Skip to main content

Training the Next Generation of European Visual Neuroscientists for the benefit of innovation in health care and high-tech industry

Periodic Reporting for period 2 - NextGenVis (Training the Next Generation of European Visual Neuroscientistsfor the benefit of innovation in health care and high-tech industry)

Reporting period: 2017-03-01 to 2019-02-28

Making sense of an ever-changing environment requires that the human visual system balances its stability and its ability to change and adapt. On the one hand, perceiving the visual world in a coherent way requires stability of the visual system. On the other hand, the ability to reorganize its neural processing may also be required to respond to changes in either the environment or even the system itself.

Our primary objective, understanding cortical stability and plasticity in the human visual system, is of significant scientific and clinical relevance. The NextGenVis Research Training program funded 15 PhD students across Europe who investigated topics ranging from basic clinical questions about eye and brain disease through to computational models of vision that inform the rapidly expanding field of machine learning. Here, we provide a short summary of our results and their potential future relevance to healthcare and industry.
"Albinism is a congenital disease characterized by reduced pigmentation in the eyes and skin. Puzniak, Ahmadi and colleagues of Otto-von-Guericke University Magdeburg have shown that MRI measurements of the optic nerves can provide a simple index that correlates with other aspects of the disorder – particularly vision loss. This information may be used to help clinicians provide a more accurate and useful diagnosis and prognosis to patients.

Bhat and colleagues of Fondazione Stella Maris, Pisa have found remarkable preservation of function in children suffering profound congenital damage to the visual brain. It implies that if visual training aimed at the restoration of visual function is to have any effect, it must be started as early in life as possible.

Patients with progressive forms of vision loss are, paradoxically, often unaware of this because the brain effectively ‘fills in’ the missing information which may cause delays in the detection of eye diseases such as glaucoma and retinitis pigmentosa (RP). Carvalho and colleagues of the University Medical Centre Groningen have developed new brain imaging techniques to objectively measure vision – even when ‘filling-in’ is occurring – which may aid the early diagnosis of these eye diseases.

Grillini and colleagues of the University Medical Centre Groningen have developed a new rapid visual assessment procedure based on eye tracking that can aid in the diagnosis of glaucoma. It is non-invasive, provides diagnosis of the disease at a relatively early stage and does not require the patient to perform a complex task.

Posterior Cortical Atrophy (PCA) is a brain disease that causes degeneration in the visual and attentional regions of the brain. De Best and colleagues of Hadassah Medical Organization have shown that the visual brain of PCA patients represents the world differently giving a new understanding of how the brain reorganizes in the face of neurodegenerative diseases which may lead to improved diagnosis and treatment in the future.

The brain’s gray matter is divided into individual layers that are fractions of a millimetre deep. The way information moves between these layers is fundamental to its operation. Van Dijk and colleagues of Utrecht University have created a state-of-the-art brain imaging pipeline that allows for the accurate analysis of functional MRI data over these layers.

Understanding how the human brain represents the world is a fundamental part of neuroscience. Edadan and colleagues of Utrecht University have now developed a new type of measurement using a magnetoencephalography scanner that tells us about the electrical activity of neurons in the brain and therefore provides a direct read-out of brain activity. This enables new ways of quantifying the activity of the brain in diseases that can affect vision.

Faces are visual stimuli with immediate relevance for everyday social interactions, yet it is still not well understood how they are detected and recognised by the brain. Gnolo and his colleagues of Brain Innovation BV developed a mathematical description of how the visual properties of faces contribute to the perception of ""faceness"". Moreover, Gnolo contributed to the development of decoding methods of brain signal which may be applied to communication with patients with lock-in syndrome.

Neurodegenerative diseases such as Parkinson’s Disease (PD) can have effects on the human visual system. Himmelberg and colleagues of University of York have shown that changes in the visual system of animal models of PD are complex and reliable enough to allow us to diagnose the presence, progression and even the subtype of the disease. In parallel, Østergaard and colleagues of Lundbeck A/S, Copenhagen have examined the effect of genes relevant for PD pathology on deep brain structures in rats using fMRI.

Amblyopia or ‘lazy eye’ is a leading cause of vision loss. The conventional therapy for this brain disease in children is to occlude or ‘patch’ the unaffected eye. Kurzawski and colleagues of Fondazione Stella Maris, Pisa have shown that a radical new type of therapy shows promise for treating this disease now also in adults with this this visual disorder.

A remarkable new gene therapy may restore function to their retina, but to predict its effectiveness on restoring vision, we also need to understand how the brains of patients change. Molz and colleagues of the University of York have used MRI to map both structure and visual function in brains of patients with achromatopsia, a rare congenital eye disorder that leaves patients with reduced central vision and complete colour blindness. Their results will allow clinicians to predict more accurately how individual patients might benefit from gene therapy.

Semeniuta and colleagues of Pattern Recognition Company GmbH have developed novel machine vision algorithms for object recognition and scene segmentation. These new algorithms may benefit the manufacturers of ‘smart visual prosthetics’ – devices that interpret the visual world for patients with visual impairments.

Hernández-García and colleagues of Whitematter Labs GmbH have addressed the problem of training deep neural networks with little data by artificially creating new training examples out of available data (“data augmentation”). The high performance of their data augmentation techniques can also be adapted to other types of network and training data.
We have shown how observers can learn to adapt to changes in visual context and have found various evidence on how the architecture of the visual cortex adapts to major deviations in developmental as well as severe consequences of ophthalmologic and neurodegenerative diseases. NextGenVis has created impact in terms of the following:

• Gaining deeper understanding of stability and plasticity at the level of neuronal populations
• Designing innovative adaptive algorithms for use in automated recognition
• Designing new brain analysis tools
• Identification of possible new biomarkers for neurological disease
• Identification of new diagnostics
• Identification of directions for increasing the effectiveness of visual rehabilitation

We have disseminated the results of our research through communication and transfer of knowledge to other research settings, namely through presentations at various types of scientific meetings and conferences, as well as presentations and events targeted at the general public.