Skip to main content
Fig. 4 | BMC Medical Imaging

Fig. 4

From: A deep learning framework for automated detection and quantitative assessment of liver trauma

Fig. 4

Liver trauma and organ segmentation results as well as ground truth annotations in patients with trauma or pre-existing conditions. (a) Axial contrast-enhanced CT image shows an intraparenchymal hematoma in the left liver lobe. (b) Axial contrast-enhanced CT image shows diffuse low attenuation of the right liver lobe relative to the spleen, consistent with fat deposition. This is a non-traumatic pre-existing condition. (c) Axial contrast-enhanced CT image shows heterogeneous enhancement of the right liver dome due to congestive hepatopathy, a non-traumatic pre-existing condition. In (a), (b), and (c), the first column corresponds to the original CT image; the second column corresponds to the ground truth annotations; and the third column corresponds to the automated segmentation results before post-processing. In both ground truth annotations and segmentation results, the green line shows the liver contour while the red line marks the contour of trauma regions. The fourth column compares the pixel intensity distribution inside the segmented liver and the segmented trauma region. As the -test indicated the difference between the two means of the aforementioned distributions was small for both examples of (b) and (c), the corresponding segmented injured regions were excluded during the post-processing step

Back to article page