Machine learning characterization of cancer patients-derived extracellular vesicles using vibrational spectroscopies: results from a pilot study.

Abstract

Background: Patient-derived extracellular vesicles (EVs) that contains a complex biological cargo is a valuable source of liquid biopsy diagnostics to aid in early detection, cancer screening, and precision nanotherapeutics. In this study, we predicted that coupling cancer patient blood-derived EVs to timeresolved spectroscopy and artificial intelligence (AI) could provide a robust cancer screening and followup tools. Methods:In our pilot study, fluorescence correlation spectroscopy (FCS) measurements were performed on 24 blood samples-derived EVs. Blood samples were obtained from 15 cancer patients (presenting 5 different types of cancers), and 9 healthy controls (including patients with benign lesions). EVs samples were labeled with PKH67 dye. The obtained FCS autocorrelation spectra were processed into power spectra using the Fast-Fourier Transform algorithm. The processed power spectra were subjected to various machine learning algorithms to distinguish cancer spectra from healthy control spectra. Results and Applications:The performance of AdaBoost Random Forest (RF) classifier, support vector machine, and multilayer perceptron, were tested on selected frequencies in the N=118 power spectra. The RF classifier exhibited a 90% classification accuracy and high sensitivity and specificity in distinguishing the FCS power spectra of cancer patients from those of healthy controls. Further, an image convolutional neural network (CNN), ResNet network, and a quantum CNN were assessed on the power spectral images as additional validation tools. All image-based CNNs exhibited a nearly equal classification performance with an accuracy of roughly 82% and reasonably high sensitivity and specificity scores. Conclusion:Our pilot study demonstrates that AI-algorithms coupled to time-resolved FCS power spectra can accurately and differentially classify the complex patient-derived EVs from different cancer samples of distinct tissue subtypes. As such, our findings hold promise in the diagnostic and prognostic screening in clinical medicine. INTRODUCTION.Cancers globally remain amidst the leading-cause of disease-related mortality. Conventional therapies may be successful for certain subtypes of the disease, while others are complex adaptive systems progressing to clinically aggressive stages causing a paramount disease burden. Further, the long-term health complications and side-effects, successfully treated patients must live with, must be emphasized. Within this pilot study, in efforts to advance precision oncology and patient-centered clinical medicine, we explored the application of artificial intelligence (AI) in tackling one of the greatest challenges in preventive and diagnostic medicine: early cancer detection and prognostic screening. Cancer biomarker discovery was pioneered by Gold and Freedman (1965) with their recognition of the first tumor marker, Carcinoembryonic Antigen (CEA), which remains to date the most used clinically-relevant, blood-based cancer screening and diagnostic in patient-care. Their co-discovery o...

Discrimination of Healthy and Cancerous Colon Cells Based on FTIR Spectroscopy and Machine Learning Algorithms

“…Also, the SVM method was successfully used for the classification of spectra with good accuracy: D. Carvalho Caixeta et al used the ATR-FTIR tool associated with the SVM classifier in order to detect modifications to salivary components to be used as biomarkers for the diagnosis of type 2 diabetes mellitus with an accuracy of 87% [48]. The SVM algorithm was also able to distinguish the Raman spectra of extracellular vesicles in the serum of cancer patients from those of healthy controls with a classification accuracy of 100% when reduced to the spectral frequency range from 1800 to 1940 cm −1 , although the accuracy values significantly decreased to 67% and 57% when the complete Raman spectrum and FTIR spectrum, respectively, were used [49]. Good classification performances were also reported for the kNN model.…”

Surface‐Enhanced Raman Spectroscopy‐Based Optical Biosensor for Liquid Biopsy: Toward Precision Medicine

“…[97] Features can be extracted and mathematical models can be built from a large amount of SERS spectral data using suitable ML methods, thereby achieving precise diagnosis or even disease staging and classification when combined with the proposed SERS strategies to guide clinical sample analysis. [95,[98][99][100][101] As shown in Figure 7A, Hou et al combined SERS with principal component analysis and a support vector method (PCA-SVM) to realize the detection of Sjögren's syndrome (SS) and diabetic nephropathy (DN), with excellent diagnostic accuracy (90.7% for SS and 89.3% for DN), sensitivity (93.4% for SS and 95.6% for DN), and selectivity (86.7% for SS and 80% for DN). [102] In synergy with SVM, Liz-Marzán et al identified and extracted features of cellular secretions under different conditions using special microfluidic chips, enabling high-throughput assessment of anticancer treatment efficacy (Figure 7B).…”

Advances in Machine Learning Processing of Big Data from Disease Diagnosis Sensors

“…Biosensors based on DNA nanotechnology, nanomaterials, nanoplasma technology, mass spectrometry, and single-molecule technology are rapidly advancing because of their inherent advantages, including high sensitivity, specificity, and controllability. These advancements facilitate joint diagnosis of multiple biomarker sensors. ,,,,,,,,− For example, the electrochemical methylation-specific polymerase chain reaction (E-MSP) constructed on the basis of framework nucleic acids enables quantitative and highly selective DNA methylation analysis (Figure C) . The investigators employed this method to determine the levels of DNA methylation in 12 selected gene promoters across prostate cancer cell lines, cancer tissues, serum samples from both cancer and benign prostatic hypertrophy (BPH) cases, and serum samples from normal individuals.…”