Colon Cancer Grading Using Infrared Spectroscopic Imaging-Based Deep Learning.

Area of Science:

• Oncology
• Biomedical Engineering
• Spectroscopy

Background:

• Accurate tumor grading is essential for effective cancer treatment planning.
• Traditional histopathology relies on hematoxylin and eosin (H&E) staining for morphologic evaluation.
• Fourier transform infrared (FT-IR) imaging offers molecular insights from unstained tissue.

Purpose of the Study:

• To explore the utility of FT-IR imaging for grading colon cancer in biopsy samples.
• To develop and validate a deep learning model for tumor grade estimation using FT-IR data.

Main Methods:

• A deep learning classifier was developed using FT-IR absorption data from 148 patients.
• The classifier was trained to estimate tumor grade based on molecular signatures.
• Validation was performed on an independent cohort of surgical resection samples.

Main Results:

• FT-IR imaging demonstrated potential as a viable tool for colorectal cancer grading.
• The developed deep learning model accurately estimated tumor grade.
• Successful validation on an independent cohort confirmed the model's robustness.

Conclusions:

• FT-IR imaging provides valuable molecular information for cancer grading.
• Combining FT-IR imaging with morphometry can lead to clinically relevant grade prediction models.
• This approach offers a promising alternative or adjunct to traditional histopathology for cancer assessment.