Transfer learning approach based on computed tomography images for predicting late xerostomia after radiotherapy in patients with oropharyngeal cancer

Annarita Fanizzi; Giovanni Scognamillo; Alessandra Nestola; Santa Bambace; Samantha Bove; Maria Colomba Comes; Cristian Cristofaro; Vittorio Didonna; Alessia Di Rito; Angelo Errico; Loredana Palermo; Pasquale Tamborra; Michele Troiano; Salvatore Parisi; Rossella Villani; Alfredo Zito; Marco Lioce; Raffaella Massafra

doi:10.3389/fmed.2022.993395

Transfer learning approach based on computed tomography images for predicting late xerostomia after radiotherapy in patients with oropharyngeal cancer

Front Med (Lausanne). 2022 Sep 23:9:993395. doi: 10.3389/fmed.2022.993395. eCollection 2022.

Authors

Annarita Fanizzi¹, Giovanni Scognamillo¹, Alessandra Nestola¹, Santa Bambace², Samantha Bove¹, Maria Colomba Comes¹, Cristian Cristofaro¹, Vittorio Didonna¹, Alessia Di Rito², Angelo Errico², Loredana Palermo¹, Pasquale Tamborra¹, Michele Troiano³, Salvatore Parisi³, Rossella Villani¹, Alfredo Zito¹, Marco Lioce¹, Raffaella Massafra¹

Affiliations

¹ IRCCS Istituto Tumori "Giovanni Paolo II," Bari, Italy.
² Ospedale Monsignor Raffaele Dimiccoli, Barletta, Italy.
³ IRCCS Casa Sollievo della Sofferenza, Opera di San Pio da Pietrelcina Viale Cappuccini, Foggia, Italy.

Abstract

Background and purpose: Although the latest breakthroughs in radiotherapy (RT) techniques have led to a decrease in adverse event rates, these techniques are still associated with substantial toxicity, including xerostomia. Imaging biomarkers could be useful to predict the toxicity risk related to each individual patient. Our preliminary work aims to develop a radiomic-based support tool exploiting pre-treatment CT images to predict late xerostomia risk in 3 months after RT in patients with oropharyngeal cancer (OPC).

Materials and methods: We performed a multicenter data collection. We enrolled 61 patients referred to three care centers in Apulia, Italy, out of which 22 patients experienced at least mild xerostomia 3 months after the end of the RT cycle. Pre-treatment CT images, clinical and dose features, and alcohol-smoking habits were collected. We proposed a transfer learning approach to extract quantitative imaging features from CT images by means of a pre-trained convolutional neural network (CNN) architecture. An optimal feature subset was then identified to train an SVM classifier. To evaluate the robustness of the proposed model with respect to different manual contouring practices on CTs, we repeated the same image analysis pipeline on "fake" parotid contours.

Results: The best performances were achieved by the model exploiting the radiomic features alone. On the independent test, the model reached median AUC, accuracy, sensitivity, and specificity values of 81.17, 83.33, 71.43, and 90.91%, respectively. The model was robust with respect to diverse manual parotid contouring procedures.

Conclusion: Radiomic analysis could help to develop a valid support tool for clinicians in planning radiotherapy treatment, by providing a risk score of the toxicity development for each individual patient, thus improving the quality of life of the same patient, without compromising patient care.

Keywords: CNN–convolutional neural network; CT images; deep learning; oropharyngeal cancer; xerostomia.