Connect
Handong News
Handong Global University and S-Pohang Hospital Collaborate on AI Technology Development for Diagnosing Dysphagia
writer: 영문 관리자   |   date: 2024.03.28   |   count: 109

(Photo 1) Presentation Material for Bolus Detection AI Model for Swallow Studies Featured in Paper

Professor In-Jung Kim’s research team at Handong Global University, led by President Dosoung Choi, has developed a medical image analysis AI model for diagnosing dysphagia (difficulty in swallowing) in collaboration with S Pohang Hospital.

If someone frequently experiences choking or has significant difficulty swallowing food, dysphagia may be suspected. Dysphagia refers to the inability to swallow food properly due to problems with nerves or muscles. Particularly, if there are symptoms of food entering the airway instead of the esophagus, serious complications such as pneumonia or suffocation can occur, highlighting the need for early diagnosis and treatment.

Diagnosis of this condition is usually achieved through a Video-Fluoroscopic Swallow Study (VFSS), which tracks the movement path of the bolus (a food bolus including contrast material) swallowed by the patient in X-ray images. However, because the shape of the swallowed bolus is not consistent, existing AI models often fail to accurately detect the bolus from unclear X-ray images.

(Photo 2) Professor In-Jung Kim introduced medical image analysis through deep learning technology at S-Pohang Hospital

The PECI-Net (Preprocessing Ensemble and Cascaded Inference Network), jointly developed by Handong Global University and S-Pohang Hospital, utilizes artificial neural networks to combine various image processing algorithms, improving the quality of X-ray images, and then accurately detecting the bolus through multi-stage inference, outperforming existing AI models. Professor In-Jung Kim, the principal investigator, stated, "PECI-Net consists of a Preprocessing Ensemble Network (PEN) to improve the quality of medical images and a Cascaded Inference Network (CIN) to resolve the ambiguity of medical images. The two networks are trained together to maximize synergy." He added, "We expect it to be effective not only in swallow studies but also in overcoming low-quality and ambiguous medical images in various medical scenarios."

This research involved several student researchers from Handong Global University, including Young Hun Kim, Ha Rim Kang, Jun Myung Lee, Jin Young Choi, and numerous medical professionals from S-Pohang Hospital, including Manager Deok Ho Park. It was published in the prestigious medical AI journal, 'Computers in Biology and Medicine' (IF 7.7).

Handong Global University focuses on actively introducing and utilizing artificial intelligence (AI) to strengthen educational and research capabilities required in the era of the Fourth Industrial Revolution, supporting various research activities.

facebook 공유하기 twitter 공유하기 kakaostory 공유하기 링크 공유
목록
Handong Global University
558 Handong-ro Buk-gu, Pohang Gyeongbuk 37554 Republic of Korea
Copyright (c) Handong Global University. All Rights Reserved.
Apply Get Info Connect