Spectroscopic Nuclear Magnetic Resonance and Fourier Transform–Infrared Approach Used for the Evaluation of Healing After Surgical Interventions for Patients with Colorectal Cancer: A Pilot Study

TL;DR

This pilot study explores using NMR and FT-IR spectroscopy to evaluate healing in colorectal cancer patients after surgery, showing a nonlinear recovery process.

Contribution

The study introduces a novel combination of NMR and FT-IR with machine learning to assess post-surgical healing in CRC patients.

Findings

Native blood plasma samples better predict CRC patient evolution 7 days post-surgery.

Nonlinear healing states were observed between preoperative and healthy states.

Machine learning maps successfully identified medical state probabilities.

Abstract

Native and deproteinized blood plasma collected from 10 patients with confirmed CRC, before and 7 days after surgery, and from 20 healthy volunteers were measured by 1H NMR T2 relaxometry and FT-IR spectroscopy and statistically analyzed by PCA, ROC and AUC and by prediction maps using machine learning-based ANN. 1H NMR relaxometry and FT-IR spectroscopy methods combined with numeric analysis methods demonstrated that the native blood plasma samples can be better used to predict the evolution of patients with colorectal cancer at 7 days after surgery. Successful individual and group evolutions were discussed and a nonlinear healing evolution was observed and evaluated.

Background/Objectives: Colorectal cancer (CRC) is one of the most common and deadly types of cancer. Compared with the classical histopathological approach, this study discusses the application of 1H NMR and FT-IR techniques for the fast evaluation degree of healing of patients with CRC after surgical intervention. Methods: Native and deproteinized blood plasma collected from 10 patients with confirmed CRC and 20 healthy volunteers were analyzed using 1H NMR T2 distributions and FT-IR spectra measured for samples collected before and 7 days after surgery. The average FT-IR spectrum from 20 healthy volunteers is also presented. Principal component analysis (PCA) was performed on the FT-IR spectra. The results were used for further statistical analysis using receiver operating characteristic (ROC) and area under the curve (AUC) and to produce a series of prediction maps using a machine learning library. Results: Both experimental methods combined with analysis methods demonstrated that the native blood plasma samples can be better used to predict the CRC patients’ evolution 7 days after surgery. Three patients showed a significant evolution by 1H NMR T2 distribution, correlated to the observation of FT-IR–PCA analysis. Maps of medical state probability were generated using a trained machine learning-based ANN. Conclusions: The experimental measurements combined with an advanced statistical analysis and machine learning were successfully used and show that the healing process of patients with CRC is not linear, from the preoperative state to the state associated with healthy volunteers, but passes through a distinct healing state

Figures