[8a-PA3-1~13] 7 Beam Technology and Nanofabrication (Poster)

[8a-PA3-7] Segmentation of X-ray CT Images Using Integrated Features from Absorption and Phase-Contrast Images

〇Ryosuke Ueda¹, Yuki Goto², Hideki Tsuruta², Chika Kamezawa¹, Junya Kowa², Wataru Yashiro¹ (1.Tohoku Univ., 2.IHI)

Keywords:

X-ray Computed Tomography,X-ray Phase Imaging,Segmentation

X-ray CT enables three-dimensional visualization of internal structures; however, segmentation is often difficult for samples composed mainly of light elements because of their low absorption contrast. In this study, we investigated a segmentation method using integrated features from absorption-contrast and phase-contrast CT images. Segmentation was performed using a random forest classifier and evaluated using CT data acquired at the BL09W beamline of NanoTerasu. The results showed that phase-contrast features achieved higher classification performance than absorption-contrast features alone, while the integrated feature set provided the best performance. These findings suggest that absorption and phase information complement each other and improve segmentation accuracy.