본문 바로가기
  • Home

A Cloud-Based Training-Free Retinal Vessel Analysis System with Automatic Clinical Report Generation

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2026, 31(1), pp.91~97
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : November 6, 2025
  • Accepted : January 9, 2026
  • Published : January 30, 2026

Ji-Won Han 1 Ji-yeon So 2 Eun-sol Noh 2 Ye-na Lee 2 Sejong Lee 3

1한양대학교
2한양대학교 ERICA캠퍼스
3영남대학교

Accredited

ABSTRACT

This study proposes a cloud-based diagnostic system that segments retinal vessels from fundus images and automatically generates clinical diagnostic reports. The proposed system operates without any additional model training by utilizing a training-free analysis pipeline based on image preprocessing and postprocessing techniques. In particular, it is designed to be integrated with hospital cloud infrastructure to enable real-time diagnosis and includes a reporting module that produces clinically applicable outputs. This paper describes the image processing pipeline for vessel segmentation, the text-based report generation process, and the cloud integration architecture, along with both qualitative and quantitative evaluations. Experimental results demonstrate that even without model training, the proposed system achieves meaningful vessel segmentation performance and reliable report generation capability.

Citation status

* References for papers published after 2024 are currently being built.