Quantitative analysis of brain herniation from non-contrast CT images using deep learning.

BACKGROUND: Brain herniation is one of the fatal outcomes of increased intracranial pressure (ICP). It is caused due to the presence of hematoma or tumor mass in the brain. Ideal midline (iML) divides the healthy brain into two (right and left) nearly equal hemispheres. In the presence of hematoma, the midline tends to shift from its original position to the contralateral side of the mass and thus develops a deformed midline (dML). NEW METHOD: In this study, a convolutional neural network (CNN) was used to predict the deformed left and right hemispheres. The proposed algorithm was validated with non-contrast computed tomography (NCCT) of (n = 45) subjects with two types of brain hemorrhages - epidural hemorrhage (EDH): (n = 5) and intra-parenchymal hemorrhage (IPH): (n = 40)). RESULTS: The method demonstrated excellent potential in automatically predicting MLS with the average errors of 1.29 mm by location, 66.4 mm2 by 2D area, and 253.73 mm3 by 3D volume. Estimated MLS could be well correlated with other clinical markers including hematoma volume - R2 = 0.86 (EDH); 0.48 (IPH) and a Radiologist-defined severity score (RSS) - R2 = 0.62 (EDH); 0.57 (IPH). RSS was found to be even better correlated (R2 = 0.98 (EDH); 0.70 (IPH)), hence better predictable by a joint correlation between hematoma volume, midline pixel- or voxel-shift, and minimum distance of (ideal or deformed) midline from the hematoma (boundary or centroid). CONCLUSION: All these predictors were computed automatically, which highlighted the excellent clinical potential of the proposed automated method in midline shift (MLS) estimation and severity prediction in hematoma decision support systems.

Computer-assisted delineation of hematoma from CT volume using autoencoder and Chan Vese model.

PURPOSE: To reduce the inter- and intra- rater variability as well as time and effort, a method for computer-assisted delineation of hematoma is proposed. Delineation of hematoma is done for further automated analysis such as the volume of hematoma, anatomical location of hematoma, etc. for proper surgical planning. METHODS: Fuzzy-based intensifier was used as a pre-processing technique for enhancing the computed tomography (CT) volume. Autoencoder was trained to detect the CT slices with hematoma for initialization. Then active contour Chan-Vese model was used for automated delineation of hematoma from CT volume. RESULTS: The proposed algorithm was tested on 48 hemorrhagic patients. Two radiologists have independently segmented the hematoma manually from CT volume. The intersection of two volumes was used as ground-truth for comparison with the segmentation performed by the proposed method. The accuracy was determined by using similarity matrices. The result of sensitivity, positive predictive value, Jaccard index and Dice similarity index were calculated as 0.71 ± 0.12, 0.73 ± 0.18, 0.55 ± 0.14, and 0.70 ± 0.12 respectively. CONCLUSIONS: A new approach for delineation of hematoma is proposed. The algorithm works well with the whole volume. Similarity indices of the proposed method are comparable with the existing state of art.

Computer-assisted delineation of cerebral infarct from diffusion-weighted MRI using Gaussian mixture model.

PURPOSE: Diffusion-weighted imaging (DWI) is a widely used medical imaging modality for diagnosis and monitoring of cerebral stroke. The identification of exact location of stroke lesion helps in perceiving its characteristics, an essential part of diagnosis and treatment planning. This task is challenging due to the typical shape of the stroke lesion. This paper proposes an efficient method for computer-aided delineation of stroke lesions from DWI images. METHOD: Proposed methodology comprises of three steps. At the initial step, image contrast has been improved by applying fuzzy intensifier leading to the better visual quality of the stroke lesion. In the following step, a two-class (stroke lesion area vs. non-stroke lesion area) segmentation technique based on Gaussian mixture model has been designed for the localization of stroke lesion. To eliminate the artifacts which would appear during segmentation process, a binary morphological post-processing through area operator has been defined for exact delineation of the lesion area. RESULT: The performance of the proposed methodology has been compared with the manually delineated images (ground truth) obtained from different experts, individually. Quantitative evaluation with respect to various performance measures (such as dice coefficient, Jaccard score, and correlation coefficient) shows the efficient performance of the proposed technique.