Computer‐aided detection of metastatic brain tumors using automated three‐dimensional template matching

Abstract

Purpose: To demonstrate the efficacy of an automated three-dimensional (3D) template matching-based algorithm in detecting brain metastases on conventional MR scans and the potential of our algorithm to be developed into a computer-aided detection tool that will allow radiologists to maintain a high level of detection sensitivity while reducing image reading time. Materials and Methods:Spherical tumor appearance models were created to match the expected geometry of brain metastases while accounting for partial volume effects and offsets due to the cut of MRI sampling planes. A 3D normalized cross-correlation coefficient was calculated between the brain volume and spherical templates of varying radii using a fast frequency domain algorithm to identify likely positions of brain metastases.Results: Algorithm parameters were optimized on training datasets, and then data were collected on 22 patient datasets containing 79 total brain metastases producing a sensitivity of 89.9% with a false positive rate of 0.22 per image slice when restricted to the brain mass.Conclusion: Study results demonstrate that the 3D template matching-based method can be an effective, fast, and accurate approach that could serve as a useful tool for assisting radiologists in providing earlier and more definitive diagnoses of metastases within the brain.

3D blob based brain tumor detection and segmentation in MR images

“…The true positive over false positive ratio T p : F p in [10] is approximately 1:10 versus 1:2.7 per brain in our case. Fig.…”

“…Subsequent affine adaptation and asymmetrybased pruning stages result in low false positive F p tumor detections. Our average 95.3% detection rate, average 3-10 false positives F p per brain (Table 1), and under 3 minute run-time (on a standard PC running Intel Core i7) are an improvement over state-of-the-art algorithms (average detection rate 90%, average F p per brain 34.8 by [9][10], and 30 minutes using template matching in [10]). The detection results can also be applied to other brain abnormality detection tasks using the final 3D blobs as features.…”

Radiomics in Brain Tumor: Image Assessment, Quantitative Feature Descriptors, and Machine-Learning Approaches

“…Computational image descriptors quantify visual characteristics at different scales from ROIs, which can be readily translated into radiological image analysis pertaining to tumor volumetric shapes and visual appearance dynamics. For example, the scale-invariant feature transform (SIFT) 2122 is computed through key point detection by using a difference-of-gaussians (DoG) function and local image gradient measurment with radius and scale selections (as illustrated in Fig.1). This permits a quantitative measurement of the tumor shape so that subtle variation during treatment (i.e.…”

Brain metastases: neuroimaging

“…The addition of these algorithms to human interpretation could potentially lead to greater sensitivity and improved accuracy of intracranial metastases characterization, based both on standard postcontrast exams as well as advanced physiologic imaging such as perfusion-related acquisitions (Ambrosini et al, 2010; Yang et al, 2013; Szwarc et al, 2015). …”