Automated Brain Metastases Detection Framework for T1-Weighted Contrast-Enhanced 3D MRI

Abstract

Brain Metastases (BM) complicate 20-40% of cancer cases. BM lesions can present as punctate (1 mm) foci, requiring high-precision Magnetic Resonance Imaging (MRI) in order to prevent inadequate or delayed BM treatment. However, BM lesion detection remains challenging partly due to their structural similarities to normal structures (e.g., vasculature). We propose a BMdetection framework using a single-sequence gadoliniumenhanced T1-weighted 3D MRI dataset. The framework focuses on detection of smaller (< 15 mm) BM lesions and consists of: (1) candidate-selection stage, using Laplacian of Gaussian approach for highlighting parts of a MRI volume holding higher BM occurrence probabilities, and (2) detection stage that iteratively processes cropped region-of-interest volumes centered by candidates using a custom-built 3D convolutional neural network ("CropNet"). Data is augmented extensively during training via a pipeline consisting of random gamma correction and elastic deformation stages; the framework thereby maintains its invariance for a plausible range of BM shape and intensity representations. This approach is tested using five-fold cross-validation on 217 datasets from 158 patients, with training and testing groups randomized per patient to eliminate learning bias. The BM database included lesions with a mean diameter of ~5.4 mm and a mean volume of ~160 mm 3 . For 90% BM-detection sensitivity, the framework produced on average 9.12 falsepositive BM detections per patient (standard deviation of 3.49); for 85% sensitivity, the average number of falsepositives declined to 5.85. Comparative analysis showed that the framework produces comparable BM-detection accuracy with the state-of-art approaches validated for significantly larger lesions.Index Terms-magnetic resonance imaging, brain metastases, convolutional neural networks, deep learning, scale-space representations, computer-aided detection, medical image analysis.

Deep convolutional neural networks for automated segmentation of brain metastases trained on clinical data

“…The final model was tested in two configurations, with the more sensitive one reaching an overall sensitivity of 0.82. Compared to previous studies the same model was highly specific with a specificity of 0.83 [7]. The data contained a broad range of metastases in regards to both primary cancer type and volume.…”

“…The first application which produced state-of-the-art results in automated segmentation of BM in MRI was published in 2015 by Losch et al [5]. Since then, a large variety of network architectures for deep learning including GoogLeNet [6], CropNet [7], DeepMedic [8] and En-DeepMedic [9] have been tested. A common limitation is the high number of false positives and the small sample sizes used for training.…”

“…Also, the performance of most of the algorithms heavily depended on the volume of the target lesion. Dikici et al identified this problem and developed an algorithm which was trained and performed well only on small BM (mean volume of 160 mm 3 ) [7], which suggests that a single model is unlikely to solve the segmentation problem for metastases of arbitrary size.…”

Integrating AI into radiology workflow: levels of research, production, and feedback maturity

“…), (2) analysis of statistical properties of the BM included in the study (e.g., lesion diameter, volume, location etc. ), and (3) adherence to data-acquisition criteria have been comprehensively described in a previous report 21 . This retrospective study was conducted under Institutional Review Board approval with waiver of informed consent.…”

“…The metric of Average False-Positives (AFP) per patient, representing the incorrectly detected BM lesions for each patient in relation to the sensitivity, was used during the validation of the algorithm for the three datasets 21 ; the AFP values were computed using a 5-fold cross validation (see Fig. 9).…”