Radiomics in Predicting Tumour Response to Radiation Therapy in Lung Cancer: A multi-institutional Study

Abstract

Radiomics is a state-of-the-art image analysis technology that has provided significant insight into the assessment of response to RT in lung cancer. This multimodal study provides an examination of the value of radiomic attributes for lung cancer computed tomography (CT) images acquired before therapy. In all, 500 patients in 5 related centres were studied, and the radiomic features were compared to clinical parameters such as regression, and survival rates. In an aspect, feature selection algorithms were applied to define parameters that were used in the machine learning model to make the necessary predictions for assessment. This study proves that certain texture and shape biomarkers were capable of predicting response to treatment as well as the survival of the patient. The proposed model obtained a prediction rate of 85%, providing the patients with a less invasive solution for assessing the best treatment scheme. The external dataset’s comparison with the internal model also proved the model’s credibility. As shown in this review, radiomics has significant potential for optimizing the results of radiation therapy for lung cancer by developing personalized treatment regimens. Further investigation should be directed to the expanded verification of the proposed approach and its inclusion as one of the features of clinical decision support systems.