| Creator |
|
|
|
|
|
|
|
|
|
|
|
|
|
| Language |
|
| Publisher |
|
|
|
| Date |
|
| Source Title |
|
| Vol |
|
| Issue |
|
| First Page |
|
| Last Page |
|
| Publication Type |
|
| Access Rights |
|
| JaLC DOI |
|
| Abstract |
Conventional odor discrimination is generally performed by gas chromatography–mass spectrometry (GC–MS) that identifies specific marker molecules. Such marker identification process is, however, labor...-intensive, and the limited number of identified marker molecules is often insufficient to discriminate complex odors. In this study, we have demonstrated a facile method for discriminating complex odors with GC–MS data by combining texture image analysis (TIA) and machine learning (ML). We extracted various texture features (i.e., contrast, energy, homogeneity, correlation, dissimilarity and angular second moment) of two-dimensional (2D) MS maps by TIA, and used them as datasets for ML. Each texture feature contains a lot of molecular information appeared in 2D MS maps, and thus serves as an effective parameter for discriminating complex odors. Based on this method, we successfully performed the discrimination of breath samples collected from the persons of different blood glucose levels with higher performances and reliability than the conventional approach.show more
|
Export Mendeley
4302_p042