Netter's Atlas Of Hu..
Most recently, deep learning-based segmentation has shown enormous potential in providing accurate and consistent results10,11,14,15,16, in comparison to most classification and regression approaches, such as atlas-based contouring, statistical shape modeling, and so on17,18,19,20. The most popular architecture is convolutional neural networks (CNNs)21,22,23, including U-Net24,25, V-Net26, as well as nnU-Net27, which achieve excellent performance in Medical Image Decathlon Segmentation Competition. Besides, other hybrid algorithms also have shown outstanding segmentation performance28,29,30, i.e., Swin UNETR31. However, deep learning-based algorithm needs specific computing resources such as graphics processing unit (GPU) memory, especially for 3D image processing13, thus leading to limited clinical applications in practice.
Netter's Atlas of Hu..
So far, we have demonstrated that the proposed deep learning-based segmentation framework can automatically, efficiently and accurately delineate the OARs and target volumes. There are multiple AI-based software tools that are commercially available and have been used in clinical practices to standardize and accelerate the RT procedures. They include atlas-based contouring tool for automatic segmentation12,34,35,36,37, and knowledge-based planning module for automatic treatment planning38,39,40. Here, we focus on exploring of AI-based automatic segmentation of target volumes and its integration into RT workflows. These AI solutions have reportedly achieved comparable performance with manual delineations in segmentation accuracy, with minor editing efforts needed12,35. However, majority of the studies were only evaluated on limited organs and data with specific acquisition protocols, which affects their clinical applicability when used in different hospitals or for different target volumes. Two studies have tried to address this challenge to improve the model generalizability41,42. Nikolov et al. applied 3D U-Net to delineate 21 OARs in head and neck CT scans, and achieved expert-level performance41. The study was conducted on the training set (663 scans) and testing set (21 scans) from routine clinical practice, and validation set (39 scans) from two distinct open-source datasets. Oktay et al. incorporated the AI model into the existing RT workflow, and demonstrated that AI model could reduce contouring time while yielding clinical valid structural contours for both prostate and head-and-neck RT planning42. Their study involved 6 OARs for prostate cancer and 9 OARs for head-and-neck cancer, where experiments were conducted on a set of 519 pelvic and 242 head-and-neck CT scans acquired at eight distinct clinical sites with heterogeneous population groups and diverse image acquisition protocols. In contrast to previous works, we evaluate how RTP-Net can lead to generalized performance with extensive evaluation on 67 target volumes with varying volume sizes on a large-scale dataset of 28,581 cases (Supplementary Fig. 1). This large-scale dataset was obtained from eight distinct publicly-available datasets and one local dataset with varying acquisition settings and demographics (Supplementary Table 5). Our proposed model demonstrates performance generalizability across hospitals and target volumes, while achieving superior levels of agreement with expert contours and also time savings, which can facilitate easier deployment in clinical sites.
A total of 27 anatomical structures are contoured. The anatomical definitions of 25 structures refer to the Brouwer atlas53 and neuroanatomy textbook54, i.e., brain, brainstem, eyes (left and right), parotid glands (left and right), bone mandibles (left and right), lens (left and right), oral cavity, joint TM (left and right), lips, teeth, submandibular gland (left and right), glottis, pharyngeal constrictor muscles (superior, middle, and inferior), pituitary, chiasm, and brachial plex (left and right). The contouring of temporal lobes (left and right) refers to the brain atlas55.
A total of 16 anatomical structures are contoured, in which 8 anatomical structures are defined following the Radiation Therapy Oncology Group (RTOG) guideline 110656 and the textbook of cardiothoracic anatomy57, i.e., heart, lungs (left and right), ascending aorta, esophagus, vertebral body, trachea, and rib. Breast (left and right), breast_PRV05 (left and right), mediastinal lymph nodes, and humerus head (left and right) are contoured referring to the RTOG breast cancer atlas58. Moreover, the contouring of NSCLC follows RTOG 051559.
A total of 115 participants (61 male and 54 female) were recruited by email announcement, and allocated into two groups, a VR intervention group (n = 57) and a control group (n = 58). 14 participants dropped out (10 from the intervention group and 4 from the control group) due to participation in informal ultrasound training after initial recruitment. A total of 101 participants participated in this study. (Fig 1) Participants in the VR intervention group used VR as part of their training during the course, including a self-directed VR-enhanced anatomy review of thorax and abdomen, and additional VR review sessions during ultrasonography hands-on practice. The participants in the control group took part in an ultrasound workshop of similar design. The VR anatomy component was replaced with a review session using a digital atlas. Upon conclusion of the workshop all participants (control and intervention groups) were assessed using a standardized practical multi-station ultrasonography test.
Netter's Atlas of Human Anatomy is the most loved and best selling anatomy atlas in the English language. In over 540 beautifully colored and easily understood illustrations, it teaches the complete human body with unsurpassed clarity and accuracy. This new edition features 57 revised, 200 relabeled and 17 wholly new plates, drawn fully in the tradition of Frank Netter, and includes more imaging and clinical images than ever before. Six Consulting Editors have worked together to ensure the new edition's accuracy and usefulness in the lecture theatre, classroom and dissection lab. Fifty plates from the book as well as a powerful and varied bank of ancillary material, unique to this atlas, are available online through STUDENT CONSULT.
For students and clinical professionals who are learning anatomy, participating in a dissection lab, sharing anatomy knowledge with patients, or refreshing their anatomy knowledge, the Netter Atlas of Human Anatomy illustrates the body, system by system, in clear, brilliant detail from a clinician's perspective. Unique among anatomy atlases, it contains illustrations that emphasize anatomic relationships that are most important to the clinician in training and practice. Illustrated by clinicians, for clinicians, it contains more than 550 exquisite plates plus dozens of carefully selected radiologic images for common views.
Despite recent advancements in segmentation methods for brain tissue with magnetic resonance images (MRI) , there is no automatic segmentation tool available for nonbrain tissues such as extracranial tissues like cartilages, fats, and neck muscles. This was owing to the fact that segmentation of these tissue types was often ignored since these tissues were regarded as less important as compared with the skull-brain tissue and were not usually considered in the FE head model. Based on the reference to available atlas of head anatomy , the geometry of the cartilages, namely, the cartilage of septum and the lower and upper lateral cartilages of the human nose, is reconstructed semiautomatically using an adaptive moving mesh technique and shape preserving parameterization. The models also contain some of the interior details, which are often ignored in previous models, such as air sinuses, namely, maxillary sinuses, frontal sinus, and sphenoidal sinuses (Figure 2) [5, 23].
Winkler EA, Kim CN, Ross JM, Garcia JH, Gil E, Oh I, Chen LQ, Wu D, Catapano JS, Raygor K, Narsinh K, Kim H, Weinsheimer S, Cooke DL, Walcott BP, Lawton MT, Gupta N, Zlokovic BV, Chang EF, Abla AA, Lim DA, Nowakowski TJ. A single-cell atlas of the normal and malformed human brain vasculature. Science. 2022 03 04; 375(6584):eabi7377. PubMed 041b061a72