Explore chapters and articles related to this topic
Introduction to Artificial Intelligence in Healthcare
Published in Punit Gupta, Dinesh Kumar Saini, Rohit Verma, Healthcare Solutions Using Machine Learning and Informatics, 2023
Divyani Jigyasu, Sunil Kumar, Rajveer Singh Shekhawat, Shally Vats
The process of transforming several photographs into a single coordinate system is known as registration. This step is required when analyzing photographs taken from various perspectives, at various times, and with various modalities. Previously, clinicians had to perform picture registration manually, and most registration tasks can be extremely difficult. Manual alignment quality is heavily dependent on the user’s expertise, which might be therapeutically detrimental. Automatic registration was created to compensate for the flaws in manual registration (see Figure 1.4).Several deep learning approaches for automatic image registration have been intensively investigated, transforming picture registration as shown in Figure 1.2.
A Combination of Dilated Adversarial Convolutional Neural Network and Guided Active Contour Model for Left Ventricle Segmentation
Published in Kayvan Najarian, Delaram Kahrobaei, Enrique Domínguez, Reza Soroushmehr, Artificial Intelligence in Healthcare and Medicine, 2022
Heming Yao, Jonathan Gryak, Kayvan Najarian
In this class of segmentation techniques, an atlas that describes the structure of the target object is first generated from one or more manually delineated images. Image registration is a process of mapping the coordinates of one image to those of another image. Given an atlas, a new image can be segmented by image registration between the new image and the atlas. A three-step registration framework was proposed in Zhuang et al. (2010). A global affine registration was first applied to localize the heart, then a locally affine registration method was employed to initialize the substructures. Finally, a free-form deformation with adaptive control point status was proposed to refine the local details. The method achieved good performance on heart images with diverse morphology and pathology. In Lorenzo-Valdés et al. (2004), a 4D probabilistic heart atlas was constructed from 14 healthy cases, which encodes both spatial anatomical information and temporal information. With the 4D probabilistic atlas, the expectation-maximization algorithm was applied to perform the segmentation.
Image Registration, Segmentation and Virtual Simulation
Published in W. P. M. Mayles, A. E. Nahum, J.-C. Rosenwald, Handbook of Radiotherapy Physics, 2021
Vibeke Nordmark Hansen, J.-C. Rosenwald
According to Maintz and Viergever (1998), medical image registration can be divided into two main categories: Extrinsic methods, based on additional objects visible on all images of the multimodalilty dataset;Intrinsic methods, based on anatomical information belonging to the patient and visible on all images.
A review on the applications of virtual reality, augmented reality and mixed reality in surgical simulation: an extension to different kinds of surgery
Published in Expert Review of Medical Devices, 2021
Abel J Lungu, Wout Swinkels, Luc Claesen, Puxun Tu, Jan Egger, Xiaojun Chen
Image registration is required to align the virtual data accurately with the physical scene. In [19,20], a multi-step co-registration strategy has been adopted in which four fiducial markers are placed around the surgical site. During navigation, these markers track the tumor margin that needs to be resected. To register the virtual coordinate system to that of the physical coordinate system, Wang et al. [23] adopted a point-to-point registration method. For the extraction of the fiducial landmark coordinates in the virtual coordinate system, image processing is applied. The fiducial landmark coordinates have been obtained using a positioning probe in the physical coordinate system. Liu et al. [24] adopted a similar point-based registration approach. To improve the registration accuracy, Chen et al. [22] combine fiducial point-based registration with surface-based registration. These registration methods are based upon visible markers. However, there are also marker-less methods that can be used for image registration. It has been demonstrated that the SIFT, SURF, BRISK and ORB algorithms can be used in fluorescence-to-color image registration for intraoperative AR [39]. It is also possible to use anatomical landmarks to perform image registration. Wang et al. [34] use patient tracking in combination with 3D contour matching of the teeth to obtain automatic marker-free patient-image registration. Similar approaches have been used by Suenaga et al. [36] and Wang et al. [37,38].
Towards a generalised development of synthetic CT images and assessment of their dosimetric accuracy
Published in Acta Oncologica, 2020
Josefine Handrack, Mark Bangert, Christian Möhler, Tilman Bostel, Steffen Greilich
The usage of BDA specifically for pelvic cancer has been widely investigated in the literature, however mostly for water equivalent (WE) bulk densities [4,17] or WE and bone [4,5,18] as surrogate EDs. The commercial approach described in Köhler et al. [14] increased the complexity of the sCT by using five different tissue classes for BDA. The next increase in complexity is the dual model [19,20], which separates bone (polynomial fit) and soft tissue (interpolation between fat, muscle and urine) for sCT generation. Machine learning and image registration are the most complex and computational demanding approaches. However, at some point an increase in complexity of the sCT generation might not result in a relevant increase in dosimetric accuracy any more. This trade-off has not yet been analysed systematically.
Image-guided radiation therapy for post-operative gynaecologic cancer: patient set up verification with and without implanted fiducial markers
Published in Acta Oncologica, 2018
Donna H. Murrell, Andrew Warner, Quinn Benwell, Wendy Wells, Danielle Scott, Vikram Velker, George Hajdok, David P. D’Souza
Sources of uncertainty should be considered when interpreting the reported shifts between IGRT techniques. Of note, all image registrations were performed with a compromise to maintain bony alignment within 5 mm. This may result in imperfect registration at the primary target; however, this approach is consistent with clinical implementation of IGRT techniques for treatments that include nodal volumes. In this study, if a match was not possible within the restriction of bony alignment, patients were taken off the bed, set-up again, and a new CBCT was acquired. Manual image registration reproducibility may also contribute to uncertainty but was not investigated in this study. All registrations required two therapists to reach consensus on the match, as well as subsequent review for accuracy by a Radiation Oncologist. While these sources of uncertainty should be acknowledged, it is likely that targeting error in the era of IGRT is dominated by contouring variability.