Mitigation of motion-induced artifacts in cone beam computed tomography using deep convolutional neural networks.

BACKGROUND: Cone beam computed tomography (CBCT) is often employed on radiation therapy treatment devices (linear accelerators) used in image-guided radiation therapy (IGRT). For each treatment session, it is necessary to obtain the image of the day in order to accurately position the patient and to enable adaptive treatment capabilities including auto-segmentation and dose calculation. Reconstructed CBCT images often suffer from artifacts, in particular those induced by patient motion. Deep-learning based approaches promise ways to mitigate such artifacts. PURPOSE: We propose a novel deep-learning based approach with the goal to reduce motion induced artifacts in CBCT images and improve image quality. It is based on supervised learning and includes neural network architectures employed as pre- and/or post-processing steps during CBCT reconstruction. METHODS: Our approach is based on deep convolutional neural networks which complement the standard CBCT reconstruction, which is performed either with the analytical Feldkamp-Davis-Kress (FDK) method, or with an iterative algebraic reconstruction technique (SART-TV). The neural networks, which are based on refined U-net architectures, are trained end-to-end in a supervised learning setup. Labeled training data are obtained by means of a motion simulation, which uses the two extreme phases of 4D CT scans, their deformation vector fields, as well as time-dependent amplitude signals as input. The trained networks are validated against ground truth using quantitative metrics, as well as by using real patient CBCT scans for a qualitative evaluation by clinical experts. RESULTS: The presented novel approach is able to generalize to unseen data and yields significant reductions in motion induced artifacts as well as improvements in image quality compared with existing state-of-the-art CBCT reconstruction algorithms (up to +6.3 dB and +0.19 improvements in peak signal-to-noise ratio, PSNR, and structural similarity index measure, SSIM, respectively), as evidenced by validation with an unseen test dataset, and confirmed by a clinical evaluation on real patient scans (up to 74% preference for motion artifact reduction over standard reconstruction). CONCLUSIONS: For the first time, it is demonstrated, also by means of clinical evaluation, that inserting deep neural networks as pre- and post-processing plugins in the existing 3D CBCT reconstruction and trained end-to-end yield significant improvements in image quality and reduction of motion artifacts.

A Distinct Role of the Autonomic Nervous System in Modulating the Function of Lymphatic Vessels under Physiological and Tumor-Draining Conditions.

Lymphatic vessels (LVs) are important in the regulation of tissue fluid homeostasis and the pathogenesis of tumor progression. We investigated the innervation of LVs and the response to agonists and antagonists of the autonomic nervous system in vivo. While skin-draining collecting LVs express muscarinic, α1- and ß2-adrenergic receptors on lymphatic endothelial cells and smooth muscle cells, intestinal lacteals express only ß-adrenergic receptors and muscarinic receptors on their smooth muscle cells. Quantitative in vivo near-infrared imaging of the exposed flank-collecting LV revealed that muscarinic and α1-adrenergic agonists increased LV contractility, whereas activation of ß2-adrenergic receptors inhibited contractility and initiated nitric oxide (NO)-dependent vasodilation. Tumor-draining LVs were expanded and showed a higher innervation density and contractility that was reduced by treatment with atropine, phentolamine, and, most potently, isoproterenol. These findings likely have clinical implications given the impact of lymphatic fluid drainage on intratumoral fluid pressure and thus drug delivery.

AutoTube: a novel software for the automated morphometric analysis of vascular networks in tissues.

Due to their involvement in many physiologic and pathologic processes, there is a great interest in identifying new molecular pathways that mediate the formation and function of blood and lymphatic vessels. Vascular research increasingly involves the image-based analysis and quantification of vessel networks in tissue whole-mounts or of tube-like structures formed by cultured endothelial cells in vitro. While both types of experiments deliver important mechanistic insights into (lymph)angiogenic processes, the manual analysis and quantification of such experiments are typically labour-intensive and affected by inter-experimenter variability. To bypass these problems, we developed AutoTube, a new software that quantifies parameters like the area covered by vessels, vessel width, skeleton length and branching or crossing points of vascular networks in tissues and in in vitro assays. AutoTube is freely downloadable, comprises an intuitive graphical user interface and helps to perform otherwise highly time-consuming image analyses in a rapid, automated and reproducible manner. By analysing lymphatic and blood vascular networks in whole-mounts prepared from different tissues or from gene-targeted mice with known vascular abnormalities, we demonstrate the ability of AutoTube to determine vascular parameters in close agreement to the manual analyses and to identify statistically significant differences in vascular morphology in tissues and in vascular networks formed in in vitro assays.

Mechanisms of Tumor-Induced Lymphovascular Niche Formation in Draining Lymph Nodes.

Enlargement of the lymphatic vascular network in tumor-draining lymph nodes (LNs) often precedes LN metastasis, likely providing a lymphovascular niche for tumor cells. We investigated morphological and molecular changes associated with the lymphatic remodeling process, using the 4T1 breast cancer and B16F10 melanoma models. Lymphatic expansion in tumor-draining LNs is mediated by sprouting and proliferation of lymphatic endothelial cells (LECs) as early as 4 days after tumor implantation. RNA sequencing revealed an altered transcriptional profile of LECs from tumor-draining compared to naive LNs with similar changes in both tumor models. Integrin αIIb is upregulated in LECs of tumor-draining LNs and mediates LEC adhesion to fibrinogen in vitro. LEC-associated fibrinogen was also detected in LNs in vivo, suggesting a role of integrin αIIb in lymphatic remodeling. Together, our results identify specific responses of LN LECs to tumor stimuli and provide insights into the mechanisms of lymphovascular niche formation in tumor-draining LNs.