (movement, anxiety, comfort, breath-hold) with rates per scenario
Breath-hold tolerance as a ceiling in seconds, not as respiratory mechanics
Motion as amplitude and frequency applied to tissue channels (what breathes, what pulses, the whole body), without diaphragm mechanics or organ deformation
Parallel acceleration penalizes signal-to-noise by a heuristic g-factor — it grows with the factor and falls with the number of channels —, without a real parallel reconstruction behind it
AI reconstruction gains signal-to-noise and apparent sharpness by a level multiplier (off to high), with a cost in seconds of processing added to the acquisition time — without a neural network or k-space behind it
Relaxation times tabulated by field, without variation between individuals
Coil elements and channels: manual choice changes signal-to-noise by coverage and combination, without a real spatial sensitivity map behind it

REPRESENTED, NOT SIMULATED

It is on screen, but it is appearance

k-space filled from the synthetic image of the slice — matrix, bandwidth, aliasing and motion act on it —, not encoded from spins
Anatomy from atlases labeled by region, and from a procedural model where there is no atlas (supine breast, tendon and ligament), not from the patient being examined
MPR, MIP and subtraction re-render slices, without reformatting a volume that already exists for that
No real DICOM: sending to PACS is a progress bar, not a transfer
The room code pairs console and panel, but identifies nobody: who is on the other side is the account the person signed in with

WHERE THE DATA COMES FROM

Protocols

Twelve regions — brain, lumbar spine, knee, shoulder, ankle, wrist, pelvis, prostate, upper abdomen, breast, cardiac and angio — each with its own coil and indication. Changing region reloads the queue and recalculates the total time.

Sequences

Sixty-five sequences with nominal 3 T parameters: 1 mm isotropic sagittal MPRAGE, T2 TSE with an 11-echo train, FLAIR with TI of 2500 ms, DWI EPI b0/1000, SWI, STIR with TI of 180 ms, breath-hold VIBE, TOF. Each one keeps its original values, and it is against them that SAR and time move when you edit.

Library

Ninety-one programs behind the twelve protocols: the sixty-five routine sequences plus twenty-six alternatives, other paths to the same clinical question. STIR and Dixon suppress fat without depending on shim, but cost signal or time; readout-segmented diffusion greatly reduces EPI distortion and takes three to six times longer; a 3D volume does away with the angled group and triples the acquisition; time-of-flight angiography does away with gadolinium and takes almost seven minutes against just over one. The operator adds the program to the exam card in the position they want, as in the protocol tree of a real console, and the sequence already there stays where it was — deleting, skipping or keeping both is the operator's decision, and it is recorded in the session log.

Scenarios

Fifteen, and each one is a set of numbers, not a text: restlessness, breath-hold tolerance, initial anxiety, presence of an implant, GFR and, in two of them, a time ceiling — fifteen minutes in claustrophobia, ten in acute stroke. That is what makes the same protocol behave differently.

Monitoring

Electrocardiogram as a continuous trace, sweeping the screen like a bedside monitor, with P wave, QRS complex and T wave in lead II proportion. Trace, bpm number and beep come from the same beat generator, so in atrial fibrillation the spikes are truly irregular. Next to it, oxygen saturation, which only exists with the patient at isocenter — the oximeter is fiber-optic, because a copper wire inside the magnet forms a loop and burns — and which falls while the breath-hold is sustained.

Cryostat

The quench brings down the field and takes the cryostat with it: helium falls from 82% to zero in twelve seconds and the magnet rises from 4.1 K to over 240 K, with the gauges following live. The gradient coil does not heat up — it works at room temperature with water cooling —, but goes offline with the system; and the room starts to indicate low oxygen, which is the real risk in a quench.

Image

Twenty-five tissues with T1 by field, T2, T2*, proton density and diffusion coefficient, drawn by region and plane. Brightness comes from the signal equations; spectral saturation leaves 12% of residual fat, and noise enters in the proportion of 1 ÷ signal-to-noise.

Artifacts

Seven effects, each with its cause: motion ghosting in the phase direction, aliasing, susceptibility distortion, anatomy outside the slab, fat saturation failure, spike and zipper. Reducing the phase FOV below the body diameter makes aliasing appear in the next image.

Interlocks

Eight-item checklist, coil compatible with the region, table at isocenter, transfer height, pre-scan per sequence and First Level confirmation. Guided mode — on by default — turns the checklist into a requirement to start the acquisition.

Report

The training report is generated from the session log: each finding gives what happened, what would have been right and why, followed by the table of series with the parameters used. The student can export it to hand in to the instructor, and the exam result is recorded in their account — this is what builds the history of the performance dashboard, exam by exam.

WHAT THIS MEANS IN PRACTICE

It is for training operating decisions: choosing the protocol, anticipating the cost of each adjustment, respecting the limits, understanding why a parameter changes contrast and recognizing the artifact by its cause. It is not for validating a clinical protocol, predicting the behavior of a specific scanner or replacing supervised time in the room.

Comparison with commercial consoles

By capability family, against what the four major manufacturers — Siemens, GE, Philips and Canon — offer in common at this level. Product names only appear when the feature is known by them; the comparison is not item by item against any specific scanner.

Family	Commercial console	The simulator	Verdict
Worklist	DICOM Modality Worklist from the RIS, emergency patient, record merging.	Simulated worklist with accession, indication, referring physician and priority. No emergency patient.	Consistent
Protocol tree	Hierarchical tree by region and scanner, site protocols, versioning, dragging sequences between exams.	Flat list of twelve regions with a library of ninety-one programs and positioned insertion in the exam card.	Simplified
Graphic prescription	Prescription on three simultaneous localizers, automatic alignment to anatomy (AutoAlign, SmartExam), reference copy between series, tracking.	Prescription on sagittal and coronal localizers, with a check of shift, angulation and coverage against the region reference; geometry copy for 2D sequences.	Consistent
Automatic adjustments	Frequency, transmitter gain, receiver gain, higher-order shim, coil element selection.	Automatic frequency, transmitter and volume shim, plus element check; element selection and receiver gain adjustable by hand. No higher-order shim or automatic selection.	Simplified
Physiological gating	ECG, peripheral pulse and respiratory belt; trigger delay, acceptance window, arrhythmia rejection, diaphragm navigator.	ECG, peripheral pulse and respiratory navigator, with trigger delay, acceptance window and arrhythmia rejection — a narrow window loses more beats when the rhythm is irregular. No respiratory belt.	Consistent
Contrast injector	Volume, flow rate, saline, delay; test bolus and bolus tracking (CARE Bolus, SmartPrep, BolusTrak).	Volume, flow rate, saline, delay, test bolus that measures circulation time, tracking that triggers the series when the contrast arrives, and a block for GFR below 30.	Consistent
Real-time acquisition	Adjustment of plane, TR and angle during acquisition; fluoroscopy for manual angiography triggering.	Absent. The acquisition runs to the end or is interrupted.	Gap
Reconstruction	Deep learning reconstruction with adjustable denoise level (AIR Recon DL, Deep Resolve, SmartSpeed, AiCE); compressed sensing acceleration.	AI reconstruction level as a parameter, with signal-to-noise gain and apparent sharpness against processing time, and a warning at the high level for smoothing noise aggressively. iPAT as linear acceleration. No real neural network or compressed sensing.	Simplified
Post-processing	MPR, MIP, subtraction, ADC maps, perfusion and fusion, with parameters and real results.	The four tasks exist with thickness, increment and rotation parameters, but produce a schematic derived series.	Simplified
Viewing	Full DICOM viewer, window presets, hanging protocols, ROI statistics.	Seven tools — window, zoom, pan, distance, ROI, angle and cine — in a configurable grid.	Simplified
Sending and documentation	C-STORE to PACS, printing, exam record, structured report.	Simulated transfer to PACS and printer.	Simplified
RF safety	Normal, First Level and Second Level mode; B1+rms declared for checking against the MR Conditional implant sheet; nerve stimulation model.	Normal and First Level with a block at 4.0 W/kg; dB/dt with a 100 T/s limit; B1+rms calculated and checked against the implant sheet, with an implant mode that applies the sheet limits. No Second Level.	Consistent
Coils	Automatic detection, element selection, coil combination, mixed set-ups.	Nine coils with manual choice and port connection; elements switched on and off with a signal-to-noise penalty for incomplete coverage, and adjustable receiver gain. No automatic detection or combination between different coils.	Simplified
Emergencies	Quench button, emergency stop, patient call bell, monitoring.	Five incidents with a step-by-step procedure, drawn during acquisition and with consequences on the system state — plus five additional incidents through the instructor panel.	Goes further
Patient behavior	Does not exist. The patient is real.	Movement, anxiety, comfort, breath-hold tolerance and call bell, evolving with what the operator does.	Goes further
Operator assessment	Does not exist. The console does not assess who operates it.	Twenty-three criteria in the closing report, each finding with cause and physical explanation, a log of the session decisions and a performance dashboard with each student's history.	Goes further

THE CRITERION USED ABOVE

It is not "does it have the button?". It is whether the decision the operator makes here is the same one they would make on the scanner, and whether the price of getting it wrong is the same. A feature can be missing with no harm, if the decision it represents does not exist in the simulated flow; and it can be present and teach nothing, if the mistake costs nothing.

WHAT THIS MEANS

The gaps in the previous section matter to the extent that they reduce the fidelity of the workflow — not because feature parity with a commercial console is missing. The product is in these three rows, not in the ones that are missing.

AMONG SIMULATORS

Compare the solutions

While other MRI simulators pursue image fidelity, ELARION pursues acquisition fidelity: what the operator plans, decides and does during a real exam. The planned geometry changes the slice, the injector and the patient react, and the instructor can trigger incidents.

What the student trains	ELARION	CorsmedPHYSICS SIMULATION	CLMRIRECORDED SERIES	ScanLabMRPARAMETER LABS
The planned geometry changes the image	YesSlices generated from the 3D volume at the planned position, thickness and angle	YesImage simulated from the anatomy, according to the vendor	NoChanging thickness or angle returns the same recorded series	PartialThe page says the student sees the slices the prescription produced
Contrast injector	YesLine test, bolus, dose by weight and lockout by GFR	NoNot in the public material	NoNot identified	NoNot on the page
Patient and environment that react	YesMotion, anxiety, breath-hold, alarm bulb, ECG and SpO2; quench and cryostat	PartialMotion and artifacts in the image, no live vital signs	No	No
The instructor triggers incidents during the exam	YesInstructor panel on a second screen, with incidents triggered live	NoClasses and assessment in an LMS; the instructor does not interfere in the exam	No	NoClass reports and instructor office hours; no intervention in the exam
Safety and emergencies	YesSAR, dB/dt, implant sheet and quench with the procedure assessed	PartialSAR calculated at acquisition	NoNo SAR or emergencies identified	PartialSafety quiz before the exam; no SAR or emergencies
The whole exam, from RIS to PACS	YesWorklist, registration, positioning, planning, viewer and sending	PartialPatient, planning and protocols; little RIS and PACS	PartialRegistration, queue, viewer and film, no worklist	NoTrains slice planning on its own; registration and PACS do not appear
Image formation	PartialContrast from signal equations; aliasing, chemical shift and motion through k-space	YesPhysics simulation from the anatomy, according to the vendor	NoRecorded DICOM series, the same for any parameter	PartialContrast and resolution labs; the page does not say how the image is formed

Based on the public material of Corsmed (September 2026), ScanLabMR and CLMRI (October 2026). The trademarks belong to their owners.

FOR TEACHERS AND INSTITUTIONS

The instructor follows each student and the class

Every assessed exam goes to the dashboard, and the instructor can trigger incidents during the exam.

Performance dashboard with each student's history and the class progress
Timeline of each exam: every parameter changed, the warnings and the acquisitions, with the time
Instructor panel to trigger incidents while the student operates

How to start

Start with the radiology review: it is a prerequisite for the manual and explains the reasons behind the numbers the console asks you to check, in many rich, interactive demonstrations. Then the operator manual, which describes the workflow step by step. Only then the console.

Every closed exam goes into the performance dashboard, where the instructor follows the progress of each student and of the class. The console was designed for a wide screen — from 1480 px wide it appears in full, with no horizontal scrolling — and the room and intercom audio need speakers or headphones.

FIRST SESSION — SHORT PATH

In the worklist, open the exam of Maria Aparecida (fictional patient), routine brain, tick the whole checklist, confirm the brain protocol, move the table to isocenter, run the pre-scan, acquire the localizer and a T2 sequence, send to PACS and close the exam — it is closing that generates the score and the report. It takes about ten minutes and covers the whole cycle. Then, on the second pass, raise the TE of the T2 and watch the contrast flip, or change the field to 1.5 T in the brand menu and watch SAR drop. The operator manual describes each step in detail.

A magnetic resonance console to train without the scanner

Understand, operate, practice and correct

Radiology review and MRI physics

Operator manual

Console

Exam report