Improving imaging of the heart and the cornea through modelling

Tissue palpation through imaging

While conventional imaging techniques (e.g. ultrasound, MRI, CT scan) “see” the inside of the body, shear wave elastography is a more recent breakthrough that can be used for “palpation”. What it actually does is measure local elasticity: diseased tissue, as occurs with fibrosis or a tumour, becomes more rigid than healthy tissue. 

“Elastography has been used routinely for the past decade or so, including for liver disease”, explains Sébastien Imperiale, a researcher with the project team Ananke at the Inria Saclay Centre and the École Polytechnique, a member of the Institut Polytechnique de Paris. “It is non-invasive and has a lot of advantages over biopsies, but is trickier to apply to organs such as the cornea or the heart, where it could be used to diagnose certain serious conditions affecting them.” Using it for the screening of keratoconus, for example, a neurodegenerative disease of the eye that severely affects vision, or necrosis of the heart muscle caused by myocardial infarction, is difficult.

A wave 150 times slower than ultrasound

Before we can understand these issues, let’s first take a look at the principle of elastography, which involves using ultrasonic waves to apply a tiny amount of mechanical pressure to the area under examination. In response, a complex mechanical phenomenon generates a second “shear” wave that is different in nature from the first. 

This wave passes through soft tissue at a much slower speed than ultrasonic waves (1 to 10 metres per second, compared to 1500 m/s). Most importantly, its speed increases with the rigidity of the tissue: a faster than normal shear wave is a likely indicator of disease. 

The method is well-suited to organs such as the liver, the prostate or the breasts, but things become more complicated with the cornea, subject as it is to intraocular pressure (hence its dome-like shape), and even more complicated with the heart, which is constantly in motion and whose internal pressure varies in order to generate blood flow. 

A long list of obstacles to overcome

“What this means is that the intrinsic rigidity of either of these organs cannot be used as a marker of potential disease. If elastography reveals an unexpected wave speed map, we don’t know whether it is physiological or related to the presence of disease.” 

But that’s not the only difficulty. Neither the heart nor the cornea are homogeneous environments: although less than a millimetre thick, the cornea has five layers of different cellular and biochemical compositions. Another obstacle is that the heart is protected by the ribcage, making imaging difficult. The heart also generates vibrations which can interfere with elastography measurements. As a result, examinations still fail to deliver a truly accurate diagnosis. 

Seeking to overcome these obstacles, researchers from Ananke (who have been collaborating with colleagues from the Inserm project team Physmed since 2022 as part of the French National Research Agency (ANR) project Elastoheart) set about developing complex mathematical and digital models. 

Using modelling to see and understand the invisible

These models describe the propagation of shear waves, including factors such as their trajectory through the heart or the cornea, and how the waves are altered by the properties of the tissue or the fact that the heart can be at different stages of its cycle. 

“We already have something new to offer clinicians: knowledge of the entire journey taken by the wave. With elastography you only get partial observations made by sensors on the device.” 

Improving the accuracy of models through data assimilation

Another challenge is to find a way of applying algorithms to measurements in order to determine the rigidity of tissue. This is further complicated by the fact that shear wave propagation also depends on the shape of the organ on the day and at the exact moment it is measured. Intraocular pressure also varies between individuals, in addition to changing over time. As for the heart, it never rests.

“The challenge is to make our models as faithful as possible to the physiology of each individual patient”, explains Sébastien Imperiale. “We can use images from ultrasounds to configure our models, and we also compare our simulated measurements with real-life elastography measurements. Finally, in order to further reduce the error rate, we enrich these models through data assimilation, which is commonly used to improve the accuracy of weather forecasts.” 

Data assimilation combines noisy observations from real-life (in this case, elastography measurements of the cornea or the eye) with imperfect data generated by the model, the goal being to arrive at as accurate an estimate as possible of the rigidity of the organ. “The model produced has the advantage of generating simulated data with limited noise that can be used as a reference for real-life measurements in the future.” 

Moving towards elastography devices that are much cheaper than MRI machines

This is an extremely complex undertaking: it took three years to model shear waves in the cornea, and work on modelling waves in the heart - which is currently in progress - will take a few more years. This means it will be a while before patients are able to take advantage of these new uses in elastography. But developments are promising: “we can expect devices in the future to be as compact and portable as ultrasound machines, much cheaper than MRI equipment and to be of high enough resolution to make an accurate diagnosis.” 

After a three-year collaboration with the Inserm, Ananke has made significant progress, acquiring in-depth knowledge of the elastography of the heart and the cornea from both a mathematical and a digital perspective. They also have their first real-life measurements and have improved reconstruction algorithms, which will make imaging more accurate in the future. “This is a high-risk, high-reward field of research”, says Sébastien Imperiale, “which is what makes this challenge so captivating.”