Researchers from the joint Biomedically-Informed AI
Laboratory of Skoltech (a VEB.RF Group institution) and
the University of Sharjah and their colleagues have improved
the performance of computer vision algorithms that look for
signs of epilepsy and multiple sclerosis on brain MRI scans.
Rather than merely analyzing an image as a set of pixels,
the neural network now also attends to medical features
known to physicians. This is complemented by the
machine’s capability to spot subtle details that elude the
human eye. In the future, such solutions will help doctors
diagnose diseases and select treatment strategies.
The Russian Science Foundation-backed study
was published
in the proceedings of Medical Image Computing and Computer
Assisted Intervention — an A-rated international
conference on medical imaging and AI.
Computer
vision algorithms are already fairly good at detecting signs
of drug-resistant epilepsy and multiple sclerosis in medical
images. This gives hope that neural networks will soon be
able to assist radiologists in this work, thereby
speeding up the analysis and relieving the doctors of some
of their workload, while also improving detection rate.
About 30% of cases of epilepsy alone are currently deemed
“MRI-negative”: The doctor does not see signs of
the disease on the scan, but the machine can.
Advertisement – scroll to continue reading
One of
the main difficulties, though, is that existing models are
geared toward detecting the typical presentation of the
disease. For instance, in drug-resistant epilepsy, a
blurring of the boundary between gray and white brain matter
is frequently observed, and this feature is easily detected
by neural networks. However, there is a rarer scenario where
compact areas with increased signal intensity serve as the
key sign of the disease on the scan. Today’s models
largely ignore these.
Image.
The highlighted areas on MRI scans of patients with multiple
sclerosis correspond to characteristic affected regions
of various types. Modern computer vision models
effectively detect the areas marked in red but have a hard
time spotting the green ones. Credit: Nadezhda
Alsahanova et al./Medical
Image Computing and Computer Assisted Intervention —
MICCAI 2026
While in epilepsy pinpointing
the location of a lesion is necessary to
perform surgical intervention, in multiple sclerosis
monitoring the appearance of new disease foci of various
sizes allows doctors to lay out or adapt
a treatment plan. The problem here is that models are
good at finding large lesions on scans but often miss small
ones. Thus, in both epilepsy and multiple sclerosis, the
model has to be taught that sometimes pathologies manifest
atypically, and the features observed in those cases
are significant — not merely random noise. This is
precisely what the authors of the study managed to
achieve.
“By integrating biological features into
the neural network, we make it work more in line with
the actual reasoning of physicians,” said the
study’s lead author, Nadezhda Alsahanova,
a research engineer at Skoltech and a PhD student
in the Institute’s Computational
and Data Science and Engineering program. “Further
development of this approach will make the results of
computer vision analysis more interpretable. The model
will justify its conclusions in a comment, making the
findings more trustworthy in the eyes of a
physician.”
The researchers tested their approach on
clinical data. For both diseases, the efficiency of
pathological area detection increased: by 2% for scans
showing signs of multiple sclerosis, and by 8% — from
72% to 80% — on scans showing signs of epilepsy.
Training also became more robust: The new model could
match the performance of its predecessors with 30% less
training data.
Technically, the innovation behind
the improvement involved supplementing segmentation, which
is a standard task in automated medical image analysis,
with an auxiliary procedure, classification. Now,
the model does not simply segment the image
in the manner of “these areas are healthy
tissue, and these other ones are pathological,” but also
categorizes the latter regions based on which part of the
brain they appear in, whether the suspected lesion has
blurred boundaries, and so on. Thanks to this, the model
begins to “pay attention” to biomedical features, which
improves segmentation quality and — in the long run —
allows the result to be accompanied by a comment for the
doctor’s benefit.
Assistant Professor Maxim
Sharaev, who co-directs the joint
Skoltech-University of Sharjah BIMAI-Lab and was a principal
investigator in the study, said: “Searching for
small and atypical pathologies on MRIs is a critically
important task, since missing them can cost patients the
right treatment. What makes this study’s approach
original is that when detecting a pathology, the model
begins to rely on biomedical features, much like
a human physician. Consequently, the output becomes
interpretable and more suitable for real-world diagnostics.
Importantly, Nadezhda, who came up with this idea and
brought it all the way to MICCAI level, managed
to confirm the findings on clinical
data.”
About Skoltech:
Skoltech is a
private international university (part of the VEB.RF group)
in Russia, cultivating a new generation of leaders in
technology, science, and business. As a factory of
technologies, it conducts research in breakthrough fields
and promotes technological innovation to solve critical
problems that face Russia and the world. Skoltech focuses on
six priority areas: life sciences, health, and agro;
telecommunications, photonics, and quantum technologies;
artificial intelligence; advanced materials
and engineering; energy efficiency and the energy
transition; and advanced studies. Established
in 2011 in collaboration with
the Massachusetts Institute
of Technology (MIT), Skoltech was listed among
the world’s top 100 young universities by the Nature
Index in its both editions (2019, 2021).
On Research.com, the Institute ranks as Russian
university No. 2 overall and No. 1 for genetics.
In the recent SCImago Institutions Rankings,
Skoltech placed first nationwide for computer science.
Website: https://www.skoltech.ru/.