A Comparison of Image Segmentation Methods on Low Contrast Cellular Fluorescence Images

カテゴリ: 部門大会

論文No: PS2-4

グループ名: 【C】平成29年電気学会電子・情報・システム部門大会講演論文集

発行日: 2017/09/06

タイトル(英語): A Comparison of Image Segmentation Methods on Low Contrast Cellular Fluorescence Images

著者名: Moiloa Pelonomi(東北大学),本間 経康(東北大学),小山内 実(東北大学)

著者名(英語): Pelonomi Moiloa|Noriyasu Homma|Makoto Osanai

キーワード: カルシウムイメージング|セグメンテーション|神経細胞神経細胞|calcium imaging|segmentation|neuron

要約(日本語): It is difficult to quantify which of the many segmentation methods are best suited to a particular dataset. This study compares manual ROI selection to differing semi automated segmentation methods in their ability to segment time series low contrast cellular fluorescence images. The ROI identified by application of these methods to an image stack are analysed in Matlab to quantify; number of cells, number of cells which contain a calcium trace and signal to noise ratios of said calcium traces. Manual ROI selection is found to offer the most reliable solution for the low contrast images used in this study. The remaining methods explored may help to speed up the manual ROI selection process by acting as reference but require further development in order to qualify as viable replacements.

PDFファイルサイズ: 479 Kバイト