Generative AI in orthopedics: an explainable deep few-shot image augmentation pipeline for plain knee radiographs and Kellgren-Lawrence grading.

OBJECTIVES: Recently, deep learning medical image analysis in orthopedics has become highly active. However, progress has been restricted by the absence of large-scale and standardized ground-truth images. To the best of our knowledge, this study is the first to propose an innovative solution, namely a deep few-shot image augmentation pipeline, that addresses this challenge by synthetically generating knee radiographs for training downstream tasks, with a specific focus on knee osteoarthritis Kellgren-Lawrence (KL) grading. MATERIALS AND METHODS: This study leverages a deep few-shot image augmentation pipeline to generate synthetic knee radiographs. Despite the limited availability of training samples, we demonstrate the capability of our proposed computational strategy to produce high-fidelity plain knee radiographs and use them to successfully train a KL grade classifier. RESULTS: Our experimental results showcase the effectiveness of the proposed computational pipeline. The generated synthetic radiographs exhibit remarkable fidelity, evidenced by the achieved average Frechet Inception Distance (FID) score of 26.33 for KL grading and 22.538 for bilateral knee radiographs. For KL grading classification, the classifier achieved a test Cohen's Kappa and accuracy of 0.451 and 0.727, respectively. Our computational strategy also resulted in a publicly and freely available imaging dataset of 86 000 synthetic knee radiographs. CONCLUSIONS: Our approach demonstrates the capability to produce top-notch synthetic knee radiographs and use them for KL grading classification, even when working with a constrained training dataset. The results obtained emphasize the effectiveness of the pipeline in augmenting datasets for knee osteoarthritis research, opening doors for broader applications in orthopedics, medical image analysis, and AI-powered diagnosis.

Evolution of the meta-neurosurgeon: A systematic review of the current technical capabilities, limitations, and applications of augmented reality in neurosurgery.

Background: Augmented reality (AR) applications in neurosurgery have expanded over the past decade with the introduction of headset-based platforms. Many studies have focused on either preoperative planning to tailor the approach to the patient's anatomy and pathology or intraoperative surgical navigation, primarily realized as AR navigation through microscope oculars. Additional efforts have been made to validate AR in trainee and patient education and to investigate novel surgical approaches. Our objective was to provide a systematic overview of AR in neurosurgery, provide current limitations of this technology, as well as highlight several applications of AR in neurosurgery. Methods: We performed a literature search in PubMed/Medline to identify papers that addressed the use of AR in neurosurgery. The authors screened three hundred and seventy-five papers, and 57 papers were selected, analyzed, and included in this systematic review. Results: AR has made significant inroads in neurosurgery, particularly in neuronavigation. In spinal neurosurgery, this primarily has been used for pedicle screw placement. AR-based neuronavigation also has significant applications in cranial neurosurgery, including neurovascular, neurosurgical oncology, and skull base neurosurgery. Other potential applications include operating room streamlining, trainee and patient education, and telecommunications. Conclusion: AR has already made a significant impact in neurosurgery in the above domains and has the potential to be a paradigm-altering technology. Future development in AR should focus on both validating these applications and extending the role of AR.

Feasibility and Usability of Augmented Reality Technology in the Orthopaedic Operating Room.

PURPOSE OF REVIEW: Augmented reality (AR) has gained popularity in various sectors, including gaming, entertainment, and healthcare. The desire for improved surgical navigation within orthopaedic surgery has led to the evaluation of the feasibility and usability of AR in the operating room (OR). However, the safe and effective use of AR technology in the OR necessitates a proper understanding of its capabilities and limitations. This review aims to describe the fundamental elements of AR, highlight limitations for use within the field of orthopaedic surgery, and discuss potential areas for development. RECENT FINDINGS: To date, studies have demonstrated evidence that AR technology can be used to enhance navigation and performance in orthopaedic procedures. General hardware and software limitations of the technology include the registration process, ergonomics, and battery life. Other limitations are related to the human response factors such as inattentional blindness, which may lead to the inability to see complications within the surgical field. Furthermore, the prolonged use of AR can cause eye strain and headache due to phenomena such as the vergence-convergence conflict. AR technology may prove to be a better alternative to current orthopaedic surgery navigation systems. However, the current limitations should be mitigated to further improve the feasibility and usability of AR in the OR setting. It is important for both non-clinicians and clinicians to work in conjunction to guide the development of future iterations of AR technology and its implementation into the OR workflow.

Understanding Effects of Visual Feedback Delay in AR on Fine Motor Surgical Tasks.

Latency is a pervasive issue in various systems that can significantly impact motor performance and user perception. In medical settings, latency can hinder surgeons' ability to quickly correct movements, resulting in an experience that doesn't align with user expectations and standards of care. Despite numerous studies reporting on the negative effects of latency, there is still a gap in understanding how it impacts the use of augmented reality (AR) in medical settings. This study aims to address this gap by examining how latency impacts motor task performance and subjective perceptions, such as cognitive load, on two display types: a monitor display, traditionally used inside an operating room (OR), and a Microsoft HoloLens 2 display. Our findings indicate that both level of latency and display type impact motor performance, and higher latencies on the HoloLens result in relatively poor performance. However, cognitive load was found to be unrelated to display type or latency, but was dependent on the surgeon's training level. Surgeons did not compromise accuracy to gain more speed and were generally well aware of the latency in the system irrespective of their performance on task. Our study provides valuable insights into acceptable thresholds of latency for AR displays and proposes design implications for the successful implementation and use of AR in surgical settings.

Toward the Design of Sensing-Based Medication Adherence Aids That Support Individualized Activities of Daily Living: Survey and Interviews With Patients and Providers.

BACKGROUND: Nearly half of Americans taking prescription medications do not take them properly. The resulting implications have a broad impact. Nonadhering patients develop worsened medical conditions and increased comorbidity of disease or die. OBJECTIVE: Clinical studies have shown that the most effective strategies for addressing adherence are those that are individualized to the context that each patient and situation require. However, existing aids for adherence are relatively ridged and poorly support adaptation to individual behaviors and lifestyles. The aim of our study was to better understand this design tension. METHODS: A series of 3 qualitative studies was conducted: a web-based survey of 200 Americans that investigated existing adherence strategies and behaviors and perception of how hypothetical in-home tracking technologies would assist adherence; in-person semistructured interviews with 20 medication takers from Pittsburgh, PA, that investigated personal adherence behaviors, which included demonstration of medication locations and routines as well as an assessment of hypothetical technologies; and semistructured interviews with 6 pharmacists and 3 family physicians to gain a provider perspective on patient adherence strategies, which included feedback on hypothetical technologies in the context of their patient populations. Inductive thematic coding of all interview data was performed. Studies were conducted consecutively, with the results informing the subsequent studies. RESULTS: Synthesized, the studies identified key medication adherence behaviors amenable to technological interventions, distilled important home-sensing literacy considerations, and detailed critical privacy considerations. Specifically, 4 key insights were obtained: medication routines are heavily influenced and adapted by and through the physical location and placement of medications relative to activities of daily living, routines are chosen to be inconspicuous to maintain privacy, the value of provider-involved routines is motivated by a desire to build trust in shared decision-making, and the introduction of new technologies can create further burden on patients and providers. CONCLUSIONS: There is considerable potential to improve individual medication adherence by creating behavior-focused interventions that leverage emerging artificial intelligence (AI), machine learning (ML), and in-home Internet of Things (IoT) sensing technologies. However, success will be dependent on the technology's ability to learn effectively and accurately from individual behaviors, needs, and routines and tailor interventions accordingly. Patient routines and attitudes toward adherence will likely affect the use of proactive (eg, AI-assistant routine modification) versus reactive (eg, notification of associated behaviors with missed dosages) intervention strategies. Successful technological interventions must support the detection and tracking of patient routines that can adjust to variations in patient location, schedule, independence, and habituation.

A systematic comparison of the accuracy of monocular RGB tracking and LiDAR for neuronavigation.

With the advent of augmented reality (AR), the use of AR-guided systems in the field of medicine has gained traction. However, the wide-scale adaptation of these systems requires highly accurate and reliable tracking. In this work, the tracking accuracy of two technology platforms, LiDAR and Vuforia, are developed and rigorously tested for a catheter placement neurological procedure. Several experiments (900) are performed for each technology across various combinations of catheter lengths and insertion trajectories. This analysis shows that the LiDAR platform outperformed Vuforia; which is the state-of-the-art in monocular RGB tracking solutions. LiDAR had 75% less radial distance error and 26% less angle deviation error. Results provide key insights into the value and utility of LiDAR-based tracking in AR guidance systems.