ETRI WEBZINE

News 1

ETRI Develops Two-Way Medical Sign Language Translation Technology for Hearing Impaired

- Paving the way for health checkup preliminary questionnaire in sign language
- Will upgrade and transfer technology to non-contact reservation system following verification at hospitals

A group of South Korean researchers has set out to close the digital divide for the information disadvantaged. They are taking the lead in bringing about a “compassionate digital world” through R&D to tackle social issues in a bid to lower the barriers to medical service for people with hearing impairments.

Sign Language Translation Kiosk for Preliminary Health Checkup Questionnaires

Electronics and Telecommunications Research Institute (ETRI) announced that it has developed a two-way medical sign language service kiosk for those hard of hearing in collaboration with Chungnam National University Hospital, the Korea Association of the Deaf and EQ4ALL Co., Ltd. Those hard of hearing can communicate with an avatar appearing on the screen of the kiosk in sign language.

Developed by ETRI, the medical sign language service kiosk is a device that helps these users understand and answer health checkup questionnaires in sign language at hospitals. This technology will be further enhanced to incorporate additional essential features for more effective medical services.

The medical sign language service kiosk is expected to mitigate the difficulties experienced by hearing-impaired individuals in filling out these questionnaires, and help them access medical services offered in hospitals with ease.

Post COVID-19, important medical information, such as on infectious diseases and vaccination, has been provided through various channels. However, those hard of hearing have difficulty accessing it.

For such people, there is a truly urgent need for sign language translation service in the medical field. That being said, there are limitations for every healthcare organization to offer sign language translators. In addition, the masks worn by healthcare workers hinders communication by making it more difficult for those hard of hearing to read their lips and facial expressions.

Medical sign language translation kiosk: Health checkup questionnaire screen

Developed to tackle this issue, the kiosk translates the content of health checkup questionnaires in sign language to assist these users. With help from the avatar appearing on the screen, these users can easily understand the questionnaires in sign language. This technology is expected to considerably improve accessibility to medical services for them.

The kiosk also has an interactive function to translate the sign language used by them into Korean. When someone hard of hearing answers a question in sign language, the kiosk automatically analyzes it, converts it into Korean text and enters the answer on the questionnaire. This will be an important tool to help them more accurately explain their health conditions to the healthcare workers.

On November 4, with the Korea Association of the Deaf, the team of ETRI researchers jointly conducted a satisfaction survey on the kiosk service with hearing impaired respondents at the Eroom Center in Yeouido, Seoul. Based on the survey results, the researchers aim to develop an interactive sign language - Korean language translator to facilitate effective communication between those hard of hearing and healthcare workers.

ETRI will keep striving to deliver a better quality of life to the information disadvantaged by expanding application of this technology to a wide range of public services.

The kiosk machine must be able to recognize hand gestures and movements when a user engages in sign language in front of the camera and automatically identify which words are being expressed. To ensure such functions, the kiosk is equipped with image recognition technology, human joint signal analysis technology, Korean - sign language bidirectional conversion technology, etc.

The researchers are making a difference in the lives of people with disabilities by developing technologies such as sign language broadcasting and sign language translation avatar service based on original proprietary broadcast media technologies of ETRI. The researchers explained that the development of the sign language translation service kiosk is attributed to applying this new technology as an original proprietary one.

During the COVID-19 pandemic in 2021, the researchers contributed to greater hospital accessibility by developing an avatar sign language system for those hard of hearing, which was installed at Chungnam National University Hospital, improving hospital accessibility for people with disabilities. In 2020, they also developed an avatar sign language system providing the COVID-19 guidelines to people with disabilities, and distributed it to hospitals. Posted in the websites of the Ministry of Science and ICT and the Korea Association of the Deaf during the pandemic, these guidelines in sign language provided effective assistance to people with disabilities.

The researchers explained that the technology needs to be further upgraded and refined as a prototype so as to help those hard of hearing communicate naturally in sign language without any inconvenience. After further identifying any issues through a pilot program at hospitals, the technology will be transferred to the barrier-free kiosk distributors or non-contact reservation service providers, and subsequently proliferate nationwide.

ETRI assistant vice president Tae Jin Lee of the Media Research Division said, “The artificial intelligence era must usher in equal accessibility to essential social information related to medical services, routine life, safety, etc. for all. We need continued effort for and interest in compassionate ICT development for the vulnerable segment.”

This technology was developed as part of the “Two-Way Medical Sign Language Translation Service Technology Development for the Hearing Impaired,” a project implemented with support from the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation. Chungnam National University Hospital, the Korea Association of the Deaf, EQ4ALL Co., Ltd, Korea Electronics Technology Institute and Korea Advanced Institute of Science & Technology have participated in the project as collaborative research institutes.

Han Kyu Lee, Principal Researcher
Media Intellectualization Research Section
(+82-42-860-5178, hkl@etri.re.kr)

News 2

ETRI won first place in Global Medical Image Challenge

- Achieved world-best result with AI-powered diabetic food image analysis
- Will expand AI medical technology application to healthcare for geriatric skin diseases, such as pressure ulcer

First Place Award in the DFUC 2024 Challenge of MICCAI (From left: ETRI Senior Engineer Ji Won Park, Principal Researchers Yun Ji Ban and Dr. Seihyoung Lee, Prof. Moi Hoon Yap, the host of the challenge, and Prof. Raphael)

A group of South Korean researchers won the top award at a conference hosted by the Global Medical Image Association. Thanks to this, the global competitiveness and reputation of the research team’s AI-based medical technology were demonstrated again.

Electronics and Telecommunications Research Institute (ETRI) announced that its global research competitiveness was demonstrated by being recognized as the world’s best¹⁾ at an AI performance competition using diabetic foot²⁾ medical image data (Diabetic Foot Ulcer Challenge, DFUC³⁾) held in Marrakech, Morocco for five days from October 6 under the International Conference on Medical Image Computing and Computer-Assisted Intervention(MICCAI)⁴⁾.

The key agenda of this year’s conference was “Self-supervised instance segmentation of diabetic foot ulcers⁵⁾.” In particular, segmentation of lesions using unlabeled medical data⁶⁾ was presented as a key task. In other words, this conference aims to find the exact areas of lesions that occur on the foot due to diabetes.

So far, this task has been implemented primarily through supervised learning. However, a downside of supervised learning is that it requires a massive amount data, including tens of thousands of relevant photos. In addition, according to the characteristics of medical data, data labeling has been difficult and cost prohibitive.

Self-supervised learning has emerged as a key technology to resolve these shortcomings. Through self-supervised learning, the area of a lesion can be accurately identified even with a small volume of data.

Moreover, self-supervised instance segmentation takes on a crucial role in identifying separate lesions, even of the same type, and accurately distinguishing their boundaries. This technology can make a key contribution to automatically segmenting lesions from unlabeled data.

By pinpointing the exact areas of major lesions in diabetes and other skin conditions, it serves as a helpful tool to support medical doctors in treating their patients.
1) Top-ranked worldwide: https://dfu-challenge.github.io/
2) Diabetic foot: Diabetic foot is the most common complication of diabetes that affects one in four diabetic patients. This is a condition in which poor blood circulation causes the skin on the tip of the toe or heel, which is the farthest from the heart, to become blackened or, in severe cases, necrotic.
3) DFUC (Diabetic Foot Ulcer Challenge): An AI performance competition using diabetic foot medical image data
4) MICCAI (International Conference on Medical Image Computing and Computer-Assisted Intervention): The world’s most prestigious conference in the field of medical image processing and computer-assisted therapy
5) Self-supervised instance segmentation of diabetic foot ulcers: A technology enabling computers to identify unlabeled diabetic foot ulcers through self-supervised learning
6) Labeling: An operation of identifying objects or parts in a photo to help a computer understand the image

ETRI Research Team applied its modified self-supervised learning model¹⁾ to over 2,000 test data provided by the host of the challenge, and achieved the best results. This method is praised as a key technological advancement that can be utilized for both diabetic foot image analysis and diagnosis of various chronic diseases.

The researchers attested that during their preparation for the competition, they added an algorithm through strategic scheduling over a short time span by extracting data values based on the performance of ETRI’s analysis hardware, thus achieving these remarkable results.

Their research focuses on the treatment for pressure ulcers and diabetic foot, two chronic conditions of geriatric patients. In addition to the medical image analysis technologies using artificial intelligence, the researchers are developing fabric pressure mats for pressure ulcer prevention, optical coherence tomography (OCT) for diabetic foot diagnosis, and diabetic foot moisture level measuring equipment using terahertz technology. With this, they are striving to provide advanced medical solutions for early diagnosis and treatment of chronic diseases among geriatric patients.

Enhancing its technological prowess, ETRI plans to apply AI to a wider range of medical fields and also further strengthen its global competitiveness.

This award is expected to make a major contribution to the development of their research, and is recognized as a feat that validates Korea’s AI technological competitiveness in the global arena.

The principal researcher, Dr. Seihyoung Lee from the AI Convergence Research Section of ETRI said, “Based on the results of this research, we will not only further upgrade the technology to analyze diabetic foot images, but also apply the technology to the management of geriatric skin diseases, especially the development of diagnostic and prognosticating technologies for chronic skin diseases including pressure ulcers.”

The winning paper will be published in the Medical Image Analysis, an international journal in the field of medical image analysis.

Twelve teams from seven countries, including Korea, the U.S., the U.K. and Poland, competed in the challenge, and ETRI researchers produced the best results based on their unrivaled technological prowess and AI model.

The research was conducted as part of the “XR Twin Technology Development for Chronic Skin Disease Management for the Elderly,” a project funded by the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation, along with support from a local subsidy project of the Gwangju Metropolitan City.
1) Modified self-supervised learning model: A model developed to apply Mask R-CNN which identifies the characteristics of an image using DINOv2, a self-supervised learning-based model, and classify the characteristics in a pixel unit

Seihyoung Lee, Principal Researcher
AI Convergence Research Section
(Tel. 82-62-970-6596, seihyoung@etri.re.kr)

News 3

ETRI Achieves Significant Milestones at WTSA-24

- Secures Six Leadership Positions in ITU-T Study Groups for 2025-2028
- Drives the Adoption of New Resolutions on Metaverse and Digital Transformation

Electronics and Telecommunications Research Institute (ETRI) has achieved significant milestones at the World Telecommunication Standardization Assembly (WTSA-24), the highest decision-making body for telecommunications standardization, by securing six leadership positions. This achievement further solidifies South Korea’s pivotal role in setting essential directions for ICT standardization and undertaking key activities for the advancement of the ICT industry.

ETRI announced that at the World Telecommunication Standardization Assembly (WTSA-24) of the International Telecommunication Union (ITU)¹⁾, held from October 15 to 24 in New Delhi, India, it secured six leadership positions for the next study term (2025–2028) and successfully drove the adoption of two new resolutions. These accomplishments further reinforce ETRI’s leadership in the global standardization landscape.

ETRI secured the following leadership positions in ITU-T study groups for the upcoming term (2025–2028), namely Hyoung Jun KIM as Chair of the Study Group 20 (Internet of Things (IoT) and Smart Cities), Byung Chan KIM as Vice-Chair of the Study Group 5 (ICT and Circular Economy), Namseok KO as Vice-Chair of the Study Group 11 (Signaling, Protocols, and Test Specifications), Kangchan LEE as Vice-Chair of the Study Group 13 (Future Networks), Taesik CHEUNG as Vice-Chair of the Study Group 15 (Optical Transport Networks), and Shin-Gak KANG as Vice-Chair of the Study Group 21 (Multimedia and Broadcasting).

These leadership roles at international standardization bodies are highly significant, as they determine the direction of standardization and are critical to the success of establishing global standards. Notably, Dr. Hyoung Jun KIM, who has been re-elected as Chair of SG20, has been leading the group since March 2022. His continued leadership ensures that South Korea remains at the forefront of standardizing IoT and smart city technologies, which is expected to significantly enhance national technological competitiveness.

The vice-chairpersons appointed during this assembly are also expected to play crucial roles in expanding South Korea’s influence on international standardization efforts across various ICT domains.

These six leadership positions secured by ETRI are anticipated to have a significant impact, especially as these technologies are commercialized and widely adopted, potentially generating substantial economic benefits for the nation.

Moreover, eight new resolutions were approved during this assembly, including pivotal resolutions on the Metaverse and Digital Transformation, led by ETRI. These achievements reaffirm ETRI’s leadership in advancing next-generation ICT technologies.

ETRI further enhanced international cooperation by collaborating with 38 member countries of the Asia-Pacific Telecommunity (APT)²⁾, successfully leading the APT WTSA Preparatory Meeting (chaired by Dr. Hyoung Jun KIM), and preparing a total of 45 joint contributions throughout the WTSA-24 preparation process. This collaborative effort is expected to provide a solid foundation for South Korea to sustain its leadership in ICT standardization at the global level.

ETRI President Seung Chan Bang said, “The successful election of chairpersons and the approval of new resolutions at WTSA-24 represent major achievements that underscore ETRI’s technological capabilities and influence within international standardization organizations in this era of technological competition.”

ETRI will continue to lead international standardization activities, expand South Korea’s role, and strengthen the global competitiveness of the Korean ICT industry. Through global partnerships, ETRI aims to drive innovation in ICT technologies and lead initiatives for the advancement of the global ICT ecosystem. These efforts are expected to contribute to the growth of the global ICT industry and elevate South Korea’s technological standing.

The World Telecommunication Standardization Assembly (WTSA), held every four years by the ITU, is a major global event that sets the policy direction for telecommunications standardization and fosters international cooperation. WTSA-24 was attended by approximately 3,700 participants from 164 countries, with major outcomes including the approval of new standardization procedures, study topics, and key resolutions on the Metaverse and Digital Transformation.
1) World Telecommunication Standardization Assembly (WTSA-24) of the International Telecommunication Union (ITU): The highest decision-making body of the International Telecommunication Union’s Telecommunication Standardization Sector (ITU-T) that elects leadership position of each Study Group (SG), and discusses resolutions and recommendations for ICT standardization
2) APT: Asia-Pacific Telecommunity, an intergovernmental organization promoting ICT development in the Asia-Pacific region

Kangchan Lee, Director
Strategic Standards Research Section
(Tel. 82-42-860-6659, chan@etri.re.kr)