Shannon entropy as a stable diffusion tensor parameter for evaluating normal spinal cord regions in dogs

Diffusion tensor imaging (DTI) has been used in veterinary medicine for cerebral infarctions and degenerative diseases in recent years. Defining normal values for the apparent diffusion coefficient (ADC) and fractional anisotropy (FA) to detect lesions is challenging because these values depend on anatomical differences in the spinal cord, magnetic resonance imaging (MRI) machine model, magnetic field strength, and measurement protocol. Shannon entropy is the average amount of information and is a measure of the variability or complexity, of distributed numbers. The purpose of this study is to evaluate whether Shannon entropy can serve as a useful new indicator for interpreting the structural integrity of the canine spinal cord. We used six beagle dogs with no spinal cord disease, as confirmed by neurological examination and magnetic resonance imaging. We calculated the respective values for the cervical and thoracolumbar spinal cord under the 3 T magnetic field. The ADC and FA values (mean ± standard error) of the cervical spinal cord were 1.11 ± 0.05 × 10^-3 mm²/s and 0.64 ± 0.02. The thoracolumbar spinal cord had values of 1.16 ± 0.04 × 10^-3 mm²/s and 0.58 ± 0.02. The ADC and FA values of the spinal cord reflected anatomical differences, and no fixed normal values were observed for the entire spinal cord. Shannon entropy was 5.67 ± 0.45. No significant differences were observed in the values at any intervertebral level. These results suggest that Shannon entropy is more useful than the ADC and FA values. Shannon entropy may be able to differentiate between normal tissue and lesions, independent of various MRI conditions.