Multi photon micro material analysis based on Raman spectroscopy biosensor for cancer detection using biomarker with deep learning techniques

Due to singularity of Raman Spectroscopy (RS) measurements in revealing molecular biochemical changes between cancerous and normal tissues and cells, RS was recently demonstrated to be a non-destructive method of cancer diagnosis. The quantity and quality of tissue samples for RS are crucial for accurate prediction when developing computational methods for cancer detection. The training of the classifier with a small number of samples is difficult and frequently leads to overfitting. As a result, increasing the number of samples is crucial for better training classifiers that can accurately classify cancer tissue. This study proposes a novel method for the detection of spine cancer using a high-sensitivity biosensor and edge elimination based on Raman spectroscopy by Multi photon micro material analysis. Using a high-sensitivity biosensor and Raman spectroscopy, the input tumour image edge is removed. F-fluorodeoxy glucose PET imaging (FDG-PET) applied images are used to identify the cancer-affected region in this edge-minimized image, and a Lasso regressive-based reinforcement neural network is used to analyse the 511-keV photons. The accuracy, F-measure, recall, dice coefficient, Areas under the curve (AUC), and neuron-specific enolase (NSE) are all used in the experimental analysis. proposed technique attained accuracy of 98%, F-measure of 75%, Recall of 65%, dice coefficient of 55%, AUC of 63% and NSE of 59%.