Optical advances in skeletal imaging applied to bone metastases

Prediction of major torso organs in low-contrast micro-CT images of mice using a two-stage deeply supervised fully convolutional network

Delineation of major torso organs is a key step of mouse micro-CT image analysis. This task is challenging due to low soft tissue contrast and high image noise, therefore anatomical prior knowledge is needed for accurate prediction of organ regions. In this work, we develop a deeply supervised fully convolutional network which uses the organ anatomy prior learned from independently acquired contrast-enhanced micro-CT images to assist the segmentation of non-enhanced images. The network is designed with a two-stage workflow which firstly predicts the rough regions of multiple organs and then refines the accuracy of each organ in local regions. The network is trained and evaluated with 40 mouse micro-CT images. The volumetric prediction accuracy (Dice score) varies from 0.57 for the spleen to 0.95 for the heart. Compared to a conventional atlas registration method, our method dramatically improves the Dice of the abdominal organs by 18%-26%. Moreover, the incorporation of anatomical prior leads to more accurate results for small-sized low-contrast organs (e.g. the spleen and kidneys). We also find that the localized stage of the network has better accuracy than the global stage, indicating that localized single organ prediction is more accurate than global multiple organ prediction. With this work, the accuracy and efficiency of mouse micro-CT image analysis are greatly improved and the need for using contrast agent and high x-ray dose is potentially reduced.

Imaging of bone metastasis: An update

Early detection of skeletal metastasis is critical for accurate staging and optimal treatment. This paper briefly reviews our current understanding of the biological mechanisms through which tumours metastasise to bone and describes the available imaging methods to diagnose bone metastasis and monitor response to treatment. Among the various imaging modalities currently available for imaging skeletal metastasis, hybrid techniques which fuse morphological and functional data are the most sensitive and specific, and positron emission tomography (PET)/computed tomography and PET/magnetic resonance imaging will almost certainly continue to evolve and become increasingly important in this regard.

Multimodal imaging of bone metastases: From preclinical to clinical applications

Metastases to the skeletal system are commonly observed in cancer patients, highly affecting the patients' quality of life. Imaging plays a major role in detection, follow-up, and molecular characterisation of metastatic disease. Thus, imaging techniques have been optimised and combined in a multimodal and multiparametric manner for assessment of complementary aspects in osseous metastases. This review summarises both application of the most relevant imaging techniques for bone metastasis in preclinical models and the clinical setting.

Bone metastasis imaging with SPECT/CT/MRI: a preclinical toolbox for therapy studies

Bone is one of the most common metastatic target sites in breast cancer, with more than 200 thousand new cases of invasive cancer diagnosed in the US alone in 2011. We set out to establish a multimodality imaging platform for bone metastases in small animals as a tool to non-invasively quantify metastasis growth, imaging the ensuing bone lesions and possibly the response to treatment. To this end, a mouse model of osteolytic metastatic bone tumors was characterized with SPECT/CT and MRI over time. A cell line capable of forming bone metastases, MDA-MB-231, was genetically modified to stably express the reporter gene herpes simplex virus-1 thymidine kinase (hsv-1 tk). The intracellular accumulation of the radiolabeled tracer [(123)I]FIAU promoted by HSV-1 TK specifically pinpoints the location of tumor cells which can be imaged in vivo by SPECT. First, a study using tumors implanted subcutaneously was performed. The SPECT/MRI overlays and the ex vivo γ-counting showed a linear correlation in terms of %ID/cm(3) (R(2)=0.93) and %ID/g (R(2)=0.77), respectively. Then, bone metastasis growth was imaged weekly by SPECT/CT and T2-weighted MRI over a maximum of 40 days post-intracardiac injection of tumor cells. The first activity spots detectable with SPECT, around day 20 post-cell injection, were smaller than 2mm(3) and not yet visible by MRI and increased in volume and in %ID over the weeks. Osteolytic bone lesions were visible by CT (in vivo) and μCT (ex vivo). The SPECT/MRI overlays also showed a linear correlation in terms of %ID/cm(3) (R(2)=0.86). In conclusion, a new multimodality imaging platform has been established that non-invasively combines images of active tumor areas (SPECT), tumor volume (MRI) and the corresponding bone lesions (CT and μCT). To our knowledge this is the first report where the combination of soft tissue information from MRI, bone lesions by CT, and reporter gene imaging by SPECT is used to non-invasively follow metastatic bone lesions.

Combined in vivo optical and µCT imaging to monitor infection, inflammation, and bone anatomy in an orthopaedic implant infection in mice

Multimodality imaging has emerged as a common technological approach used in both preclinical and clinical research. Advanced techniques that combine in vivo optical and μCT imaging allow the visualization of biological phenomena in an anatomical context. These imaging modalities may be especially useful to study conditions that impact bone. In particular, orthopaedic implant infections are an important problem in clinical orthopaedic surgery. These infections are difficult to treat because bacterial biofilms form on the foreign surgically implanted materials, leading to persistent inflammation, osteomyelitis and eventual osteolysis of the bone surrounding the implant, which ultimately results in implant loosening and failure. Here, a mouse model of an infected orthopaedic prosthetic implant was used that involved the surgical placement of a Kirschner-wire implant into an intramedullary canal in the femur in such a way that the end of the implant extended into the knee joint. In this model, LysEGFP mice, a mouse strain that has EGFP-fluorescent neutrophils, were employed in conjunction with a bioluminescent Staphylococcus aureus strain, which naturally emits light. The bacteria were inoculated into the knee joints of the mice prior to closing the surgical site. In vivo bioluminescent and fluorescent imaging was used to quantify the bacterial burden and neutrophil inflammatory response, respectively. In addition, μCT imaging was performed on the same mice so that the 3D location of the bioluminescent and fluorescent optical signals could be co-registered with the anatomical μCT images. To quantify the changes in the bone over time, the outer bone volume of the distal femurs were measured at specific time points using a semi-automated contour based segmentation process. Taken together, the combination of in vivo bioluminescent/fluorescent imaging with μCT imaging may be especially useful for the noninvasive monitoring of the infection, inflammatory response and anatomical changes in bone over time.

Interactive local super-resolution reconstruction of whole-body MRI mouse data: a pilot study with applications to bone and kidney metastases

In small animal imaging studies, when the locations of the micro-structures of interest are unknown a priori, there is a simultaneous need for full-body coverage and high resolution. In MRI, additional requirements to image contrast and acquisition time will often make it impossible to acquire such images directly. Recently, a resolution enhancing post-processing technique called super-resolution reconstruction (SRR) has been demonstrated to improve visualization and localization of micro-structures in small animal MRI by combining multiple low-resolution acquisitions. However, when the field-of-view is large relative to the desired voxel size, solving the SRR problem becomes very expensive, in terms of both memory requirements and computation time. In this paper we introduce a novel local approach to SRR that aims to overcome the computational problems and allow researchers to efficiently explore both global and local characteristics in whole-body small animal MRI. The method integrates state-of-the-art image processing techniques from the areas of articulated atlas-based segmentation, planar reformation, and SRR. A proof-of-concept is provided with two case studies involving CT, BLI, and MRI data of bone and kidney tumors in a mouse model. We show that local SRR-MRI is a computationally efficient complementary imaging modality for the precise characterization of tumor metastases, and that the method provides a feasible high-resolution alternative to conventional MRI.

Binding kinetics of a fluorescently labeled bisphosphonate as a tool for dynamic monitoring of bone mineral deposition in vivo

Bone mineral deposition during the modeling of new bone and remodeling of old bone can be perturbed by several pathological conditions, including osteoporosis and skeletal metastases. A site-specific marker depicting the dynamics of bone mineral deposition would provide insight into skeletal disease location and severity, and prove useful in evaluating the efficacy of pharmacological interventions. Fluorescent labels may combine advantages of both radioisotope imaging and detailed microscopic analyses. The purpose of this study was to determine if the fluorescent bisphosphonate OsteoSense could detect localized changes in bone mineral deposition in established mouse models of accelerated bone loss (ovariectomy) (OVX) and anabolic bone gain resulting from parathyroid hormone (PTH) treatment. We hypothesized that the early rate of binding, as well as the total amount of bisphosphonate, which binds over long periods of time, could be useful in evaluating changes in bone metabolism. Evaluation of the kinetic uptake of bisphosphonates revealed a significant reduction in both the rate constant and plateau binding after OVX, whereas treatment with PTH resulted in a 36-fold increase in the bisphosphonate binding rate constant compared with untreated OVX controls. Localization of bisphosphonate binding revealed initial binding at sites of ossification adjacent to the growth plate and, to a lesser extent, along more distal trabecular and cortical elements. Micro-computed tomography (CT) was used to confirm that initial bisphosphonate binding is localized to sites of low tissue mineral density, associated with new bone mineral deposition. Our results suggest monitoring binding kinetics based on fluorescently labeled bisphosphonates represents a highly sensitive, site-specific method for monitoring changes in bone mineral deposition with the potential for translation into human applications in osteoporosis and bone metastatic processes and their treatment.

Photothermal tomography for the functional and structural evaluation, and early mineral loss monitoring in bones

Salient features of a new non-ionizing bone diagnostics technique, truncated-correlation photothermal coherence tomography (TC-PCT), exhibiting optical-grade contrast and capable of resolving the trabecular network in three dimensions through the cortical region with and without a soft-tissue overlayer are presented. The absolute nature and early demineralization-detection capability of a marker called thermal wave occupation index, estimated using the proposed modality, have been established. Selective imaging of regions of a specific mineral density range has been demonstrated in a mouse femur. The method is maximum-permissible-exposure compatible. In a matrix of bone and soft-tissue a depth range of ~3.8 mm has been achieved, which can be increased through instrumental and modulation waveform optimization. Furthermore, photoacoustic microscopy, a comparable modality with TC-PCT, has been used to resolve the trabecular structure and for comparison with the photothermal tomography.

Longitudinal in vivo imaging of bone formation and resorption using fluorescence molecular tomography

Bone research often focuses on anatomical imaging of the bone microstructure, but in order to gain better understanding in how bone remodeling is modulated through interventions also bone formation and resorption processes should be investigated. With this in mind, the purpose of this study was to establish a longitudinal in vivo imaging approach of bone formation and resorption using fluorescence molecular tomography (FMT). In this study the reproducibility, accuracy and sensitivity of FMT for bone imaging were assessed by performing longitudinal measurements with FMT and comparing it to in vivo micro-computed tomography on a set of control mice, and mice in which load-adaptation was induced in the sixth caudal vertebra. The precision error for FMT measurements, expressed as coefficient of variation, was smaller than 16%, indicating acceptable reproducibility. A correlation was found between bone resorption measured with FMT and bone resorption rate measured with in vivo micro-computed tomography only over the first 14days (R=0.81, p<0.01), but not between bone formation measured with FMT and bone formation rate measured with in vivo micro-CT. Bone formation measured by FMT was 89-109% greater (p<0.05) for mice subjected to mechanical loading than control mice. Bone resorption was 5-8% lower, but did not reach a significant difference between groups, indicating moderate sensitivity for FMT. In conclusion, in vivo FMT in mouse tail bones is feasible but needs to be optimized for monitoring load adaptation in living mice.

Monitoring bacterial burden, inflammation and bone damage longitudinally using optical and μCT imaging in an orthopaedic implant infection in mice

BACKGROUND

Recent advances in non-invasive optical, radiographic and μCT imaging provide an opportunity to monitor biological processes longitudinally in an anatomical context. One particularly relevant application for combining these modalities is to study orthopaedic implant infections. These infections are characterized by the formation of persistent bacterial biofilms on the implanted materials, causing inflammation, periprosthetic osteolysis, osteomyelitis, and bone damage, resulting in implant loosening and failure.

METHODOLOGY/PRINCIPAL FINDINGS

An orthopaedic implant infection model was used in which a titanium Kirshner-wire was surgically placed in femurs of LysEGFP mice, which possess EGFP-fluorescent neutrophils, and a bioluminescent S. aureus strain (Xen29; 1×10(3) CFUs) was inoculated in the knee joint before closure. In vivo bioluminescent, fluorescent, X-ray and μCT imaging were performed on various postoperative days. The bacterial bioluminescent signals of the S. aureus-infected mice peaked on day 19, before decreasing to a basal level of light, which remained measurable for the entire 48 day experiment. Neutrophil EGFP-fluorescent signals of the S. aureus-infected mice were statistically greater than uninfected mice on days 2 and 5, but afterwards the signals for both groups approached background levels of detection. To visualize the three-dimensional location of the bacterial infection and neutrophil infiltration, a diffuse optical tomography reconstruction algorithm was used to co-register the bioluminescent and fluorescent signals with μCT images. To quantify the anatomical bone changes on the μCT images, the outer bone volume of the distal femurs were measured using a semi-automated contour based segmentation process. The outer bone volume increased through day 48, indicating that bone damage continued during the implant infection.

CONCLUSIONS/SIGNIFICANCE

Bioluminescent and fluorescent optical imaging was combined with X-ray and μCT imaging to provide noninvasive and longitudinal measurements of the dynamic changes in bacterial burden, neutrophil recruitment and bone damage in a mouse orthopaedic implant infection model.

Biomaterial delivery of morphogens to mimic the natural healing cascade in bone

Complications in treatment of large bone defects using bone grafting still remain. Our understanding of the endogenous bone regeneration cascade has inspired the exploration of a wide variety of growth factors (GFs) in an effort to mimic the natural signaling that controls bone healing. Biomaterial-based delivery of single exogenous GFs has shown therapeutic efficacy, and this likely relates to its ability to recruit and promote replication of cells involved in tissue development and the healing process. However, as the natural bone healing cascade involves the action of multiple factors, each acting in a specific spatiotemporal pattern, strategies aiming to mimic the critical aspects of this process will likely benefit from the usage of multiple therapeutic agents. This article reviews the current status of approaches to deliver single GFs, as well as ongoing efforts to develop sophisticated delivery platforms to deliver multiple lineage-directing morphogens (multiple GFs) during bone healing.

CT-based handling and analysis of preclinical multimodality imaging data of bone metastases

The pathogenesis of bone metastases is a complex and multifaceted process. Often multiple imaging modalities are needed to follow both the structural and functional changes over time during metastatic bone disease. Researchers face extended data sets of one experiment acquired with multiple modalities at multiple points in time. This review gives an overview of an integrated approach for handling these kinds of complex data. It focuses on the analysis of whole-body micro-computerized tomography and optical data handling. We show how researchers can generate side-by-side visualizations of scans taken with one imaging modality at multiple time points and with multiple modalities at one point. Moreover, we highlight methods for normalized volumes of interest selection and quantification of bone volume and thickness.

Preclinical imaging of the cellular and molecular events in the multistep process of bone metastasis

Bone metastasis is a complex process that ultimately leads to devastating metastatic bone disease. It is therefore of key interest to unravel the mechanisms underlying the multistep process of skeletal metastasis and cancer-induced bone disease, and to develop better treatment and management of patients with this devastating disease. Fortunately, novel technologies are rapidly emerging that allow real-time imaging of molecules, pathogenic processes, drug delivery and drug response in preclinical in vivo models. The outcome of these experimental studies will facilitate clinical cancer research by improving the detection of cancer cell invasion, metastasis and therapy response.

Preclinical models that illuminate the bone metastasis cascade

In this chapter currently available preclinical models of tumor progression and bone metastasis, including genetically engineered mice that develop primary and metastatic carcinomas and transplantable animal models, will be described. Understanding the multistep process of incurable bone metastasis is pivotal to the development of new therapeutic strategies. Novel technologies for imaging molecules or pathologic processes in cancers and their surrounding stroma have emerged rapidly and have greatly facilitated cancer research, in particular the cellular behavior of osteotropic tumors and their response to new and existing therapeutic agents. Optical imaging, in particular, has become an important tool in preclinical bone metastasis models, clinical trials and medical practice. Advances in experimental and clinical imaging will-in the long run-result in significant improvements in diagnosis, tumor localization, enhanced drug delivery and treatment.

In-vivo near-infrared optical imaging of growing osteosarcoma cell lesions xenografted in mice: dual-channel quantitative evaluation of volume and mineralization

BACKGROUND

In a previous study using a rodent osteosarcoma-grafted rat model, in which cell-dependent mineralization was previously demonstrated to proportionally increase with growth, we performed a quantitative analysis of mineral deposit formation using (99m)Tc-HMDP and found some weaknesses, such as longer acquisition time and narrower dynamic ranges (i.e. images easily saturated). The recently developed near-infrared (NIR) optical imaging technique is expected to non-invasively evaluate changes in living small animals in a quantitative manner.

PURPOSE

To test the feasibility of NIR imaging with a dual-channel system as a better alternative for bone scintigraphy by quantitatively evaluating mineralization along with the growth of osteosarcoma lesions in a mouse-xenograft model.

MATERIAL AND METHODS

The gross volume and mineralization of osteosarcoma lesions were evaluated in living mice simultaneously with dual-channels by NIR dye-labeled probes, 2-deoxyglucose (DG) and pamidronate (OS), respectively. To verify these quantitative data, retrieved osteosarcoma lesions were then subjected to ex-vivo imaging, weighing under wet conditions, microfocus-computed tomography (μCT) analysis, and histopathological examination.

RESULTS

Because of less scattering and no anatomical overlapping, as generally shown, specific fluorescence signals targeted to the osteosarcoma lesions could be determined clearly by ex-vivo imaging. These data were well positively correlated with the in-vivo imaging data (r > 0.8, P < 0.02). Other good to excellent correlations (r > 0.8, P < 0.02) were observed between DG accumulation and tumor gross volume and between OS accumulation and mineralization volume.

CONCLUSION

This in-vivo NIR imaging technique using DG and OS is sensitive to the level to simultaneously detect and quantitatively evaluate the growth and mineralization occuring in this type of osteosarcoma lesions of living mice without either invasion or sacrifice. By possible mutual complementation, this dual imaging system might be useful for accurate diagnosis even in the presence of overlapping tissues.

Imaging technologies for preclinical models of bone and joint disorders

Preclinical models for musculoskeletal disorders are critical for understanding the pathogenesis of bone and joint disorders in humans and the development of effective therapies. The assessment of these models primarily relies on morphological analysis which remains time consuming and costly, requiring large numbers of animals to be tested through different stages of the disease. The implementation of preclinical imaging represents a keystone in the refinement of animal models allowing longitudinal studies and enabling a powerful, non-invasive and clinically translatable way for monitoring disease progression in real time. Our aim is to highlight examples that demonstrate the advantages and limitations of different imaging modalities including magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), single-photon emission computed tomography (SPECT) and optical imaging. All of which are in current use in preclinical skeletal research. MRI can provide high resolution of soft tissue structures, but imaging requires comparatively long acquisition times; hence, animals require long-term anaesthesia. CT is extensively used in bone and joint disorders providing excellent spatial resolution and good contrast for bone imaging. Despite its excellent structural assessment of mineralized structures, CT does not provide in vivo functional information of ongoing biological processes. Nuclear medicine is a very promising tool for investigating functional and molecular processes in vivo with new tracers becoming available as biomarkers. The combined use of imaging modalities also holds significant potential for the assessment of disease pathogenesis in animal models of musculoskeletal disorders, minimising the use of conventional invasive methods and animal redundancy.

Hands-free, wireless goggles for near-infrared fluorescence and real-time image-guided surgery

BACKGROUND

Current cancer management faces several challenges, including the occurrence of a residual tumor after resection, the use of radioactive materials or high concentrations of blue dyes for sentinel lymph node biopsy, and the use of bulky systems in surgical suites for image guidance. To overcome these limitations, we developed a real-time, intraoperative imaging device that, when combined with near infrared fluorescent molecular probes, can aid in the identification of tumor margins, guide surgical resections, map sentinel lymph nodes, and transfer acquired data wirelessly for remote analysis.

METHODS

We developed a new compact, wireless, wearable, and battery-operated device that allows for hands-free operation by surgeons. A charge-coupled device-based, consumer-grade night vision viewer was used to develop the detector portion of the device, and the light source portion was developed from a compact headlamp. This piece was retrofitted to provide both near infrared excitation and white light illumination simultaneously. Wireless communication was enabled by integrating a battery-operated, miniature, radio-frequency video transmitter into the system. We applied the device in several types of oncologic surgical procedures in murine models, including sentinel lymph node mapping, fluorescence-guided tumor resection, and surgery under remote expert guidance.

RESULTS

Unlike conventional imaging instruments, the device displays fluorescence information directly on its eyepiece. When employed in sentinel lymph node mapping, the locations of sentinel lymph nodes were visualized clearly, even with tracer level dosing of a near infrared fluorescent dye (indocyanine green). When used in tumor resection, tumor margins and small nodules invisible to the naked eye were visualized readily. In a simulated, point-of-care setting, tumors were located successfully and removed under remote guidance using the wireless feature of the device. Importantly, the total cost of this prototype system ($1200) is substantially less than existing imaging instruments.

CONCLUSION

Our results demonstrate the feasibility of using the new device to aid surgical resection of tumors, map sentinel lymph nodes, and facilitate telemedicine.