Projects / Programmes
QUANTITATIVE ANALYSIS OF BRAIN WHITE MATTER LESIONS
| Code |
Science |
Field |
Subfield |
| 7.00.00 |
Interdisciplinary research |
|
|
| Code |
Science |
Field |
| T111 |
Technological sciences |
Imaging, image processing |
| Code |
Science |
Field |
| 2.06 |
Engineering and Technology |
Medical engineering
|
medical image analysis, image registration, image segmentation, quantitative analysis, validation, brain, magnetic resonance imaging, multiple sclerosis, obstructive sleep apnea, white matter lesion
Researchers (17)
Organisations (2)
Abstract
Diseases and disorders affecting the human brain are the leading cause of physical and mental disability and have a huge and increasing socioeconomic impact. The neurological and mental disorders are among the largest contributors to the years lived with disability (30% of all diseases), which is linked to enormous costs of patient treatment and care (1). Long-term reduction of the number of patients, severity of disease expression, and the related enormous costs is only possible through advanced understanding of the associated risk factors, the disease pathogenesis and progression, and through objective, early and reliable assessment of novel drugs and other innovative ways of therapy. To achieve these goals, we aim to develop, test and validate methods, technologies and systems that enable objective and reliable longitudinal monitoring of disease progression and treatment effects.
The main challenge in the assessment of disease progression and treatment effects is to develop the ability to detect small changes by short-term screening, which would reliably and consistently reflect the long-term changes of sustained disease progression or treatment effects. Besides the assessment of clinical symptoms, which are often difficult to quantify, many of the neurodegenerative and mental disorders, and cerebrovascular diseases, can be assessed by quantifying the occurrence of brain white matter lesions (WMLs), i.e. areas of damaged white matter brain cells. WMLs are one of the few paraclinical symptoms that appear well before any clinical symptoms can even be recognized. Magnetic resonance (MR) imaging is the standard technique for the detection of WMLs (2), while the analysis of MR images allows the extraction of quantitative in vivo metrics of anatomically and physiologically relevant parameters of WMLs, which are sensitive biomarkers.
Quantitative measurements of WMLs, such as their number, size, shape and location and the corresponding measurements of changes of these parameters over time are extremely valuable but less explored biomarkers compared to qualitative assessments of WMLs. Quantification of WMLs requires their outlining in threedimensional (3D) MR images, traditionally performed manually, slicebyslice. Manual outlines serve for localization and morphological measurements of WMLs, both used mainly for diagnosis according to the evolving disease specific clinical guidelines (3). However, the nature of clinical guidelines is still mostly qualitative, rather than quantitative (4). Manual outlining of WMLs in MR images is subjective, tedious, and time consuming. Hence, the manual WML outlines are inaccurate and highly variable between different raters, and thus, manual quantitative measurements cannot be fully trusted and are rarely used as a clinical guideline. The accuracy and variations in WML outlines are even more critical when monitoring patients over time.
Objective and reliable quantitative measurements of WMLs in longitudinal MR images can be obtained by the use of advanced image analysis methods. A promising approach is to perform accurate spatial MR image registration between MR imaging sessions, followed by the analysis of MR intensity differences and resulting spatial deformations (5,6). The proposed project is a prompt response to the latest recommendations for improving longitudinal imaging studies (7) and will thus mainly concentrate on the problem of quantitatively measuring, in an objective, accurate and reliable way, the WML changes through automated registration of longitudinal MR images.
The primary objective of the proposed project is to design, develop, and validate a computeraided system based on quantitative measurements of WMLs for longitudinal studying, monitoring and treatment planning of diseases and health disorders associated to the progression of lesions in white matter brain tissue. The secondary objective is the translation of the automated WML analysis system into clinical pract
Significance for science
Within this research project we have developed and thoroughly validated novel sophisticated methods for automated analysis of magnetic resonance (MR) images. For the purpose of monitoring multiple sclerosis patients, we have implemented these methods into clinical practice (https://ms.quantim.eu). The results of the research project significantly contribute to the scientific fields of medical image analysis, neurology and neuroimaging, clinical radiology, and clinical trials of novel drugs. Relevant contributions to automated medical image analysis are: • increased understanding and contribution to the awareness of the potentials of the tested image analysis method, achieved through scientifically rigorous and objective validation of existing state of the art and novel image analysis methods • promotion of further research and development of image analysis methods, and facilitation of objective evaluation and comparison studies, by making standard validation image data sets with benchmark results publicly available to the research community. We have assembled and edited two publicly available MR image databases with highly accurate reference measurements of white matter lesions. Because the field of lesion quantification is very active, and because high quality reference data is not available as their extraction is difficult and time consuming, we expect that our contribution in the form of image databases will significantly contribute to the standardization of lesion quantification and, consequently, to the use of quantitative data as important imaging biomarkers in many neurological disorders. • formulation of guidelines for quantitative measurements in longitudinal studies Relevant contributions to neurology, neuroimaging, and clinical radiology: • advancement of understanding and monitoring of progression of several diseases associated to WMLs, achieved through objective and reproducible in vivo MR imaging-based measurements • further evolution and improvement of related clinical guidelines and treatment plans, by promoting objective and quantitative measurements in addition to the more subjective neurological status. The techniques for the extraction of imaging biomarkers for multiple sclerosis patients, that we have developed and implemented in practice, can be used with little modifications in the context of other neurological disorders. Through intensive interdisciplinary networking we have come into contact with several leading researchers in Slovenia and abroad. We are recognized as the provider of quantitative analysis of medical images. Relevant contributions to clinical trials of novel drugs: • faster development and assessment of novel drugs and other innovative ways of therapy, by enabling early decision-making based on objective measurements • reduction of the numbers of patients required to detect a given treatment effect in a trial by improved accuracy and sensitivity of quantified WMLs • more reliable characterization of individual patients for personalized treatment The primary goal of this project was the automated quantification of pathological and healthy structures but during the project we have identified a novel research area, which is the analysis of human connectome. Human connectome, among others, enables establishing the anatomical and functional connectivity within the healthy human brain, and produce a wealth of data that may facilitate research of brain disorders. In this novel area we collaborate with the UKC Ljubljana. The quality results of this project facilitate numerous future collaborations with academia and health institutions, as well as industrial partners at home and abroad. As this project resulted in numerous scientific publications in peer reviewed journals we expect that they will be highly cited.
Significance for the country
Neurological disorders like multiple sclerosis, Parkinson disease, Alzheimer's dementia, etc. are one of the main causes of heavy mental and physical disabilities, with a consequence of enormous costs of patient treatment and care. In 2010, the costs of treating these disorders in Slovenia were 2.4 billion Euros, which is 7% of Slovenian GDP. The immediate benefit of the results of this project for Slovenia is the objective automated quantification of over 900 MR images of patients with multiple sclerosis facilitated by the developed internet service. All slovenian hospitals dealing with MS patients have a connection to the server at the Faculty of Electrical Engineering and use our service. The automated image analysis system provides the neurologists and neuroradiologist in our hospitals fast, accurate and reliable data presented in a report issued max. 2 hour after the upload of MR images. Besides the conventional data on patients, the quantitative data provide important and objective information that may help the doctors to more deeply understand the status of the disease and to select the most appropriate treatment. Long term benefits are several, like lower cost of treatment of patients with neurological disorders, better outcome of treatment, faster recovery, new high-tech companies and jobs (as an example, we have already established the company INTELITEH), easy application of the technology to patients with other diseases, etc. In this research project we have developed technologies and systems that enable extraction of quantitative data from MR brain images. The results indicate that automated MR image analysis methods are capable of non-invasive extraction of much more information about a patient than a neuroradiologist can detect and characterize using conventional methods. Besides, automated methods have other advantages over classic radiological examination, like objectiveness, repeatability, higher traceability and standardization. Application of the system for quantitative analysis enables independant and proof-based creation of guidelines for the treatment of patients with neurological disorders. The quantitative information that our system provides, is establishing itself as one of the important arguments that neurologists use when deciding which therapy to use. In this way, they may provide the patients with personalized treatment. In long term a more optimal therapy leads to significantly lower expenses. The above mentioned benefits are the reason that in different areas of health care image-based diagnosis is more and more important, resulting in higher numbers and frequency of acquired images. This is connected with several challenges: 1) the need for more radiological experts, 2) more finance, 3) investment into modern medical equipment and infrastructure, ect. The technology of quantitative analysis of MR images that we have developed directly addresses several of the challenges above. For instance, extraction of numerous quantitative data from medical images requires less than 30 minutes, measurements resented in the form of a structured report provide the radiologist a tool for fast, reliable and efficient diagnosis and verification, and objective data supported reporting on the status of a patient. Besides a more complete treatment of a patient the application of this technology also shortens the radiological examination of the patient and consequently the costs. In longer term the results of the project may also create new jobs and professions in the field of biomedical engineering, more specifically in R&D, implementation and maintenance of the automated image analysis technology at clinics. The research group has been and is very active in education , which is of utmost importance for the efficient translation of knowledge from academia to the clinics and industry.
Most important scientific results
Annual report
2014,
2015,
final report
Most important socioeconomically and culturally relevant results
Annual report
2014,
2015,
final report
