The ventral intermediate nucleus (Vim), as the motor thalamic nuclei section, is a generally used target in brain lesion surgery or stimulation for decreasing tremors in people with Parkinson’s disease. Determining the exact position of Vim is challenging because Vim cannot be visualized clearly in commonly used magnetic resonance imaging (MR). Indirect methods, i.e., Coordinate-based targeting and Guiot’s, utilize anterior commissure and posterior commissure to detect the location of Vim. In practice, neurosurgeons manually implement these methods in existing neurosurgical planning software, so the accuracy of the targeting depends on their memory and foresight. Afterward, Coordinate-based targeting and Guiot’s locate Vim based on anterior commissure (AC) and posterior commissure (PC), so neurosurgeons must correctly determine AC and PC. This paper proposes automatic indirect methods and measures the accuracy of indirect methods in MRIs with correct and incorrect orientations of AC-PC planes. An objective of analyzing indirect methods in MRIs with incorrect orientations of AC-PC planes is to discover the most resilient indirect method with inaccuracy AC-PC planes. To develop automatic indirect methods, the first step is redefining the plane passing through three defined points, i.e., AC, PC, and midline reference, by a quaternion. Secondly, Coordinate-based targeting and Guiot’s are implemented to determine the Vim targeting location automatically. This paper converts the rules of those methods in voxels because the rules use millimeters while the three-dimensional MRIs use voxels. The experiment shows that Vim locations obtained by Guiot’s are more accurate than those by Coordinate-based targeting in MRIs with the correct orientation of AC-PC planes. Guiot’s has 0.05 mm smaller value of average error results than Coordinate-based targeting. In contrast, Vim locations based on Coordinate-based targeting are more accurate in the MRIs with the incorrect orientation of AC-PC planes. Coordinate-based targeting has 0.032 mm smaller value of average error results than Guiot’s.