.svg){kind=small}
Subscribe to our MRI newsletter
Magnetic Resonance Imaging (MRI) is a crucial technology in modern medicine. By producing detailed images of organs and tissues, MRI helps doctors accurately diagnose a wide range of medical conditions.
However, the high cost and limited availability of MRI scanners makes it hard for students and technologists to gain hands-on training with real MRI equipment.
This training gap is what Corsmed’s online MRI simulator is designed to solve.
Accessible from any laptop, Corsmed’s simulator lets users practice MRI scanning in a virtual environment that closely mirrors a real MRI machine. Users can control the same settings, modify parameters, and view resulting images that accurately reflect those from a real scanner.
But how does Corsmed’s MRI simulator actually work?
How can a digital program replicate the complex processes of a physical MRI machine?
That is what this article answers. Below, we break down each step a real MRI scanner goes through to produce an image. And for each step, we explain how Corsmed is able to replicate all steps using the latest MRI simulation technology.
The first step of a real MRI scan is that a patient enters the scanner.
Depending on what body region needs to be imaged, the patient will be positioned differently in the scanner.
With Corsmed, there is obviously no physical patient.
Instead, the simulator uses a “digital patient” – an extremely detailed 3D model that mirrors the anatomy of a real human patient.
Just as images on the web are created from tiny 2D pixels, that each are assigned a different color, Corsmed’s patient models are created from tiny 3D pixels, that each are assigned a different tissue type.
These detailed 3D patient models allows Corsmed to simulate all the different types of MRI signals (e.g., T1 or T2-weighted) of a real patient – and with sub-millimeter detail.
MRI scanners are equipped with superconducting magnets that create a powerful magnetic field, usually at 1.5T or 3T.
Once the patient is positioned in the MRI scanner, the protons in the hydrogen atoms inside the body starts to align. Hydrogen protons – abundant in water and fat tissue in the body – naturally spin in random directions. But when the magnetic field is applied, these hydrogen atoms line up like tiny compass needles.
This alignment creates a baseline magnetization across the patient’s body, which sets the stage for generating the MRI signals used to make images.
Corsmed’s simulator replicates this initial alignment step by simulating a digital magnetic field. Users can choose field strengths of 0.25T, 0.31T, 0.4T, 0.55T, 1.5T or 3T.
Every voxel (3D pixel) in the simulator’s “digital patient” model represents a tiny region of tissue, with embedded hydrogen atoms.
These atoms are programmed to respond to the virtual magnetic field, aligning similarly to what would happen in a real MRI scanner.
Once the protons are aligned, the MRI scanner sends carefully timed radiofrequency (RF) pulses into the body. These RF pulses excite the protons so that they “tip” away from the magnetic field alignment.
When the RF pulse stops, the protons gradually relax back to their original position, releasing this energy as radio waves. These signals are captured by the MRI scanner and used as the raw data to create images.
RF pulses determine how much energy is released, but not where it’s released from. Without location data, the MRI would produce a general signal but no clear map of the body.
Gradient pulses solve this by adding controlled variations in the magnetic field across the body. These variations give each area’s protons a unique frequency or phase, allowing the MRI to identify the exact origin of each signal.
By combining RF pulses (for signal generation) with gradient pulses (for location tagging), MRI can create highly detailed, location-specific images, like T1- or T2-weighted scans, that highlight specific tissues and conditions.
Corsmed’s simulator recreates pulse sequences and proton responses digitally, using sophisticated math and compute processing that mimic real MRI physics.
This is done in two steps:
Bloch equations are a set of mathematical equations that describe how protons excite and relax in response to magnetic fields and RF energy.
Since each voxel (a 3D pixel of tissue) in the simulator is assigned magnetic properties like T1 and T2 relaxation times, these equations can model the physics of MRI in astonishing detail.
Bloch equations allow Corsmed to replicate all the key stages of the MRI process, including:
The animation below illustrates a Bloch equation graph that simulates how a proton – excited with RF energy – relaxes back to alignment. The proton precesses (wobbles) around the direction of the magnetic field as it realigns.
When protons instead are excited with RF pulses, their spin will travel in the opposite path as shown in this simulation.
Corsmed uses Bloch equations to manage an incredible volume of calculations. This includes simulating:
Solving Bloch equations is an extremely difficult task. Even simple ones require a massive amount of compute power.
To speed up the simulations, Corsmed uses massively parallel GPUs (Graphics Processing Units) to handle these calculations for millions of voxels at once. Normal CPUs (Central Processing Units), in contrast, can only work on a handful calculations at once.
Simulations that would have taken 30 minutes – or even several hours – can thus be done in a few seconds.
In fact, pulse sequences on Corsmed’s simulator return images up to 10X faster than real scanners. This near-instant feedback allows users to get far more practice than they ever could on a real MRI machine.
How much compute power does Corsmed use to simulate a realistic MRI scan?
To produce an image near instantly, even the simplest MRI sequences require many trillions (1012) of operations per second. (1 trillion calculations per second is called a “TeraFlop”).
The more advanced MRI sequences can consume many hundreds of TeraFlops of compute power.
So how much is 100 TeraFlops? It’s enough to:
Navigate 160 SpaceX Falcon 9 rocket launches at the same time. 1
Power 8 high-fidelity flight simulators simultaneously. 2
Generate virtual environments for 110 VR headsets. 3
Simulate the traffic of all the cars in Los Angeles in real time – 128 times over. 4
This is how much compute power Corsmed uses to ensure the highest degree of accuracy in its MRI simulations – while producing images in mere seconds.
After the RF signals are acquired, they’re mapped into a unique data format called k-space, which stores the raw data representing the scanned anatomy. In k-space, signals are arranged in a grid based on their frequency and phase, rather than in a recognizable image format.
To transform this frequency-based data into a visual DICOM image, MRI scanners use a mathematical process called the Fourier Transform.
Corsmed’s simulator performs the same data processing and image reconstruction. But it uses GPU processing in the cloud without the need for any scanner-specific hardware.
Using a cluster of GPUs, the simulator rapidly maps every voxel's frequency and phase data into k-space. It then performs the Fourier Transform just as a real scanner to reconstruct the signals into a final DICOM image.
This table summarizes the 5 core steps in MRI scanning, and compares how they’re performed on a real scanner vs in Corsmed’s virtual simulation:
The answer is "yes".
Corsmed can accurately reflect the output of a real scanner, because it follows the exact same MRI physics and steps as a real scanner.
Furthermore, it simulates every step with extreme realistic accuracy. Including:
This is how every user input can be captured in the resulting image, including details like anatomy, resolution, SNR, scan time, contrast, SAR, and artifacts.
Whatever you modify — parameters, slices, or settings — the simulator reflects in the image.
For example, if you change the slice angle from 0° to 20° in a brain protocol, the image will adjust as illustrated below:
You could also plan the slices at an extreme oblique angle — even an angle you would never do in real scan — and the simulator will accurately reflect that too.
This responsiveness to every input is possible because Corsmed’s simulator creates every image from scratch, following the exact same MRI physics and steps as a real scanner.
And since everything happens in the cloud, you can train on the simulator 24/7, just like using any ordinary web app. Corsmed becomes like a personal MRI scanner on your laptop.
Each pulse sequence also returns an image up to 10X faster than on a real scanner – allowing you to get far more effective practice in the same amount of time.
That is why leading MRI colleges and hospitals use Corsmed to educate their students and upskill their MRI technologists, including:
Want to learn more about how Corsmed can enhance your MRI education or training program?
Book a free consultation with one of our experts today, and discover how Corsmed can help you take your MRI skills to the next level.
Content
Content
Content
Content