Multimodlal Apparent Diffusion and Fuzzy C Means

Introduction

This site provides the ImageJ plugin(s) used in the papers Multimodal apparent diffusion (MAD) weighted magnetic resonance imaging, Magn Reson Imaging, 2020 Dec 10;77:213-233. doi:10.1016/j.mri.2020.12.007. (the MAD paper). The fuzzy MAD stroke conjecture, using Fuzzy C Means to classify multimodal apparent diffusion for ischemic stroke lesion stratification, Magn Reson Imaging. 2025 Apr:117:110294. doi: 10.1016/j.mri.2024.110294 (the FuzzyMAD paper). Note that this was only tested in ImageJ1, e.g., not ImageJ2 or FIJI.

There are two ZIP files provided, the first was used in the production of the MAD paper, and the second was used in the production of the FuzzyMAD paper. The first ZIP file contains two plugins provided to process a multi b-value diffusion dataset, the first one to open the dataset and the second to process and intergate the resuls. A plugin called DICOM_open is provided to open a series of DICOM files containing multi b-value images, where each frame is a different b-value and is arranged in monotonically increasing order. Any other input method can be used as long as it places the data in this format. The second plugin called Diffusion_MultiModal processes the hyperstack and produces and displays the parameter images, and, provides the means to draw Roi(s) and evaluate the processing. The second ZIP file contains the aformentioned plugins, with the additions of the Fuzzy_C_Means plugin, and supporting plugins.

Installation

DICOM_open

Plugin to open a series of DICOM files containing multi b-value diffusion weighted images.

Diffusion_MultiModal

Plugin to process a hyperstack of diffusion weighted images. Frames need to be ordered in monotonically increasing b-values starting with b=0.

F_Project

Plugin to process hyperstack frames, i.e., mean, etc. F_Project is used by Diffusion_MultiModal.

F_Profiler

Plugin interactively plots an Roi in the frame direction of a hyperstack. Useful in quickly verifying your diffusion dataset.

Frame_Slider

Plugin used to interactively in unison slide through the frames.

Fuzzy C Means

Plugin to interactively interogate the MAD parameters and use Fuzzy C Means to clusterize the MAD parameters into clusters and display these clusters.

Colorizer_Overlay

Plugin to produce colorized images, with color bars, of quantified images, e.g., cluster maps.

Images_Calculator

Plugin to process a series of images, e.g. sumation.

The_Combiner

Plugin to to combine a set of images into a matrix of images, e.g., a montage.

Walk Through

Open the multi b-value diffusion weighted dataset.

DICOM_open will appear either in File/Import/DICOM find/open or in Plugins/DICOM/DICOM_open. DICOM_open will start by asking for the root directory or your DICOM files. Assuming the directory structure is subject/study/series it will provide you with the information for selection of a dataset. You must indicate that this is a diffusion dataset.

Run Diffusion_MultiModal.

Diffusion_MultiModal will appear in Plugins. With the multi b-value diffusion dataset as the current image, run Diffusio_MultiModal and select OK. N.B., some versions of ImageJ did not properly process Short dataset, most likely this does not apply to you. The resulting parameter images will appear arranged to the left of the multi b-value diffusion dataset window.

Interogate the parameter images

If you draw Rois within the multi b-value diffusion dataset window the graphs of the processing of the Roi will appear to the right of the multi b-value dissuion dataset window.

With MAD parameter datasets open, run Fuzzy_C_Means

It is assumed that you will be running Fuzzy_C_Means repeatedly. First time running, on the initial dialog, specify the MAD parameter images. After identifying some clusters, create a Centers snapshot. Thereafter on the initial dialog, specify the Centers file that was previously saved. The plugin was developed from the perspective of interactive interagation of the source data (MAD parameters) and recognize clusters; the best way to learn is to play with it.