A deep convolutional neural network-based automatic delineation strategy for multiple brain metastases stereotactic...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurosurgery

Background:

• Accurate delineation of brain metastases is crucial for effective stereotactic radiosurgery (SRS).
• Manual segmentation is time-consuming and prone to inter-observer variability.
• Automated methods are needed to improve efficiency and consistency in SRS treatment planning.

Purpose of the Study:

• To develop and validate a deep learning convolutional neural network (CNN) algorithm for automatic brain metastases segmentation.
• To integrate the CNN algorithm into an automated workflow for clinical application.
• To evaluate the performance of the developed algorithm on benchmark and clinical datasets.

Main Methods:

• A deep learning convolutional neural network (CNN) algorithm was developed for segmenting brain metastases.
• The algorithm processed contrast-enhanced T1-weighted magnetic resonance imaging (MRI) datasets.
• The CNN-based algorithm was integrated into an automatic segmentation workflow and validated on BRATS and clinical data.

Main Results:

• Validation on BRATS data achieved average DICE coefficients (DCs) of 0.75±0.07 (tumor core) and 0.81±0.04 (enhancing tumor).
• Performance on clinical patient data showed an average DC of 0.67±0.03 and an area under the ROC curve of 0.98±0.01.
• The algorithm outperformed most techniques in the 2015 BRATS challenge.

Conclusions:

• The developed automatic segmentation strategy surpasses current benchmark levels.
• The CNN-based algorithm offers a promising tool for SRS treatment planning for multiple brain metastases.
• This automated approach enhances accuracy and efficiency in radiotherapy planning for brain tumors.