# Radiomic Analysis for Ki-67 Classification in Small Bowel Neuroendocrine Tumors

## TL;DR

This study explores using radiomic features from CT scans to predict Ki-67 levels in small bowel neuroendocrine tumors, offering a non-invasive way to assess tumor aggressiveness.

## Contribution

The study introduces a radiomic approach for non-invasive Ki-67 classification in small bowel neuroendocrine tumors, demonstrating promising predictive performance.

## Key findings

- Random Forest achieved the best performance with an AUC of 0.80 and a low false negative rate of 15.3%.
- Eight top-performing radiomic features were identified through multiple ranking methods.
- Radiomic analysis showed potential for non-invasive assessment of proliferative rates in small bowel neuroendocrine tumors.

## Abstract

Although neuroendocrine tumors are generally rare, they are the most common malignant neoplasm in the small intestine and the second most common gastrointestinal neuroendocrine location. Contrary to the historical conception of neuroendocrine neoplasms as indolent and non-aggressive, a significant percentage of cases present with lymph node or distant metastases at diagnosis. Nevertheless, they are significantly less studied in the literature than pancreatic neuroendocrine tumors. Histopathological evaluation still plays a crucial role in determining the prognosis and tailoring the treatment of patients with NETs. Radiomics is a quantitative analysis technique that enables the extraction and analysis of features imperceptible to the human eye from medical images with the aim of quantifying tumor imaging characteristics. In this study, we decided to investigate radiomic features extracted from CT images, focusing on small bowel NETs and evaluating their association with Ki-67 expression.

Objective: To analyze radiomic features extracted from CT images of small bowel neuroendocrine tumors and evaluate their association with Ki-67 expression. Methods: 128 small bowel NET primary and secondary lesions from 34 patients were analyzed. Manual segmentation of the lesions was conducted on portal-phase CT images using ITK-SNAP v. 4.0®, and 107 radiomic features were extracted using the PyRadiomics library. The lesions were categorized into two groups based on their Ki-67 index expression (≤1% and >1%). Correlation filtering reduced the set of 107 to 41 radiomic features. Inferential statistical analyses (t-test and Mann–Whitney U, following Shapiro–Wilk and Levene’s tests) identified 19 significant features (p < 0.05) that were predominantly texture related. A ranking procedure further reduced these to eight top-performing variables across multiple selection methods (Information Gain, Gini, ANOVA, χ2). Five supervised Machine Learning models (Logistic Regression, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), XGBoost, and Random Forest) were trained and validated using 5-fold cross-validation. The evaluation metrics employed included AUC, accuracy, precision, recall, F1 score, and a confusion matrix. Results: Random Forest exhibited the best overall performance (AUC = 0.80; F1 score = 0.813; Recall = 0.847). The model’s low false negative rate (15.3%) suggests potential clinical utility in minimizing the risk of underestimating more aggressive lesions. Conclusions: Radiomics represents a promising frontier to identify patterns associated with histopathological markers. This study highlights its potential for non-invasive assessment of proliferative rate in small bowel neuroendocrine tumors, confirming the performance in the literature, and posing an interesting prospect for future research.