Prospective Observational Study to Predict Severe Oral Mucositis Associated With Chemoradiotherapy in Nasopharyngeal Carcinoma Based on Deep Learning

Prospective Observational Study to Predict Severe Oral Mucositis Associated With Chemoradiotherapy in Nasopharyngeal Carcinoma Based on Deep Learning

The goal of this observational study is to apply the CNN-based DL method to extract the three-dimensional spatial information of IMRT dose distribution to predict the occurrence probability of serious radiotherapy and chemotherapy induced oral mucositis(SRCOM), and compare with a model based on dosimetry, NTCP or doseomics to improve the prediction accuracy of SRCOM, thus guiding the clinical planning design, reducing the occurrence probability of OM, and may have the potential value of preventing serious complications and improving the quality of life in patients with nasopharyngeal carcinoma.

Inclusion Criteria:

• Initial diagnosis, pathological histological diagnosis, the pathological type is non-keratotic carcinoma (according to the WHO pathological classification).
• Initial intensity-modulated radiotherapy (Intensity modulated radiation therapy, IMRT).
• No previous radiotherapy was received.

Exclusion Criteria:

• Patients with recurrent nasopharyngeal carcinoma.
• Radiotherapy plan cannot be obtained.
• Previous history of malignancy; previous radiotherapy.
• The primary lesion and cervical metastatic lesions have received surgical treatment (except for diagnostic treatment).