Bioluminescence tomography system for in vivo irradiation guidance

Abstract

We constructed a bioluminescence tomography(BLT) to localize soft tissue targets for preclinical radiotherapy study. With the threshold and margin designed for target volume, BLT can provide opportunity to perform conformal irradiation to malignancy.

1. Introduction

Several groups, including ours, have developed small animal irradiators (SAI) that mimic clinical radiotherapy (RT). Cone-beam computed tomography (CBCT) is a major imaging modality to guide SAIs for irradiation, but is less adept at localizing soft tissue target that grows in a low image contrast environment. Bioluminescence imaging (BLI) offers a solution for soft tissue targeting. However, because optical transport is susceptible to irregular animal torsos and tissue optical properties, BLI is inadequate to localize internal source for precise irradiation guidance. This led us to innovate bioluminescence tomography (BLT) as an image guidance modality. In tandem with an optimization algorithm, BLT employs a light propagation model for reconstruction, which minimizes deviation between calculated and measured surface BLI, to recover underlying source distribution.

Our BLT system is designed in standalone and mobile mode with non-contact imaging geometry. It can be integrated with a SAI via a transportable mouse bed that allows the animal transferred between these two systems. The standalone design with the transportable bed minimizes the need of modifying existing commercial SAIs and ensures BLI-only applications without occupying the SAI. For our BLT-guided RT course, a mouse first undergoes BLI in the BLT system, and later CBCT imaging in SAI. Surface BLI are mapped to a numerical mesh established from the CBCT image and used as the inputs for BLT reconstruction to retrieve target distribution for SAI radiation guidance. Under well-defined threshold and radiation margin for the reconstructed volume, the uncertainties of target localization can be effectively minimized, which lead to precise volumetric-guided focal irradiation.

2. Materials and Methods

Figure 1 shows a schematic of the BLT system, which consists of an optical assembly, a transportable mouse bed, and a 1D linear stage. The optical assembly comprises a CCD and a light-tight enclosure. The front end of enclosure is an imaging chamber, including a motorized 3-mirror system with four LEDs mounted at its corners. The back end of enclosure includes a 50-mm f/1.2 lens installed at the CCD and a filter wheel. The 3-mirror system can rotate 180° around the imaged object. Four 20-nm band-pass filters at 590, 610, 630, and 650 nm were mounted in the filter wheel. For BLI acquisition, the optical assembly is driven by the 1D linear stage to dock onto the mouse bed, and the optical signal is directed as red dash line in Fig. 1. The transportable mouse bed is used for holding an imaged object in the BLT system and SAI, and supporting the object transport between these two systems. Eight ball bearings (BBs) were attached to the bed for registering the coordinates of the BLT system and SAI. The LEDs illuminate object and BBs to identify their positions. Inside the optical assembly, we installed a thermostatic system, used for controlling the temperature around the imaged object at 37 °C. A heat gun is first to boost the chamber temperature to 37 °C before imaging. A resistor, a thermocouple, and seven fans are assembled inside imaging chamber for maintaining temperature during the imaging course.

Fig. 1.

Schematic of the BLT system in mobile design; The red dash line indicates the optical path, which is 45 cm, from the focal plane to the front surface of the lens.

An orthotopic glioblastoma (GBM)-bearing mouse underwent multi-projection at 0°, 90°, and −90° and multi-spectral at 610, 630, and 650 nm BLI acquisition with pixel scale of 0.936 mm/pixel. The mouse was anesthetized and immobilized on the bed at prone position. After acquiring BLI, photos at −90°, −45°, 0°, 45°, and 90° projections were taken for the BB positions for the coordinate registration between the BLT and SAI. We proposed a geometry calibration method to map the 2D BLIs, which were first corrected with cell spectrum and time-resolved variation [1, 2], onto the mesh surface generated from the SAI CBCT image [3]. The mapped BLIs were then used as the input for BLT reconstruction. Diffusion approximation was applied to model light propagation in tissue. We used spectral derivative of the BLIs for BLT reconstruction to bypass free spacing light propagation modeling [4]. The tumor distribution (${\text{GTV}}_{\text{BLT}}$) is iteratively solved by applying a compressive sensing optimization algorithm. We assessed optimal threshold for ${\text{GTV}}_{\text{BLT}}$ using Dice coefficient between gross target volume (GTV) delineated by contrast CBCT, taken as ground truth, and the ${\text{GTV}}_{\text{BLT}}$, as $2\left({\text{GTV}}_{\text{BLT}}\cap \text{GTV}\right)/\left({\text{GTV}}_{\text{BLT}}+\text{GTV}\right)$. A uniform margin that accounts for the uncertainties of BLT target positioning and volume delineation was added to ${\text{GTV}}_{\text{BLT}}$ to form a planning target volume (${\text{PTV}}_{\text{BLT}}$) for radiation guidance. The margin size is determined by achieving full coverage to GTV and reducing normal tissue (NT) radiation toxicity. The coverage to GTV and NT is defined as $\left({\text{PTV}}_{\text{BLT}}\cap \text{GTV}\right)/\text{GTV}$ and $\left({\text{PTV}}_{\text{BLT}}-{\text{PTV}}_{\text{BLT}}\cap \text{GTV}\right)/{\text{V}}_{\text{head}}$, respectively, where ${\text{V}}_{\text{head}}$ is the volume of mouse head. The dosimetric goal for tumor coverage is 5 Gy prescribed dose covering 95% of the ${\text{PTV}}_{\text{BLT}}$.

3. Results

Figure 2 shows an example of applying BLT to guide irradiation for GBM in vivo. With 0.3 threshold, which provided the highest Dice coefficient, the ${\text{GTV}}_{\text{BLT}}$ is overlapped with the GTV at 0.16 mm center of mass deviation. Treatment margin of 0.5 mm was added on ${\text{GTV}}_{\text{BLT}}$ to form the ${\text{PTV}}_{\text{BLT}}$ and increase the GTV coverage from 84.7% to 100% with < 1% NT inclusion. A 12 non-coplanar radiation beams are used to illustrate that with the availability of ${\text{PTV}}_{\text{BLT}}$, we are able to guide irradiation with highly conformal plan with 99.1% of GTV/tumor coverage by the 5 Gy prescribed dose.

Fig. 2.

(a-b) Two views of CBCT image overlapped with GTV (green contour), GTV_BLT (heat map), and PTV_BLT (purple contour) with 0.3 threshold and 0.5-mm margin for an in vivo GBM case; isodoses of a 12 hybrid-beam plans were displayed.

4. Conclusions

We have developed a novel BLT-guided system readily compatible with commercial SAI to provide focal irradiation guidance. Our BLT platform can offer an attractive option of high contrast optical-image-guided irradiation for researchers in radiobiology to mimic clinical RT for orthotopic and spontaneous models. We expect the BLT-guided system enabling conformal irradiation for soft tissue targets, reducing normal tissue involvement and thus facilitating experimental reproducibility.