MRI images are extremely rich in anatomical details and MR is therefore the most exact imaging modality when
it comes to visualizing the human body. But this feature also makes segmentation procedures extremely time
consuming. This is because standard thresholding or maximum intensity projections (MIP) methods only work for
a few MR sequences.
But with the new dawn of Artificial Intelligence (AI), what has not been practical in an everyday clinical
setting we at NordicCAD prove is achievable by applying convolutional neural networks (CNN) to your own MR
images.
By using the software annotation tool you can either remove false positive elements or add parts of structures
that are missing and then save the corrected segmentation. Also, you have the possibility to draw or mark
slices with the same tool, save the changes as a new label and study the label in 3D.
Finally, with segmented data the exciting possibilities of 3D anatomical printing and virtual or augmented
reality (VR/AR) visualization lie readily available.