Predicting histopathological features of aggressiveness in lung cancer using CT radiomics: a systematic review

Plain Language Summary

A systematic review examining whether CT radiomics can predict aggressive histopathological features in lung cancer. Eleven studies were included, with radiomic model accuracies ranging from 0.67 to 0.94. Due to high risk of bias and concerns about applicability, evidence remains inconclusive, and prospective studies with external validation are needed before clinical implementation.

Abstract

CT radiomics has been proposed to predict histopathological features of aggressiveness in lung cancer. This systematic review examined the accuracy of radiomic models for predicting spread through air spaces, adenocarcinoma patterns, lymphovascular invasion, and other tumour characteristics. Eleven studies were included, predominantly from East Asia. Reported accuracies ranged from 0.67 to 0.94. However, due to high risk of bias and concerns regarding applicability, the evidence is inconclusive. Rigorously conducted prospective studies with external validation are needed before these models can be used to improve patient outcomes.