Integrating Diagnostic B-Mode Ultrasonography Into CT-Based Radiation Treatment Planning

Abstract

Abstract-This paper presents methods and a clinical procedure for integrating B-mode ultrasound images tagged with position information with a planning computed tomography (CT) scan for radiotherapy. A workflow is described that allows the integration of these modalities into the clinic. A surface mapping approach provides a preregistration of the ultrasound image borders onto the patient's skin. Successively, a set of individual ultrasound images from a freehand sweep is chosen by the physician. These images are automatically registered with the planning CT scan using novel intensity-based methods. We put a particular focus on deriving an appropriate similarity measure based on the physical properties and artifacts of ultrasound. A combination of a weighted mutual information term, edge correlation, clamping to the skin surface, and occlusion detection is able to assess the alignment of structures in ultrasound images and information reconstructed from the CT data. We demonstrate the practicality of our methods on five patients with head and neck tumors and cervical lymph node metastases and provide a detailed report on the conducted experiments, including the setup, calibration, acquisition, and verification of our algorithms. The mean target registration error on nine data sets is 3.9 mm. Thus, the additional information about intranodal architecture and fulfillment of malignancy criteria derived from a high-resolution ultrasonography of lymph nodes can be localized and visualized in the CT scan coordinate space and is made available for further radiation treatment planning.

Segmentation of the liver in ultrasound: a dynamic texture approach

“…Feature based registration techniques also suffer from major difficulties encountered during the intensity based segmentation of US. There are several studies reported where MR or CT was registered either to sequence of tracked 2D US images [8,9] or to 3DUS volumes [10][11][12]. Both approaches ignore the dynamic nature of the US modality, that is, they treat the US images as static and rely on the persistence of features over time.…”

Keep Breathing! Common Motion Helps Multi-modal Mapping

“…Using this approach, we predicted the location of anatomical landmarks by relying on a temporal pre-therapeutical MRI sequence based on the observation of US images acquired during therapy with an average accuracy of 4.2 mm. Our performance is comparable to state-of-the-art methods (3.7 mm for multi-modal registration [11,17,6], 1.5 mm for US tracking [2,3]) while being less complex. Lowest mean errors were achieved when employing SFA.…”

“…A main challenge is to define an effective image similarity measure. For liver US images, approaches include gradient images [9], matching of extracted vessels [13,11], simulations of US from CT images [17,5] and hybrids [6]. Assuming the same breathing phase, alignments are often optimized with respect to a rigid transformation.…”

Registration of renal SPECT and 2.5D US images

“…A study that uses US and CT (computer tomography) images of kidneys is described in [16], where the registration is carried out by minimizing a correlation ratio (CR) between the images pre-processed to remove noise and enhance edges. There are also registrations that use US and CT images in different organs, such as in [17], where images of head and neck are registered using a multi-component similarity measure involving weighted MI, intensities and edge maps. Another example using CT images is [18], in which US images of liver and kidney are registered with CT by a method that simulates US images using the CT data, and then registers the US image applying a similarity measure between it and the simulation.…”