Page 58 - חוברת תזות הנדסה ירוקה 2023

76.8%-85.1%. Since most of the misclassified samples were between the pre-cancerous and cancerous categories, it was decided to also perform binary classification. To that end, the NIHs and MBMs were combined into one ‘abnormal category’. This second stage, meant to distinguish between the ‘normal’ and ‘abnormal’ systems, was based on our research database; the classifier’s success rate was 96.9%. When analyses were performed between the NIHs and the MBMs )based on our research database(, the classifier’s success rate was 82.2%. When the analyses were based on different ranges of the Raman spectrum, a success rate of 97.8% was achieved for the differentiation of the NIHs from the MBMs, in the carbohydrate range 1195-600 cm-1, based on measurements taken from the cytoplasm. The results of this study are encouraging and show that there is great potential for the use of Raman spectroscopy-based machine learning in the correct identification and diagnosis of pre-cancerous and cancerous cells.
Keywords: Cancer, Machine learning, Medical waste, Raman spectroscopy.
Peer reviewed papers and Poster presentations in conferences SharahaU.,HaniaD.,LapidotI.,HuleihelM.,SalmanA. EarlyDetectionofPre-cancerousand Cancerous Cells Using Raman Spectroscopy-Based Machine Learning. Analyst, Accepted 2023. In press.
Sharaha U., Hania D., Lapidot I., Huleihel M., Salman A. Characterization and Detection of Precancerous and Cancerous Cells Using Raman Spectroscopy and Machine Learning Algorithms. Analytical Chemistry, Accepted 2023. In press.
Raman Spectroscopy Combined with Machine Learning for Differentiation among Cancerous, Pre-cancerous and Primary cells. 12th SPEC 2022, Dublin, Ireland.
Characterization and Detection of Mouse Primary and Malignant Cells Using Raman Spectroscopy and Machine Learning. 55th ISM 2022, Ben-Gurion University of the Negev, Beer-Sheva, Israel.
Book Of Abstracts | Class 2022
58