The advent of artificial intelligence (AI) has been transformative across various domains, and nuclear medicine is no exception. The integration of AI into nuclear medicine has opened new avenues for enhancing diagnostic accuracy, personalizing treatment and optimizing workflows. However, this technological advancement is not without its challenges, which include technical, ethical and practical considerations that must be addressed to ensure AI's successful integration into clinical practice (Laudicella et al., 2023; Saboury et al., 2023). We read with great interest the recent article by Saban and colleagues entitled ‘A comparative vignette study: Evaluating the potential role of a generative AI model in enhancing clinical decision-making in nursing’ (Saban & Dubovi, 2024), which investigated the potential of a generative AI tool as clinical support for nurses. The findings of this study show that a generative AI tool demonstrated indecisiveness and a tendency towards over-triage compared to human clinicians. We are writing to express my interest and discuss the utilization of AI in nuclear medicine, focusing on its applications and challenges. Over the years, AI has been integrated into various medical fields, including nuclear medicine, to enhance diagnostic accuracy and improve patient outcomes (Jha et al., 2022; Saboury et al., 2023).

AI's applications in nuclear medicine are vast, ranging from image enhancement to predictive analytics. In diagnostic imaging, AI algorithms have been developed to improve the quality of PET and SPECT images. These algorithms reduce noise and enhance spatial resolution, leading to more accurate diagnostics (Saboury et al., 2023). AI's ability to process and analyse large data sets allows for the development of predictive models that can assist in disease staging and prognosis. One of the significant advancements in AI is its application in PSMA PET/CT for prostate cancer imaging (Lindgren Belal et al., 2024). AI techniques, including machine learning and deep learning algorithms, have demonstrated the potential to match or even surpass human interpretation in detecting primary tumours, local recurrences and metastatic lesions (Saboury et al., 2023). These tools not only improve diagnostic accuracy but also reduce inter-reader variability and save valuable time. AI also plays a crucial role in the development of theranostics, where it aids in designing and optimizing radiopharmaceuticals for targeted therapy. By analysing large data sets of molecular interactions, AI can identify potential therapeutic targets and optimize the synthesis of radiopharmaceuticals. This capability enhances the precision and efficacy of nuclear medicine therapies, leading to better patient outcomes (Laudicella et al., 2023; Saboury et al., 2023).

Despite its potential, the integration of AI in nuclear medicine presents several challenges. One of the primary technical challenges is the need for high-quality, curated data sets for training AI models. The availability of such data is often limited, hindering the development and validation of robust AI algorithms. Additionally, AI models require constant updates and retraining to incorporate new data and adapt to changing clinical practices (Laudicella et al., 2023). Another significant challenge is the ethical and legal implications of AI in health care. The use of AI in patient care raises concerns about data privacy, informed consent and the potential for algorithmic bias. These issues necessitate the establishment of clear regulatory frameworks and guidelines to ensure that AI technologies are used responsibly and ethically (Laudicella et al., 2023; Saban & Dubovi, 2024; Saboury et al., 2023). The implementation of AI in clinical workflows also faces practical barriers, such as resistance from healthcare providers and the need for significant investments in infrastructure and training. To overcome these challenges, it is essential to foster collaboration between AI developers, healthcare professionals and regulatory bodies to create a supportive ecosystem for AI innovation (Laudicella et al., 2023).

Building a trustworthy AI ecosystem in nuclear medicine requires a multi-faceted approach (Jha et al., 2022; Laudicella et al., 2023; Saboury et al., 2023). First, it is crucial to establish standards for the development, evaluation and deployment of AI algorithms. This includes creating guidelines for data management, algorithm validation and post-deployment monitoring. Second, transparency and accountability must be prioritized throughout the AI lifecycle. Artificial intelligence developers should be transparent about the limitations and potential biases of their models, and mechanisms should be in place to hold developers accountable for the performance and outcomes of AI systems. Finally, education and training programs should be implemented to equip healthcare professionals with the skills and knowledge needed to effectively use AI tools. This will not only facilitate the adoption of AI technologies but also empower clinicians to critically evaluate and integrate AI insights into their practice. The future of AI in nuclear medicine is promising, with ongoing research aimed at enhancing the capabilities of AI systems and overcoming existing challenges. Key areas of focus include (Jha et al., 2022; Laudicella et al., 2023; Saboury et al., 2023) (1) Advanced algorithm development: Continued advancements in machine learning and deep learning techniques will enhance the accuracy and efficiency of AI tools in nuclear medicine. (2) Interdisciplinary collaboration: Collaboration between AI experts, clinicians and policymakers is essential to address the multifaceted challenges associated with AI implementation and to develop guidelines that ensure the safe and effective use of AI in healthcare. (3) Real-world implementation: Translating AI research into clinical practice requires pilot studies and real-world evaluations to assess the impact of AI tools on patient outcomes and healthcare systems. (4) Patient-centric approaches: Incorporating patient perspectives and preferences into the design and deployment of AI systems can enhance the relevance and acceptability of AI-driven healthcare solutions.

AI has the potential to revolutionize nuclear medicine by enhancing diagnostic accuracy, personalizing treatment and optimizing clinical workflows. However, realizing this potential requires addressing the technical, ethical and practical challenges associated with AI integration. By fostering collaboration, establishing standards and promoting transparency, we can build a trustworthy AI ecosystem that enhances patient care and advances the field of nuclear medicine.

人工智能（AI）的出现给各个领域带来了变革，核医学也不例外。人工智能与核医学的整合为提高诊断准确性、个性化治疗和优化工作流程开辟了新途径。然而，这种技术进步并非没有挑战，其中包括必须解决技术、伦理和实践方面的考虑，以确保人工智能成功融入临床实践（Laudicella 等人， 2023 年；Saboury 等人， 2023 年）。我们饶有兴趣地阅读了 Saban 及其同事最近发表的题为“比较小插图研究：评估生成式 AI 模型在增强护理临床决策方面的潜在作用”的文章（Saban 和 Dubovi， 2024 年），该文章研究了为护士提供临床支持的生成式人工智能工具。这项研究的结果表明，与人类临床医生相比，生成式人工智能工具表现出优柔寡断和过度分类的倾向。我们写信是为了表达我的兴趣并讨论人工智能在核医学中的应用，重点关注其应用和挑战。多年来，人工智能已融入包括核医学在内的各个医疗领域，以提高诊断准确性并改善患者治疗结果（Jha 等， 2022 ；Saboury 等， 2023 ）。

人工智能在核医学中的应用非常广泛，从图像增强到预测分析。在诊断成像中，人工智能算法已被开发来提高 PET 和 SPECT 图像的质量。这些算法可降低噪声并增强空间分辨率，从而实现更准确的诊断（Saboury 等人， 2023 ）。人工智能处理和分析大数据集的能力可以开发有助于疾病分期和预后的预测模型。 AI 的重大进步之一是其在用于前列腺癌成像的 PSMA PET/CT 中的应用（Lindgren Belal 等人， 2024 ）。包括机器学习和深度学习算法在内的人工智能技术已证明在检测原发性肿瘤、局部复发和转移性病变方面具有匹配甚至超越人类解释的潜力（Saboury 等人， 2023 ）。这些工具不仅提高了诊断准确性，还减少了阅读器之间的差异并节省了宝贵的时间。人工智能在治疗诊断学的发展中也发挥着至关重要的作用，它有助于设计和优化用于靶向治疗的放射性药物。通过分析分子相互作用的大数据集，人工智能可以识别潜在的治疗靶点并优化放射性药物的合成。这种能力提高了核医学治疗的精确度和有效性，从而改善患者的治疗效果（Laudicella 等人， 2023 年；Saboury 等人， 2023 年）。

尽管具有潜力，但人工智能在核医学中的整合面临着一些挑战。主要的技术挑战之一是需要高质量、精心策划的数据集来训练人工智能模型。此类数据的可用性通常有限，阻碍了强大的人工智能算法的开发和验证。此外，人工智能模型需要不断更新和再训练，以纳入新数据并适应不断变化的临床实践（Laudicella 等人， 2023 ）。另一个重大挑战是人工智能在医疗保健领域的伦理和法律影响。人工智能在患者护理中的使用引起了人们对数据隐私、知情同意和潜在算法偏差的担忧。这些问题需要建立明确的监管框架和指南，以确保人工智能技术得到负责任和合乎道德的使用（Laudicella 等人， 2023 年；Saban 和 Dubovi， 2024 年；Saboury 等人， 2023 年）。在临床工作流程中实施人工智能也面临实际障碍，例如医疗保健提供者的抵制以及需要对基础设施和培训进行大量投资。为了克服这些挑战，必须促进人工智能开发人员、医疗保健专业人员和监管机构之间的合作，为人工智能创新创建一个支持性生态系统（Laudicella 等人， 2023 ）。

在核医学领域建立值得信赖的人工智能生态系统需要采取多方面的方法（Jha 等人， 2022 ；Laudicella 等人， 2023 ；Saboury 等人， 2023 ）。首先，建立人工智能算法的开发、评估和部署标准至关重要。这包括为数据管理、算法验证和部署后监控创建指南。其次，在整个人工智能生命周期中必须优先考虑透明度和问责制。人工智能开发人员应该对其模型的局限性和潜在偏差保持透明，并且应该建立机制让开发人员对人工智能系统的性能和结果负责。最后，应实施教育和培训计划，为医疗保健专业人员提供有效使用人工智能工具所需的技能和知识。这不仅将促进人工智能技术的采用，而且使临床医生能够批判性地评估人工智能见解并将其整合到他们的实践中。人工智能在核医学领域的前景广阔，正在进行的研究旨在增强人工智能系统的能力并克服现有的挑战。重点关注领域包括（Jha 等人， 2022 ；Laudicella 等人， 2023 ；Saboury 等人， 2023 ）（1）高级算法开发：机器学习和深度学习技术的持续进步将提高算法的准确性和效率。核医学中的人工智能工具。 (2) 跨学科合作：人工智能专家、临床医生和政策制定者之间的合作对于解决与人工智能实施相关的多方面挑战以及制定确保人工智能在医疗保健中安全有效使用的指南至关重要。 (3) 现实世界的实施：将人工智能研究转化为临床实践需要进行试点研究和现实世界评估，以评估人工智能工具对患者治疗结果和医疗保健系统的影响。 (4) 以患者为中心的方法：将患者的观点和偏好纳入人工智能系统的设计和部署中，可以增强人工智能驱动的医疗保健解决方案的相关性和可接受性。

人工智能有潜力通过提高诊断准确性、个性化治疗和优化临床工作流程来彻底改变核医学。然而，实现这一潜力需要解决与人工智能集成相关的技术、道德和实践挑战。通过促进合作、建立标准和提高透明度，我们可以建立一个值得信赖的人工智能生态系统，以增强患者护理并推动核医学领域的发展。