=Paper=
{{Paper
|id=Vol-2743/paper-4
|storemode=property
|title=Development of a Service for Conducting Radiobiological Studies on the HybriLIT Platform
|pdfUrl=https://ceur-ws.org/Vol-2743/26-33-paper-4.pdf
|volume=Vol-2743
|authors=Yuriy Butenko,Dmitry Marov,Andrey Nechaevskiy,Dmitry Podgainy
}}
==Development of a Service for Conducting Radiobiological Studies on the HybriLIT Platform==
https://ceur-ws.org/Vol-2743/26-33-paper-4.pdf
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
DEVELOPMENT OF A SERVICE FOR CONDUCTING
RADIOBIOLOGICAL STUDIES ON THE HYBRILIT
PLATFORM
Yu.A. Butenko1, D.M. Marov1,2, A.V. Nechaevskiy1, D.V. Podgainy1
1
Laboratory of Information Technologies, Joint Institute for Nuclear Research, Dubna, Russia
2
Federal State-Funded Educational Institution of Higher Education of Moscow Region “Dubna
University”, Dubna, Russia
E-mail: ybutenko@jinr.ru
The article is devoted to the development of an information system for analyzing
experimental data on behavioral and pathomorphological changes in the central nervous
system in the study of the effects of ionizing radiation and other factors on laboratory
animals. Based on the analysis of similar existing systems, their major capacities and
disadvantages are defined, and a technology stack for developing a prototype is determined.
For data acquisition and storage, a client-server architecture of the system is elaborated, and a
database is designed. The prototype of the user web interface, the main components and tools
for analyzing experimental data are presented. The system is built on the basis of the
HybriLIT heterogeneous platform.
Keywords: web service, web interface, client-server application
Yuriy Butenko, Dmitriy Marov, Andrey Nechaevskiy, Dmitry Podgainy
Copyright © 2020 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
26
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
1. Introduction
The use of beams of protons and nuclei of other chemical elements presents a highly
promising direction for the tasks of space radiobiology (violation of the operator’s functions
of astronauts), as well as for the development and testing of drugs for prevention and relief of
post-radiation effects. At present, LRB JINR actively carries out experiments to study the
biological effects of ionizing radiation on the central nervous system of an experimental
animal at different periods after irradiation [1]. In the course of the studies, a large amount of
heterogeneous data (images, videos, text descriptions) from biomedical experiments appears.
The analysis of a large number of images obtained during biomedical studies is a
priority objective related to computer diagnostics. The following tasks are to be solved for
this:
- creation of a database for acquiring, storing and systemizing experimental data
corresponding to biological objects of the study;
- elaboration of algorithms for processing experimental data on the basis of machine
and deep learning methods for the tasks of pathomorphology and behavioral analysis in the
study of the effects of ionizing radiation and other factors;
- development of an information system based on modern IT solutions, including web
technologies, developed algorithms for statistical analysis and deep learning models, modern
solutions for inference and components for the visualization of data analysis.
The transition to advanced digital technologies, such as machine learning, neural
network approaches and high-performance computing technologies, for solving this class of
tasks will enhance and speed up the process of analyzing and obtaining results from the data.
To process and analyze data, it is planned to use the resources of the HybriLIT [2]
heterogeneous platform, which is a part of the Multifunctional Information and Computing
Complex (MICC) of the Laboratory of Information Technologies of JINR [3]. The
heterogeneous platform comprises the “Govorun” supercomputer [4] and the HybriLIT
education and testing polygon combined by the unified software and information
environment. The “Govorun” supercomputer includes two components, namely, the CPU
component based on the latest Intel architectures (Intel Xeon Phi and Intel Cascade Lake
processors) and the GPU component based on the NVIDIA DGX-1 Volta nodes. The
NVIDIA Volta architecture is the first in the world designed specifically for machine and
deep learning tasks.
The implementation of the given information system will result in the systematization
of the accumulated results, the identification of hidden patterns that appear in the response of
biological systems to the effects of damaging factors and, as a result, a significant
simplification and acceleration of diagnosis, particularly in the early stages of diseases, of the
elaboration of effective methods of prevention and counteraction to the negative effects of
ionizing radiation.
2. Analysis of existing services
Medicine and radiobiology are already considered as one of the strategic and
promising areas from the viewpoint of the effective implementation of machine learning,
neural network approaches and high-performance computing. The application of these
methods can massively increase the accuracy of diagnostics, make life easier for patients with
different diseases, accelerate the development and release of new drugs, etc.
Today, platforms and online services to support decision-making in the diagnosis and
analysis of different medical images exist and are developed. It is noteworthy that during the
27
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
COVID-19 pandemic, such online services have become particularly relevant and gained
great popularity.
For example, HistoScan is a cloud storage of a digital archive of histological
preparations, clinical information and lecture material with a high degree of data protection
[5]. The given service opens up wide opportunities for remote consultations and training of
specialists. HistoScan allows users to upload real-life cases from practice in their personal
account, while protecting personal data. Then one can get a consultation from doctors around
the world. The platform also enables remote training seminars and lectures.
The Botkin.AI online platform is a system based on artificial intelligence, it analyzes
images and helps to diagnose and assess the risks of the development of diseases, including
cancer [6]. For example, Botkin.AI can detect tumors on computer tomography, MRI, X-ray
or mammography of a patient. It is based on its own patented technology for constructing
mathematical models, with the help of which diagnostics is performed and disease risks are
assessed. The program can be easily integrated with the diagnostic equipment, therefore it
can be easily implemented in the work of a medical institution. Access to the system is
carried out after passing the authentication procedure, after which the user can upload the
necessary image in his personal account, analyze it and obtain the result based on neural
network approaches.
HistoScan and Botkin.AI operate as online tools and provide the user with a web
interface to interact with the software product; moreover, full access to the system is obtained
only after passing authentication.
Having studied the functions and capacities of the services described above, a service
architecture has been proposed, and a prototype of an online service to solve the assigned
tasks has been implemented.
3. Development of the service architecture
The service is implemented as a web application and has a client-server architecture.
The server is Node.js [7], and the client is written using the React.js library [8]. The service
will provide convenient and safe access to the HybriLIT platform and the ML/DL ecosystem,
which comprises a large number of frameworks and tools. This approach has the following
advantages:
● the service provides authorized access to the computing components of the HybriLIT
platform using the SLURM batch system;
● on the basis of web technologies, a common interface for researchers is implemented
for interacting with the MariaDB database [9] and file storages (HLIT-storage);
● the service provides a user interface for storing, modifying and adding data obtained
in the course of experiments.
In addition, all calculations will be performed on a remote dedicated server, which
does not require computing resources from the user’s device.
The general architecture of the service is illustrated below (Fig. 1).
The server-side (backend) is the software and hardware part of the service, which
implements application programming interfaces (APIs) for interaction of the client-side
(frontend) with the service components without directly accessing them from the frontend
using POST requests. It simplifies work with the platform for a group of researchers and
isolates different use cases depending on access rights. The core of the backend is the Node.js
technology. This software platform allows one to implement a fast, asynchronous, event-
driven API server. Node.js also adds the ability for JavaScript to interact with I/O devices via
its programming interface and to connect other external libraries, which are written in
different programming languages (C/C++, Python, etc.). The SLURM [10] scheduler
28
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
provides access to the computing power of the platform on a competitive basis. Due to these
features, the API server must track the status of running tasks and keep the current state for
each task in the database.
Fig. 1. General architecture of the service
To organize the storage of information about experiments, the API server under
development must interact with the metadata storage implemented on the basis of MariaDB
and file storages of the HybriLIT platform. Node.JS has a multitude of built-in and third-
party packages to interact with these components. There is a client with non-blocking I/O
MariaDB Node.js connector for interacting with the database, and there is a built-in module
for working with POSIX-compatible file systems to interact with mounted file storages. An
example of the simplified scheme of the API server operation is shown in Figure 2.
Fig. 2. Scheme of the simplified operation of the API server
The developed service presupposes a hierarchical data structure, namely, Experiment
– Group – “Object” of research, which corresponds to the logic of conducting an experiment
by researchers of LRB JINR.
29
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
Experiment is the major element inside of which all the contents of the experiment
will be stored. It has a standard set of attributes: name, creator’s name, creation date,
description; and a special attribute such as exposure (for example, gamma radiation).
Group is an auxiliary element related to a specific experiment, inside of which data on
“objects” of research (experimental animals) will be stored. In addition to the standard
attributes, it has special ones: organ under study, dye, microscope magnification, preparation.
“Object” of research (experimental animal) is an auxiliary element related to a
specific group; metadata of the “objects” of the experiment (photos and video files) will be
stored inside.
The database scheme for the service under development, which is implemented in
accordance with the presented hierarchical data structure, is illustrated in Figure 3. The
database contains information about experiments, groups, objects, general data (creation date,
creator, name, description), special attributes (organ under study, dye, exposure, microscope
magnification, preparation).
Fig. 3. Database scheme
4. Frontend and development of the service prototype
The frontend is the graphical interface that the user sees on the page. It is responsible
for the appearance of the components (stylization) and their position on the page (layout).
With the help of the frontend, the user can interact with the backend and the database using
HTTP requests to the API. From the user’s side it happens unnoticed and feels like, for
example, pressing a button, entering data into a form, clicking on a page element. The
following technologies have been used as means for implementing the frontend of the
service:
● HTML5/SCSS;
● JavaScript, WebPack [11];
● React.js, Redux [12].
The service prototype has been implemented using these technologies. To start
interacting with the service, the user needs to go through the authentication procedure by
entering the login and password in the form shown in Figure 5 (1. Authorization).
30
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
If the login and password are correct, a valid token is issued, and the user goes to the
main page of the service. The given page has a table that contains information about existing
experiments in the database. Data from the database is transferred to the client as a result of
an API request. In the request body, the required parameters and the token are passed to
check that the user is authorized in the system, and a sample from the database in the JSON
format appears as a response. The sample is further processed by React.js and displayed on
the page as a table, as shown in Figure 4 (2. Page with experiments).
Fig. 4. Table with experimental data
The table contains basic information about experiments: name, creation date and
author of the experiment. In addition, on this page, the user can manage experiments and
make changes in the database, for example, add a new one or delete an existing one. To go to
groups within the experiment, one should click on the “Details” button. The page with groups
also provides information obtained from the database as a result of an API request. On the
page with groups of the experiment (3. Page with groups in the experiment), there is a card
that contains full information about the selected experiment. By clicking on the “Edit” button,
the fields become active, and one can make changes, which will also be reflected in the
database.
The user can create a new group and delete an existing one. To go to the page for
filling in information about laboratory animals in a group, one should click on the “Details”
button.
During experiments, video materials and images are accumulated in a group of
pathomorphologists. To store data of a specific animal, the user must create a folder, for
example, mouse1. Downloading and storing digital materials (videos and images) is also
distributed on the server in the corresponding directories. For example, in the directory
mouse1, two folders, namely, morphology (for images) and behavior (for videos), will be
created. To download, for example, images, the user needs to click on the “Morphology”
button, after that he will go to the page with a form for downloading files.
To view the image, the user needs to click on the “Preview” button in the line with the
selected image. After that a window, in which the selected image is loaded from the server,
opens (Fig. 5).
31
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
Fig. 5. Window for previewing the file
The functions of analyzing images and video materials are available to the service
user. Results of image analysis can be obtained and presented in different forms. For
example, one can obtain a color mask of the image, on which normochromic, lightly
modified and degenerative neutrons will be marked with colors. Since the service under
development is aimed not only at pathomorphologists, but also at developers, the user
interface enables the view of intermediate results of algorithms, for example, primary
segmentation (Fig. 6). This is necessary in order to understand how the algorithm works at
each stage and to correct its work. The result of the work of the image analysis algorithm will
be the statistics on objects of interest to the scientist.
Fig. 6. Example of image segmentation
32
Proceedings of the Information System for the Tasks of Radiation Biology Workshop (ISRB2020)
Dubna, Russia, June 18, 2020
Conclusion
Thus, the article presents a prototype of a web service for conducting radiobiological
studies on the HybriLIT platform. A client-server architecture has been implemented, and a
database has been designed. The interaction with the service is performed using the
developed user interface. Currently, the prototype provides the ability to store, create and
modify data, as well as to store videos and images obtained as a result of experiments in a
structured form. The present results confirm the relevance of the current project, which can
be considered as a new promising solution for radiobiological studies underway at LRB
JINR.
In the near future, it is planned to develop a module for presenting results of the
analysis of images and video files after applying neural network algorithms, as well as to
refine and enhance the user interface.
References
[1] Kolesnikova I.A., Budennaya N.N., Severiukhin Yu.S. et al. Effect of proton irradiation
on the morphofunctional state of the rat brain [Electronic resource] //
http://brainres.ru/work/Sudak%202019_15%20Congress%20Proceeding_electron%20version
.pdf#page=227
[2] Adam Gh., et al. IT-ecosystem of the HybriLIT heterogeneous platform for high
performance computing and training of IT-specialists // Selected Papers of the 8th
International Conference «Distributed Computing and Grid-technologies in Science and
Education» (GRID 2018), Dubna, Russia, September 10-14, 2018, CEUR-WS.org/Vol. 2267
[3] MICC - INFRASTRUCTURE, n.d. Available at: https://micc.jinr.ru/?id=29 (accessed
15.08.2020)
[4] Supercomputer “Govorun” – Heterogeneous cluster | LIT/JINR, n.d. Available at:
http://hlit.jinr.ru/en/ (accessed 15.07.2020)
[5] Histoscan.com is a fast and comfortable way to receive and make a second opinion for
scanned digital slides. Electronic resource: http://www.biovitrum.ru/glavnaya/press-
center/all_news/histoscancom/
[6] Platform for the analysis and processing of medical images using AI technologies.
Electronic resource: https://botkin.ai/
[7] Node.js is a JavaScript runtime built on Chrome's V8 JavaScript engine. Electronic
resource: https://nodejs.org/
[8] React is a JavaScript library for building user interfaces. Electronic resource:
https://reactjs.org/
[8] MariaDB Server: The open source relational database. Electronic resource:
https://mariadb.org/
[9] Slurm Workload Manager - Overview, n.d. Available at:
https://slurm.schedmd.com/overview.html (accessed 18.09.2020)
[10] Webpack is a static module bundler for modern JavaScript applications. Electronic
resource: https://webpack.js.org/
[11] Redux is a predictable state container for JavaScript applications. Electronic resource:
https://redux.js.org/
33