Development of programs to predict the occurrence of mucositis from digital imaging and communications in medicine data by machine learning in head and neck volumetric modulated radiotherapy

Toshiya Rachi ¹, Takaki Ariji ¹, Shinichi Takahashi ¹

⁰Department of Radiological Technology, National Cancer Center Hospital East, Kashiwa, Japan.

Journal of Applied Clinical Medical Physics +

|

August 21, 2023

View abstract on PubMed

Summary

Machine learning accurately predicts severe mucositis in head and neck cancer patients undergoing VMAT with cisplatin. This predictive tool aids in preventing treatment delays and supports early intervention for radiotherapy patients.

Area Of Science

Oncology
Medical Physics
Radiotherapy

Background

Head and neck cancer patients treated with VMAT and cisplatin often develop severe mucositis.
Current management lacks standardized prevention and treatment protocols for mucositis.
Adaptive radiotherapy (ART) is used for severe mucositis but requires re-planning during treatment.

Purpose Of The Study

To develop a machine learning program for predicting mucositis requiring ART in head and neck cancer patients.
To identify quantitative features from treatment plans associated with severe mucositis development.

Main Methods

Retrospective analysis of treatment plans (DICOM data) from 61 patients receiving concurrent chemotherapy and radiotherapy (RT).
Extraction of quantitative features including equivalent square field size, dose per segment, clinical target volume, and mean oral cavity dose.
Application of Support Vector Machine (SVM) and K-nearest neighbor (KNN) algorithms for machine learning classification.

Main Results

Machine learning models (SVM and KNN) achieved high accuracy (0.981 ± 0.020 and 0.972 ± 0.033, respectively) in classifying the need for ART due to mucositis.
The developed program accurately predicted the onset of mucositis requiring ART before treatment initiation.
Classification of a five-dimensional data list was achieved with high accuracy.

Conclusions

A novel machine learning program can accurately predict mucositis requiring ART in head and neck cancer patients.
This predictive capability may facilitate early preventive measures and ensure uninterrupted radiotherapy completion.
The study supports the potential for proactive management of radiotherapy-induced mucositis.

Keywords:

head and neck cancer machine learning mucositis volumetric modulated arc therapy