Effect of Anatomic Contextual Information on the Performance of a Convolutional Neural Network Tasked With Brain...

Modern management of brain metastases requires close surveillance with serial magnetic resonance imaging, creating interest in techniques for automated detection. The convolutional neural network (CNN) is currently the dominant approach to automated metastasis detection, and multiple variations of the CNN have been investigated to improve performance. In this work, we looked beyond the impact of network architecture and assessed the impact on performance of providing anatomic contextual information to a CNN during the training period.

The nnU-Net, a widely adopted CNN, was selected for this study. The nnU-Net was trained on an institutional data set comprising 301 high-resolution, T1-weighted, contrast-enhanced magnetic resonance images and associated structure labels consisting of both target brain metastases (1111 total) and several organs at risk (OARs) that were labeled in the process of radiosurgery planning. During the training period, the model was presented with either labeled brain metastases (BM) alone or with the complete structure set of metastases and OARs (BM+OAR). The test set contained 100 cases and 421 total metastases.

We found that, by multiple standard performance metrics, the BM+OAR model outperformed the BM model. Median case-wise Dice coefficient for BM was 0.75 (IQR, 0.51-0.86) and for BM+OAR was 0.79 (IQR, 0.58-0.86) (P < .001). Median sensitivity for BM was 1.0 (IQR, 0.71-1.0) and for BM+OAR was 1.0 (IQR, 0.75-1.0) (P < .01). Median positive predictive value for BM was 0.5 (IQR, 0.25-0.74) and for BM+OAR was 0.77 (IQR, 0.50-1.0) (P < .001). Importantly, among lesions scored as "false-positives," 47/151 detected by BM+OAR could be retrospectively verified to be real tumors, whereas only 22/231 detected by BM could be verified to be real (P < .001).

These findings suggest that providing anatomic contextual information can improve the accuracy of CNN-based automated brain metastasis detection algorithms and open new opportunities for research into the optimization of automated detection approaches for brain metastases.