Brain metastasis detection using machine learning: a systematic review and meta-analysis

Abstract

Background Accurate detection of brain metastasis (BM) is important for cancer patients. We aimed to systematically review the performance and quality of machine-learning-based BM detection on MRI in the relevant literature. Methods A systematic literature search was performed for relevant studies reported before April 27, 2020. We assessed the quality of the studies using modified tailored questionnaires of the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) criteria and the Checklist for Artificial Intelligence in Medical Imaging. Pooled detectability was calculated using an inverse-variance weighting model. Results A total of 12 studies were included, which showed a clear transition from classical machine learning (cML) to deep learning (DL) after 2018. The studies on DL used a larger sample size than those on cML. The cML and DL groups also differed in the composition of the data set, and technical details such as data augmentation. The pooled proportion of detectability of BM were 88.7% (95% CI, 84–93%) and 90.1% (95% CI, 84–95%) in the cML and DL groups, respectively. The false-positive rate per person was lower in the DL group than the cML group (10 vs. 135, P < 0.001). In the patient selection domain of QUADAS-2, three studies (25%) were designated as high risk due to non-consecutive enrollment and arbitrary exclusion of nodules. Conclusion A comparable detectability of BM with a low false-positive rate per person was found in the DL group compared to the cML group. Improvements are required in terms of quality and study design.

Deep Learning-Based Computer-Aided Detection System for Automated Treatment Response Assessment of Brain Metastases on 3D MRI

“…In addition, we further performed validations using another temporally separated dataset and data from another institution. Cho et al conducted a systematic review and meta-analysis of 12 studies on BM detection by machine learning ( 46 ). They found that only two studies included consecutive patients and conducted an external validation or temporal separation of test data.…”

Systematic review and meta-analysis of lung cancer brain metastasis and primary tumor receptor expression discordance

“…EGFR resistant mutations can often be discovered in plasma from NSCLC patients before any clinical symptoms of progression, suggesting that monitoring circulating DNA levels and mutational profiles during the course of the disease could lead to earlier treatment intervention [27,28]. Advanced imaging and radiomics research could potentially represent a non-invasive approach for predicting tumor immunophenotype, but these approaches are still in an early developmental phase [29]. Till date no model has been developed to predict the LCBM immunophenotype based on patient features and treatment details.…”

Artificial Intelligence in Brain Tumour Surgery—An Emerging Paradigm

“…In recent years, there has been a marked transition away from conventional ML programs to deep learning programs, which integrate numerous layers of neural networks akin to human cortical processing, resulting in powerful systems capable of more complex and subtle pattern recognition [ 72 ]. Comparisons in the accuracy of brain metastases detection showed that the deep learning group had a statistically significantly lower rate of false-positives per person when compared with conventional ML, suggesting an ever growing accuracy [ 72 ].…”