Employment of open integrated brain-vessel imaging database with standardization

● Primary Data: Raw Database

Multimodality raw data (ADC, DWI, TOF, GRE, FLAIR, CT angio, TFCA, TCD) of disease-free healthy subjects and cerebrovascular patients while observing the regulations related to personal information protection, the personally identifiable information of all cohorts is transmitted to a 3rd party. It is provided through a thorough anonymization process in a removal state by 3rd party.

● Secondary Data: Integrated brain-vessel imaging database

Research and diagnosis of various cerebrovascular diseases based on the provided secondary database of length, tortuosity, SOAM, change in curvature, thickness, cross-sectional area, volume, and characteristics of brain lesions by territory extracted through the developed software and applicable to treatment.

● Large-scale big data cohort

It is possible to build a large-scale data cohort by continuously reinforcing and expanding the database by collecting brain images and clinical data from multi-institutional and in-hospital subjects.

Using an artificial intelligence-based open brain-vessel fusion image data warehouse platform to improve the accuracy of quantitative evaluation of cerebral blood vessels and diagnosis of various pathological conditions

● Qualitative radiology evaluation is based on the diagnosis based on the medical doctors’ personal experience and intuition. Therefore, it is possible to quantitatively evaluate the morphology of blood vessels in three dimensions using an AI-based open brain-vessel fusion image data warehouse platform. Three-dimensional morphological characteristics of brain blood vessels and small changes in brain lesions are extractable, including the length, tortuosity, SOAM, curvature, thickness, cross-sectional area change, volume, and characteristics of brain lesions by territory through the developed software. It creates a cerebrovascular data analysis technology that can sensitively find and probabilistically express the features.

Research scalability: Utilization of a program that automatically extracts lesions from patients with acute cerebral infarction

● A solution to interrogate the linkage between the morphological abnormalities of the cerebrovascular diseases and patients with them is needed to confirm the exact location of the cerebral infarction where the cerebrovascular disease occurred, which is a manual method by existing experts. We developed a deep learning-based cerebral infarction lesion extraction program and a patent-pending to create an automatic and consistent manner. Thus, it is possible to expand the cerebrovascular data analysis platform by developing a model that can measure the location of lesions by the territory of the patient’s cerebral blood vessels.

● Through the construction of a database applied with an AI-based cerebrovascular analysis algorithm, it is possible to present a standard (draft) for establishing a clinical database to domestic and foreign researchers. In the long term, it is possible to unify the database into an integrated database for patient diagnosis by vitalizing the establishment of a database for each clinical field in Korea.

● Through the research of various domestic and foreign researchers using the developed database, it is possible to present an improved diagnostic biomarker for vascular brain disease. This can be expected to serve as a diagnostic aid guideline for applying the optimal treatment to the patient. Finally, after clinical verification as an auxiliary function of the patient diagnosis system, it is intended to be used in the medical field.

● Using the database developed in this project, we intend to expand the multidisciplinary research field with other disciplines by grafting genomic medicine technology. It is possible to establish a new concept of brain disease diagnosis and treatment platform through the convergence of this task and other disciplines.

Contents and plans of patent and technology transfer for cerebrovascular image convergence database construction

● We intend to transfer technology to venture companies in related fields by deriving patients based on the following contents.

- Automated preprocessing algorithm of various brain MR images for standardized database construction

- Software that uses deep learning technology to improve the existing brain blood vessel image segmentation and skeletal algorithms at the same time and to segment them

- Multi-scale quantitative feature extraction algorithm from patient image data

- Analysis algorithm for discovering high-dimensional feature biomarkers using the developed cerebrovascular fusion image database

- User-friendly GUI advancement and visualization technology

Disclosure of R&D result DB

● Web-based information sharing and research sharing for building a convergence database platform

- Opened data sharing site for related researchers

* This study is based on three fundamental principles (① utilization of health and big medical data for public purposes, ② establishment of a discussion structure based on citizen participation/professionalism, ③ thorough protection of the rights of data subjects based on current laws and regulations). To provide data to researchers who have been approved by the Institutional Bioethics Committee (IRB) through the

- Plan to build a user interface (U) to share and visualize the research results of researchers using the open database built in this project

Application of R&D results and expansion of convergence research with other disciplines