JSAI2025

Presentation information

Poster Session

Poster session » Poster Session

[2Win5] Poster session 2

Wed. May 28, 2025 3:30 PM - 5:30 PM Room W (Event hall D-E)

[2Win5-71] The method of detecting asbestos in electron microscopy images by focusing background.

〇Masahiko Takei1, Mitsuyoshi Yoshida1, Fuminori Uematsu1 (1.JEOL Ltd.)

Keywords:AI, segmentation

Asbestos monitoring is mandatory at demolition sites to prevent atmospheric leakage. One procedure involves using electron microscopy images to observe air-sampled filters, counting fibrous substances, and then identifying asbestos through elemental analysis. Although several hundred images are taken per filter, we need reducing false positives to minimize time-consuming elemental analyses.
Images from electron microscopy are monochrome, making asbestos detection difficult due to its thin, variable shape and the similarity to filter material. Manual processing is time-consuming, and existing automatic detection methods often make many false positives. Higher magnification images improve asbestos shape clarity but reduce detection speed.
Therefore, we have considered a method for detecting asbestos using an AI-based segmentation technology without fine-tuning nor retraining. Our method aims to reduce false positives and total processing time so that we report on the method with the segmentation technology.

Authentication for paper PDF access
A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password