Periodic Reporting for period 1 - MATEDD (MRI analysis tool for early detection of dementia)
Reporting period: 2023-06-01 to 2024-04-30
More than 55 million people around the world are currently living with dementia. Due to the progressive aging of the population, it is estimated that the number will reach 75 million by 2030, with about 10 million new cases every year. Dementia is a disease described by accelerated brain ageing which leads to decline in both memory and cognitive skills, thus affecting everyday activities of those suffering. On average over 60% of patients do not receive a diagnosis and average cost of misdiagnosis is estimated to be over 2K € leading to annual care costs of 32K € per patient.
Current diagnostic pipeline allows only to start treatment and lifestyle changes once the symptoms have occurred and progressed. Early detection of dementia allows to reduce long-term patient-related care costs and improve the quality of life of the patients as there are scientifically proven lifestyle changes enabling to decelerate the course of disease.
Existing early diagnostics methods lack speed and accuracy, are not automated enough, and detect only already existing symptoms when it is late for prevention. In clinical practice, cognitive test are used for early diagnostics which lack accuracy and only detect changes once the symptoms have already progressed.
Neurosalience aims at tackling the challenges posed by the late detection of dementia due to the inadequacy of existing methods. It was previously shown that accelerated brain ageing is associated with the onset of dementia. At Neurosalience we develop a software for assessing brain ageing for early detection of dementia from low-resolution MRI data. Our current prototype detects the early signs of dementia starting from structural MRI scans and is a deep learning-based classifier to predict brain age from an MRI scan and determine whether a patient has dementia, its type and stage.
Difference between predicted and chronological subject’s age and comparison of important features for making a prediction are used to diagnose dementia.
The tool allows to decrease patient-related costs and improve the quality of life of patients. The tool is first in the world to be capable of processing even low-resolution MRI data from older scanners, thus allowing analysis of the images regardless of the scanner used, resulting resolution and parameters of scanning. This is of particular importance in launching the product to be used in governmental healthcare system as a diagnostic tool. Furthermore, our method is fully automated compared to diagnostic methods based on cognitive tests, resulting in accurate diagnoses obtained in a short time.
The project objective is to boost the market readiness of the Neurosalience tool by developing the project from business, technical and critical perspectives. For this purpose, the project involves improving product models and preparing documentation for intellectual property rights application.
Current diagnostic pipeline allows only to start treatment and lifestyle changes once the symptoms have occurred and progressed. Early detection of dementia allows to reduce long-term patient-related care costs and improve the quality of life of the patients as there are scientifically proven lifestyle changes enabling to decelerate the course of disease.
Existing early diagnostics methods lack speed and accuracy, are not automated enough, and detect only already existing symptoms when it is late for prevention. In clinical practice, cognitive test are used for early diagnostics which lack accuracy and only detect changes once the symptoms have already progressed.
Neurosalience aims at tackling the challenges posed by the late detection of dementia due to the inadequacy of existing methods. It was previously shown that accelerated brain ageing is associated with the onset of dementia. At Neurosalience we develop a software for assessing brain ageing for early detection of dementia from low-resolution MRI data. Our current prototype detects the early signs of dementia starting from structural MRI scans and is a deep learning-based classifier to predict brain age from an MRI scan and determine whether a patient has dementia, its type and stage.
Difference between predicted and chronological subject’s age and comparison of important features for making a prediction are used to diagnose dementia.
The tool allows to decrease patient-related costs and improve the quality of life of patients. The tool is first in the world to be capable of processing even low-resolution MRI data from older scanners, thus allowing analysis of the images regardless of the scanner used, resulting resolution and parameters of scanning. This is of particular importance in launching the product to be used in governmental healthcare system as a diagnostic tool. Furthermore, our method is fully automated compared to diagnostic methods based on cognitive tests, resulting in accurate diagnoses obtained in a short time.
The project objective is to boost the market readiness of the Neurosalience tool by developing the project from business, technical and critical perspectives. For this purpose, the project involves improving product models and preparing documentation for intellectual property rights application.
Technical work was performed on model improvements and tool’s backend development. In order to improve the model new data sources has been achieved using European Platform for Neurodegenerative Diseases. The datasets included MRI data for both healthy and diseased subjects, including all clinically defined stages of Alzheimer’s disease.
Further, the model was improved by training. Model training resulted in improved model performance in for all product’s models, including determining patient’s brain age, understanding whether accelerated brain ageing has occurred due to dementia, identifying type and stage of dementia.
As part of achieving regulatory approvals and CE mark, we have completed the first technical pilot project using clinical data in 2023 in collaboration with Warwick Univeristy. In this step we have confirmed tools usability for clinical data. The pilot study allowed us to reach TRL 5 since previously the prototype was developed using data deriving from 34 publicly available datasets collected in Europe, North America and Asia and used samples collected from healthy subjects. Research data is collected using imaging protocols leading to high quality scans which are not usually used in hospitals to save time and money. Also, older subjects in the research datasets are usually represented by subjects with exceptionally good health and this is not representative of overall patient population. Therefore, pilot project is an important step before starting full scale clinical trials. Along with completing the pilot project we have also improved clinical protocol used in the Neurosalience product trials. This was done with help of our partner clinical research organization – Adhoc Clinical.
Another aspect of the project was developing IP strategy and patent application for the product. For this purpose Neurosalience partners with Moosedog IP attorneys. In this project the results achieved in technical pilot contribute towards applying for regulatory approvals, hence the activities planned to achieve regulatory approval for the product are considered to be fulfilled.
The main outcome from project activities – is improved market-readiness for the Neurosalience product. IP strategy and improved models for the product along running a technical pilot allow us to move significantly further towards market-readiness. Hence, this project has accelerate the development of Neurosalience’s tool, impacting on Neurosalience as company and elevating its scalability potential.
Further, the model was improved by training. Model training resulted in improved model performance in for all product’s models, including determining patient’s brain age, understanding whether accelerated brain ageing has occurred due to dementia, identifying type and stage of dementia.
As part of achieving regulatory approvals and CE mark, we have completed the first technical pilot project using clinical data in 2023 in collaboration with Warwick Univeristy. In this step we have confirmed tools usability for clinical data. The pilot study allowed us to reach TRL 5 since previously the prototype was developed using data deriving from 34 publicly available datasets collected in Europe, North America and Asia and used samples collected from healthy subjects. Research data is collected using imaging protocols leading to high quality scans which are not usually used in hospitals to save time and money. Also, older subjects in the research datasets are usually represented by subjects with exceptionally good health and this is not representative of overall patient population. Therefore, pilot project is an important step before starting full scale clinical trials. Along with completing the pilot project we have also improved clinical protocol used in the Neurosalience product trials. This was done with help of our partner clinical research organization – Adhoc Clinical.
Another aspect of the project was developing IP strategy and patent application for the product. For this purpose Neurosalience partners with Moosedog IP attorneys. In this project the results achieved in technical pilot contribute towards applying for regulatory approvals, hence the activities planned to achieve regulatory approval for the product are considered to be fulfilled.
The main outcome from project activities – is improved market-readiness for the Neurosalience product. IP strategy and improved models for the product along running a technical pilot allow us to move significantly further towards market-readiness. Hence, this project has accelerate the development of Neurosalience’s tool, impacting on Neurosalience as company and elevating its scalability potential.
Main result of this project is improved market-readiness for the Neurosalience product. Updated models for the product along with the pilot study allowed us to move significantly further towards market-readiness.
Project results have covered three main categories – technical, business and clinical. At this stage of our company development, improving technical capabilities of the product and building clinical evidence are crucial steps as part of the go-to-market activities. Also in 2023 the main focus of Neurosalience has been on building first partnerships. During this project the first research collaboration was formed with Warwick University. Warwick University develops a screening tool for Alzheimer's disease using a flavour test. The test can help diagnose the beginnings of Alzheimer's years before symptoms of memory loss through a loss of taste or smell. Neurosalience works with Warwick University in order to establish unified pipeline for Alzheimer's disease screening and diagnostics in collaboration with UK National Health Service (NHS). In 2024 we plan to extend the pilot performed into large scale study. For this purpose and in order to build our first partnerships with hospitals for both validating and developing the tool through having access to new datasets, Neurosalience in collaboration with Warwick University have increased existing traction by organising the event for introducing our work to UK parliament.
In this project the IP strategy for the tool was developed. Our IP strategy is based on the combination of patenting the technology and trade secrets.
In terms of long-term wider impact, we expect this project to elevate the scalability potential of dementia early detection solutions from both technological and market points of view. Road-to-market for Neurosalience further involves performing large scale clinical trial for applying for CE mark for the dementia detection tool. Further company funding will be a combination of private investors and seeking for EIC funding.
Project results have covered three main categories – technical, business and clinical. At this stage of our company development, improving technical capabilities of the product and building clinical evidence are crucial steps as part of the go-to-market activities. Also in 2023 the main focus of Neurosalience has been on building first partnerships. During this project the first research collaboration was formed with Warwick University. Warwick University develops a screening tool for Alzheimer's disease using a flavour test. The test can help diagnose the beginnings of Alzheimer's years before symptoms of memory loss through a loss of taste or smell. Neurosalience works with Warwick University in order to establish unified pipeline for Alzheimer's disease screening and diagnostics in collaboration with UK National Health Service (NHS). In 2024 we plan to extend the pilot performed into large scale study. For this purpose and in order to build our first partnerships with hospitals for both validating and developing the tool through having access to new datasets, Neurosalience in collaboration with Warwick University have increased existing traction by organising the event for introducing our work to UK parliament.
In this project the IP strategy for the tool was developed. Our IP strategy is based on the combination of patenting the technology and trade secrets.
In terms of long-term wider impact, we expect this project to elevate the scalability potential of dementia early detection solutions from both technological and market points of view. Road-to-market for Neurosalience further involves performing large scale clinical trial for applying for CE mark for the dementia detection tool. Further company funding will be a combination of private investors and seeking for EIC funding.