Abstract
To analyze multimodal three-dimensional medical images, interpolation is required for resampling which—unavoidably—introduces an interpolation error. In this work we describe the interpolation method used for imaging and neuroimaging and we characterize the Gibbs effect occurring when using such methods. In the experimental section we consider three segmented three-dimensional images resampled with three different neuroimaging software tools for comparing undersampling and oversampling strategies and to identify where the oversampling error lies. The experimental results indicate that undersampling to the lowest image size is advantageous in terms of mean value per segment errors and that the oversampling error is larger where the gradient is steeper, showing a Gibbs effect.
Highlights
In the context of multimodal medical imaging [1,2,3], image data of the same physical body are obtained from different imaging systems
A functional image is typically obtained by Single Photon Emission Computed Tomography (SPECT), Positron Emission Tomography (PET) [3], functional Magnetic resonance Imaging (MRI) [5] or by emerging systems as Magnetic Particle Imaging (MPI) [6]
These data indicate that the interpolation error occurring in oversampling is a Gibbs effect, confirming the results proven in the theoretical section
Summary
In the context of multimodal medical imaging [1,2,3], image data of the same physical body are obtained from different imaging systems. A functional image is typically obtained by Single Photon Emission Computed Tomography (SPECT), Positron Emission Tomography (PET) [3], functional MRI (fMRI) [5] or by emerging systems as Magnetic Particle Imaging (MPI) [6] Thanks to their high spatial resolution, morphological images can be used for identifying different structures of the physical body under examination through segmentation. Despite what common sense may suggest, the latter is preferable, due to the bigger interpolation errors occurring in oversampling This represents a paradox (we will later refer to this effect as the resampling paradox), and a big waste of time since an accurate segmentation image at high resolution (in the case of the human brain) is obtained typically after 10–20 h of manual work or 3–10 h of automatic segmentation pipelines [9]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
