MNI SISCOM: an Open-Source Tool for Computing Subtraction Ictal Single-Photon Emission CT Coregistered to MRI

Associated Data

Data Availability Statement

All code is available for download at https://github.com/jeremymoreau/mnisiscom.

Abstract

Subtraction ictal single-photon emission computed tomography (SPECT) coregistered to MRI (SISCOM) is a well-established technique for quantitative analysis of ictal vs interictal SPECT images that can contribute to the identification of the seizure onset zone in patients with drug‐resistant epilepsy. However, there is presently a lack of user-friendly free and open-source software to compute SISCOM results from raw SPECT and MRI images. We aimed to develop a simple graphical desktop application for computing SISCOM. MNI SISCOM is a new free and open-source software application for computing SISCOM and producing practical MRI/SPECT/SISCOM image panels for review and reporting. The graphical interface allows any user to quickly and easily obtain SISCOM images with minimal user interaction. Additionally, MNI SISCOM provides command line and Python interfaces for users who would like to integrate these features into their own scripts and pipelines. MNI SISCOM is freely available for download from: https://github.com/jeremymoreau/mnisiscom.

Keywords:

Epilepsy, SISCOM, SPECT, MRI, Software , Open-source, Python, Nuclear Imaging, Functional Neurosurgery

Background

Subtraction ictal single-photon emission CT coregistered to MRI (SISCOM) [1, 2] is a widely used and well-established technique for quantitative analysis of ictal vs interictal SPECT images. The technique consists of computing difference images between coregistered and standardized (“z-scored”) SPECT scans captured in the interictal and ictal phases. The technique highlights areas of hyperperfusion in the ictal as compared to the interictal scan, which has been demonstrated to be valuable in helping localize the seizure onset zone [3–6]. In a prospective study evaluating SISCOM in patients with either nonlesional MRIs or discordant data in the presurgical evaluation (e.g., discordant EEG and MRI), SISCOM was found to be concordant with the surgical resection in 82% of patients and 22/26 patients with post-surgical follow-up achieved Engel class I (15) or class II (7) outcomes [3]. In one recent meta-analysis [2], concordance between SISCOM and the surgical resection as compared to nonconcordant SISCOM was associated with a 3.28 times higher seizure-free odds ratio for temporal cases (2.44 for extra-temporal cases). Timing of the tracer injection however remains a critical determinant of the localization sensitivity of SISCOM [2, 4].

The conceptual basis of SISCOM is relatively simple, but there is presently a lack of user-friendly free and open-source software to compute SISCOM results from raw SPECT and MRI images. General-purpose neuroimaging data analysis packages such as SPM [7] already provide tools (e.g., coregistration and image calculators) that allow for the computation of SISCOM results [8], but obtaining these results typically requires several time-consuming manual steps and necessitates a certain level of technical expertise. There have been previous efforts to design purpose-built software for computing SISCOM [9], but we could not find any open source software program that is actively maintained and runs on current versions of Mac/Windows/Linux operating systems. Here, we present a newly developed cross-platform and open-source application to facilitate the process of computing SISCOM images. The goal of this project is to provide a freely available single-purpose and user-friendly tool to implement SISCOM.

Methods

The MNI SISCOM desktop application (Fig.  ) runs on Windows, Mac, and Linux computers and can be downloaded here: https://github.com/jeremymoreau/mnisiscom. Detailed installation instructions are provided on the download page linked above. In addition to MNI SISCOM, the SPM software package [7] must also be installed. SPM is a popular general-purpose brain imaging data analysis program and is used by MNI SISCOM for SPECT and MRI image coregistration (i.e., aligning the SPECT images to the T1 MRI) and normalization (warping MRI and SPECT images into standard MNI coordinate space).

Open in a separate window

Usage of the desktop application is very straightforward. Simply launch the app and, after setting the SPM installation path in the “Settings” menu, select the T1 MRI, interictal SPECT, ictal SPECT, and folder where results will be saved. The other options do not generally require tweaking, but detailed explanations of each option can be viewed by hovering over the option label in the app. Once the “Compute” button is clicked, MNI SISCOM will take ~ 2–5 min to compute the SISCOM results, depending on the speed of the computer. Of note, currently only MRI and SPECT images in NIfTI format (https://nifti.nimh.nih.gov) are supported. If exporting original DICOM images from PACS, we recommend using MRIcron [10] to convert DICOM images to NIfTI. MRIcron can be downloaded here: https://www.nitrc.org/projects/mricron. In MRIcron click on the “Import” menu and select “Convert DICOM to NIfTI.”

In addition to the desktop application, a command line interface and a user scriptable library written in the Python programming language are also made available for more technically inclined users. These interfaces allow for the integration of MNI SISCOM into software pipelines or tools developed by others. An example Python script is provided to illustrate how these tools can be used to run MNI SISCOM on a group of patient MRI and interictal/ictal SPECT images without any user intervention (https://github.com/jeremymoreau/mnisiscom/tree/master/examples). The command line tool and Python library can easily be installed via the standard Python Package Index (PyPI): https://pypi.org/project/mnisiscom/.

Results

MNI SISCOM outputs scrollable 3D volumes of SISCOM results in NIfTI format, but also produces convenient image panel slides for rapid review and inclusion in presentations/reports. These panel slides include large panels showing interictal/ictal SPECT and SISCOM results side-by-side (Fig.  ) as well as a series of compact panels in axial, coronal, and sagittal orientation showing only interictal/ictal SPECT or SISCOM results (Fig.  a). Moreover, MNI SISCOM can optionally output, using the bundled Nilearn module [11], schematic maximum intensity projection (“glass brain”) images showing thresholded SISCOM maps superimposed over an anatomical reference drawing (Fig.  b). For group studies, MNI SISCOM also provides the option to produce 3D NIfTI volumes in the standard MNI coordinate space [12, 13], which can then be used in SPM [7] or other neuroimaging software packages to perform statistical comparisons between groups of patients. The generated NIfTI volumes can be visualized using any of the many popular open source image viewers such as MRIcron (https://www.nitrc.org/projects/mricron) and Mango (https://www.nitrc.org/projects/mango). A concise description of each result file generated by MNI SISCOM is presented in Table .

Open in a separate window

Open in a separate window

Table 1

FilenameDescriptionsiscom_z.nii.gzZ-scored SISCOM results volume coregistered to the input T1 MRI (NIfTI)siscom_z_MNI152.nii.gzZ-scored SISCOM results volume in standard MNI coordinate space (NIfTI)interictal_z.nii.gzZ-scored interictal SPECT volume coregistered to the input T1 MRI (NIfTI)ictal_z.nii.gzZ-scored ictal SPECT volume coregistered to the input T1 MRI (NIfTI)interictal_mask.nii.gzMask generated from interictal SPECT (NIfTI)interictal_coregistered.nii.gzOriginal interictal SPECT volume coregistered to the input T1 MRI (NIfTI)ictal_coregistered.nii.gzOriginal ictal SPECT volume coregistered to the input T1 MRI (NIfTI)ICTAL-{ax/cor/sag}_mri_slide.pngImages showing axial, coronal, and sagittal slices of the ictal SPECT resultsINTERICTAL-{ax/cor/sag}_mri_slide.pngImages showing axial, coronal, and sagittal slices of the interictal SPECT resultsSISCOM-{ax/cor/sag}_mri_slide.pngImages showing axial, coronal, and sagittal slices of the SISCOM results (Fig.  a)SISCOM-{ax/cor/sag}_mri_panel.pngImages showing side-by-side axial, coronal, and sagittal slices comparing interictal, ictal, and SISCOM results (Fig.  )SISCOM-glass_brain.pngSchematic maximum intensity projection (“glass brain”) images showing thresholded SISCOM maps superimposed over an anatomical reference drawing (Fig.  b)Open in a separate window

Discussion

Computation of SISCOM results is possible using existing general-purpose brain imaging data analysis programs, but obtaining such results is often time-consuming and labor-intensive. Our aim was to eliminate all the steps usually involved in obtaining SISCOM with a simple and modern desktop application. MNI SISCOM greatly simplifies the process by providing nearly entirely automatic processing of SPECT and MRI images in order to generate SISCOM results with minimal user interaction. The application is also completely free and open-source, and we will continue contributing new features and usability improvements. We are also happy to receive suggested feature requests and help tailor the functionality of the application to the community’s needs. Limitations include currently only supporting the NIfTI file format and requiring another program to convert DICOMs to NIfTI as well as only supporting the classical SISCOM algorithm. Additionally, MNI SISCOM currently depends on SPM for image coregistration, which requires another program to be installed. Future steps will include implementation of additional algorithms such as STATISCOM, which has been demonstrated to be superior to the more commonly used classical SISCOM algorithm in many cases [14, 15]. We are also planning on building and bundling a database of interictal SPECT images in standard MNI space [12, 13] to allow for statistical comparison of individual patient ictal SPECT scans against the norm without the need for an interictal SPECT.

Conclusion

MNI SISCOM is a user-friendly free and open-source application for computing SISCOM. It provides a straightforward graphical desktop interface and helps minimize manual image manipulation tasks as compared to more multi-purpose brain imaging processing tools. Finally, the MRI/SPECT/SISCOM image panels generated by MNI SISCOM are a useful addition to help increase the efficiency of review and reporting of SISCOM results.

Author Contributions

JTM: conception and design, app development, figures, drafting the manuscript, critically revising the manuscript. CSM: supervision, critically revising the manuscript. SB: supervision, funding and administration, critically revising the manuscript. RWRD: conception and design, supervision, drafting the manuscript, funding and administration, critically revising the manuscript.

Code Availability

All code is available for download at https://github.com/jeremymoreau/mnisiscom.

Compliance with Ethical Standard

Conflict of Interest

The authors declare that they have no conflict of interest.

Ethical Approval

This study received full approval by the McGill University Health Centre’s Research Institute Ethics Board, and all involved patients and/or parents/guardians signed an informed consent form to be enrolled in the study.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.