Non-invasive PNET grading using CT radiomics and machine learning

Pancreatic cancer is a major cause of cancer-related fatalities globally, with a poor prognosis. Machine learning-based medical image analysis has emerged as a promising approach for improving clinical decision-making. The purpose is to determine the most effective machine learning method and phase of CT scan to provide clinicians with an efficient tool for accurately identifying pathological grades of pancreatic neuroendocrine tumours (PNET). This will be achieved by analysing contrast-enhanced computed tomography scans of both arterial and portal phases. An investigation was conducted on a cohort of 100 patients diagnosed with pancreatic neuroendocrine tumours. Radiomic features were extracted using Pyradiomics. These features were subsequently utilised in different machine learning classifiers. The classification model's performance was assessed using sensitivity, specificity, area under the curve (AUC) and accuracy metrics. Our analysis demonstrates that combining CT-based radiomic features with a machine-learning approach can identify the pathological grades of pancreatic neuroendocrine tumours. the combination of Portal_RFE and K-Nearest Neighbour (KNN) demonstrated the highest predictive performance with an AUC of 0.76 and 0.69 in training and validation models, respectively. The use of CT radiomic features and machine learning effectively determines PNET pathological grades, aiding in classifying patients for clinical decisions.