Classification pipeline for in vivo Raman spectroscopy-aided colorectal cancer detection

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60461373%3A22340%2F23%3A43927335" target="_blank" >RIV/60461373:22340/23:43927335 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Classification pipeline for in vivo Raman spectroscopy-aided colorectal cancer detection

Popis výsledku v původním jazyce

With more than 1.9 million new diagnoses per year [1], colorectal cancer is among the leading globalcauses of death in cancer patients. While the definitive diagnosis usually involves biopsy, in vivo Ramanspectroscopy, a less invasive examination method, has shown great potential to discriminate betweennormal and cancerous tissue [2] . However, the absence of a suitable classifier as well as the timeconsuming and expertise demanding pre-processing of such in vivo Raman spectra are the mainobstacles to the adoption of this minimally invasive technique in clinical practice. By developing a real-time classification pipeline coupled with a user-friendly utility for non-spectroscopists, we look toremedy these obstacles. In addition to routine colonoscopy, in vivo Raman spectra of healthy andcancerous colorectal tissue were acquired using a custom-made microprobe. The spectra were thenloaded into the pipeline and pre-processed in several steps, including normalisation and finite impulseresponse filtration. The quality of the pre-processed spectra was checked using signal-to-noise ratiobefore the suitable spectra were decomposed and classified using principal component analysis andrandom forest, respectively. Additionally, a utility with a graphical user interface was developed tofacilitate the use of our data pipeline by non-spectroscopist in a clinical environment. Overall, thecombination of algorithmic preprocessing of in vivo measured Raman spectra with supervised andunsupervised machine learning appears to be a viable way of reducing the relatively large number ofbiopsies currently needed to definitively diagnose colorectal cancer.

Název v anglickém jazyce

Classification pipeline for in vivo Raman spectroscopy-aided colorectal cancer detection

Popis výsledku anglicky

With more than 1.9 million new diagnoses per year [1], colorectal cancer is among the leading globalcauses of death in cancer patients. While the definitive diagnosis usually involves biopsy, in vivo Ramanspectroscopy, a less invasive examination method, has shown great potential to discriminate betweennormal and cancerous tissue [2] . However, the absence of a suitable classifier as well as the timeconsuming and expertise demanding pre-processing of such in vivo Raman spectra are the mainobstacles to the adoption of this minimally invasive technique in clinical practice. By developing a real-time classification pipeline coupled with a user-friendly utility for non-spectroscopists, we look toremedy these obstacles. In addition to routine colonoscopy, in vivo Raman spectra of healthy andcancerous colorectal tissue were acquired using a custom-made microprobe. The spectra were thenloaded into the pipeline and pre-processed in several steps, including normalisation and finite impulseresponse filtration. The quality of the pre-processed spectra was checked using signal-to-noise ratiobefore the suitable spectra were decomposed and classified using principal component analysis andrandom forest, respectively. Additionally, a utility with a graphical user interface was developed tofacilitate the use of our data pipeline by non-spectroscopist in a clinical environment. Overall, thecombination of algorithmic preprocessing of in vivo measured Raman spectra with supervised andunsupervised machine learning appears to be a viable way of reducing the relatively large number ofbiopsies currently needed to definitively diagnose colorectal cancer.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10406 - Analytical chemistry

Projekt

<a href="/cs/project/NU20-09-00229" target="_blank" >NU20-09-00229: Vývoj nových analytických přístupů pro včasné odhalení adenomatózních polypů a prevenci kolorektálního karcinomu</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů