Abstract
<h3>Purpose/Objective(s)</h3> Dissociation between clinical and pathologic lymph node status is always noted in patients with squamous cell carcinoma of oral cavity (OSCC), especially at extranodal extension (ENE) status. Also, the lymph node texture after underwent radiotherapy and chemotherapy is considered different and cause difficulty in recognition. We used a deep learning neural network from preoperative computed tomography (CT scan) for accuracy improvement of ENE status for patients with OSCC. <h3>Materials/Methods</h3> Patients who diagnosed as OSCC with in our hospital were enrolled from 2019 to 2022. All enrolled patients underwent curative or salvage ipsilateral/bilateral neck lymph node dissection with preoperative contrast enhanced head and neck CT scan within 30 days before operation. The radiation-oncologist segmented, and annotated lymph nodes based on pathologic reports. Total two hundred and fourteen patients were enrolled, and 196 metastatic lymph nodes were identified from 402 dissected lymph nodes. Positive ENE status found in 130 metastatic lymph nodes. We divided them into two groups: curative treatment and salvage treatment. In curative group, 203 nodes in training set and 60 nodes in testing set. In salvage group 139 nodes in training set and 60 nodes in testing set. We used deep learning neural network (DLNN) of 2 different size input layers (dual-input), with framework of convolutional layers, the model with 74 epochs. <h3>Results</h3> The result AUC is 0.84 on salvage group and 0.82 on curative treatment patients, outperforming 0.69 of clinical experts. Under stratified analysis, patient with higher pathological nodal stage, achieved higher accuracy. <h3>Conclusion</h3> Deep learning algorithm is a useful tool in identification of lymph node ENE status from CT scan for patients with OSCC. Our work proved the deep learning model can also utilize on patient experienced radiotherapy and chemotherapy. The future prospective trial can evaluate the practicality when deploy in clinical decision making.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: International Journal of Radiation Oncology*Biology*Physics
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.