Hybrid µCT-FMT imaging and image analysis

Fluorescence mediated tomography is a highly sensitive imaging technique to quantitatively assess the fluorescence distribution. In anesthetized mice, many targeted fluorescent probes are available, which allow imaging of angiogenesis, apoptosis, inflammation, and others. In this movie, we will show how hybrid imaging with FMT and Micro City is performed at our institute.

In chapter one, the devices, this is our F-M-T-F-M-T stands for fluorescence molecular tomography. Tomography means that 3D images are generated. The FMT is highly sensitive for imaging near inate fluorescence in mice.

Now the front panel is opened to show the inside of the FMT. At the bottom is a laser mounted on a 2D server. The laser emits light into the mouse, which is held in a partially transparent mouse bed.

Above the mouse bed is an array of LEDs as an alternative light source, which allows the FMT to operate as a regular reflectance imaging device. The filter wheel is mounted below the lens, which captures the light for the detector on top of the device. This is our micro ct.

The micro CT is equipped with two x-ray tubes and two flat panel detectors, which allow acquisition of dual energy scans. Chapter two, micro CT FMT scanning protocol. Both FMT and micro CT are built to image mice before imaging mice are anesthetized with isof fluorine.

For the FMT scanning, the hair of the mouse needs to be removed, which works well with the herring cream. Some strains of mice can develop rashes from the deha cream. Therefore, it is recommended to monitor the mice for skin changes and to contact the veterinary staff for care if needed.

Also, test the tolerance on a small batch of any new strains of mice to start with. To avoid hypothermia, the mouse is placed on a heating pad. To inject contrast agent, A catheter is placed into the tail vein due to the small size.

This is quite difficult and requires some experience. If blood flows back into the catheter, it's placed properly through the catheter. Fluorescent contrast agents and CT contrast agents can be injected to avoid volume overloading.

At most. Five milliliter per kilogram body weight should be injected, which means 150 microliter for a 30 gram mouse. For scanning, the mouse is placed inside a multimodal mouse bed.

The mouse may have some symbols drawn onto the tail for identification. It is important to avoid such things on the torso as it may impact the optical scanning. The mouse bed is closed and the depth adjusted to tightly hold the mouse in a fixed position.

Be careful and monitor the breathing cause tightening the bed too much can suffocate the mouse. Nude mice are commonly used for oncologic research due to their immune suppression. The fact that they're nude is a fortunate side effect of the genetic mutation.

Therefore, the laborious hair removal procedure can be omitted. It should be checked that the mouse breathes properly, and if necessary, the mouse bed should be adjusted accordingly. Then the mouse bed is placed inside the micro ct.

The tubes which transport the ISO fluorine gas are switched to maintain the gas flow inside the device. Then the micro C is closed to enable the X-ray shielding. The micro C will only start scanning if the lid is closed.

Using the buttons on the micro ct, the mouse can be driven into the micro CT.At the micro city control computer, a topo gram is acquired and shown the windowing settings are adjusted to better see the mouse. One or more ZUP scans can be placed. Their position is indicated by the light blue regions.

Usually one to three ZUP scans are sufficient. After starting the scanning, the progress is shown with dark blue progress bars. Our flat panel micro CT scans ZUP scans subsequently, which is different from a clinical spiral ct.

Using the buttons, the mouse bed is moved to the front. Again, the shielding lid is opened and the anesthetic tubes are switched. The holder is carefully removed from the mouse bed and the anesthetic tube is pulled out.

This is necessary because the AN and the FMT does not depend on this small tube. Instead, the small chamber inside the FMT is flooded with anesthetic gas. Now the mouse bed with the mouse is brought over to the FMT and inserted on the FMT control computer.

The scanning field of view is adjusted as well as the sampling density. Usually around 120 points are used. The scan starts after pressing a button.

The first pass of the FMT acquires a trans illumination or excitation image for each laser source point. In this movie, it is shown in fast forward mode. Usually it takes around five minutes.

You can see that much less light passes through the upper body compared to the lower body. This is because the organs with higher relative blood volume, such as heart, liver, and kidneys are more in the upper body. Blood is the main absorber for near infrared light.

The second pass runs of the same source points with a different filter, which only lets the fluorescent light pass through Chapter three, interactive oil and segmentation. To fuse the data from both devices. Markers are used, which are built into the mouse bed.

The markers are also visible in a reflection image acquired by the FMT. The markers are actually simple holes and do not need to be filled with any fluorescence or CT contrast agent. At our institute, we developed a software program that performs the marker detection and fusion automatically.

The shape of the mouse, as well as heterogeneous absorption and scattering maps are automatically estimated using the micro CT data as described in our recent Theranostics publication. These parameters are important for quantitative fluorescence reconstruction. To measure the biodistribution of the fluorescence, an organ segmentation is needed.

We generate such a segmentation interactively using a software named imulitic preclinical developed at our institute in the following. Such a segmentation is shown in fast forward and experienced person can perform it in around 10 to 20 minutes. First, the CT data set is loaded.

It can be inspected in 3D using an ISO surface rendering. By changing the windowing settings, the ISO value can be changed. For example, to visualize bones of the entire mouse body with the mouse bed, then the overlay is loaded.

The signal for this example appears in the bladder. Now the overlay visualization is turned off. To concentrate on the anatomical segmentation.

In native microt scans, the lung is easily found due to the strong negative contrast to the other soft tissue. The big structure inside the lung is the heart. Let's segment the lung first.

All voxels below a certain value are segmented using thresholding. This appears as green. The lung is a connected region, which can be separated using a filling operation.

Similar to the paint bucket. In a painting program, the can be separated from the lung by cutting and filling. Convex organs, such as the bladder, can be segmented by drawing scribbles to delineate the borders of the organ.

More scribbles be added until a sufficient accuracy is reached. Non convex organs, such as the intestine, can be segmented piece by piece. The liver appears as a homogeneous region and has a more complex structure.

Since it consists of multiple lobes, the segmentation can be saved to disk and loaded into the program. Using such a segmentation, the fluorescent signal can be assigned to the organs. The program computes these amounts and saves them as an extra sheet for longitudinal scans, the segmentation needs to be performed for each time point again, because the intervals are usually too long to keep the mouse anesthetized in a fixed position.

Therefore, the segmentation is a laborious task if many mice and many time points are involved. To quantify the results, load an overlay and a segmentation. Click set Batch settings to let the program remember the current settings.

Now, click batch statistics to tell the program to compute the values for all regions in all micro CT FMT scans. This will take a few seconds. Then the statistics are saved in one spreadsheet file.

This is convenient cause the user does not have to merge dozens of files himself. Based on this file, the organ curves can be computed. Chapter four, representative results.

To test that the fusion works properly, we used an aros phantom. Some titanium exide powder was added for scattering. To realize an irregular shape, we cut away some parts.

Several small inclusions filled with fluorescence and CT contrast agent were built into the phantom. Since the FMT does not know the true shape of the object and assumes a simplified shape, the reconstruction is not accurate for objects with irregular shapes. Therefore, we implemented another reconstruction that uses the shape derived from the micro city data.

As you can see, the signal localization in the phantom is much better. To inspect the in vivo data, let's go through the imaging time points. This is the pre-scan.

What we see is basically just noise and artifacts. If we go to the next time point, the one after injection, we see much more signal, but the windowing settings are much too hard. Using the windowing dialogue, you can adjust the windowing settings.

We see most of the signal in the bladder. If we go to the next time, 0.2 hours after injection, we see some signal outside the mouse. This is because the mouse urinated the fluorescence onto the mouse bed.

By further adjusting the windowing, we can see some signal at the spine and knees. Now let's go to time 0.4 hours after injection, the signal from urine is gone now and we see fluorescence at the spine and the knees for the next time points six hours and 24 hours after injection. We see the same.

Now let's go over to the next nose. The pre-scan shows nothing using these windowing settings. The scan 15 minutes after injection shows strong bladder signal and so on.

In this study, we found high concentrations in the urinary bladder shortly after injection, as a consequence of the fast renal excretion of this probe. Furthermore, the signal in the spine quickly rises and remains relatively stable throughout the later imaging time points. Chapter five, conclusion.

In conclusion, we show a multimodal imaging protocol to combine the strengths of fluorescence, molecular tomography and microcomputer tomography. The anatomical micro CT data enables an improved fluorescence reconstruction by using the shape of the mouse. Furthermore, it is useful to generate organ segmentations, which are needed to extract quantitative measurements from the image data.