How Artificial Intelligence Is Shaping Medical Imaging Technology: A Survey of Innovations and Applications

Bioengineering, Journal Year: 2023, Volume and Issue: 10(12), P. 1435 - 1435

Published: Dec. 18, 2023

The integration of artificial intelligence (AI) into medical imaging has guided in an era transformation healthcare. This literature review explores the latest innovations and applications AI field, highlighting its profound impact on diagnosis patient care. innovation segment cutting-edge developments AI, such as deep learning algorithms, convolutional neural networks, generative adversarial which have significantly improved accuracy efficiency image analysis. These enabled rapid accurate detection abnormalities, from identifying tumors during radiological examinations to detecting early signs eye disease retinal images. article also highlights various imaging, including radiology, pathology, cardiology, more. AI-based diagnostic tools not only speed up interpretation complex images but improve disease, ultimately delivering better outcomes for patients. Additionally, processing facilitates personalized treatment plans, thereby optimizing healthcare delivery. paradigm shift that brought role revolutionizing By combining techniques their practical applications, it is clear will continue shaping future positive ways.

Language: Английский

---

Advances in medical image analysis with vision Transformers: A comprehensive review

Medical Image Analysis, Journal Year: 2023, Volume and Issue: 91, P. 103000 - 103000

Published: Oct. 19, 2023

Language: Английский

---

A Survey on Generative Diffusion Models

IEEE Transactions on Knowledge and Data Engineering, Journal Year: 2024, Volume and Issue: 36(7), P. 2814 - 2830

Published: Feb. 2, 2024

Deep generative models have unlocked another profound realm of human creativity. By capturing and generalizing patterns within data, we entered the epoch all-encompassing Artificial Intelligence for General Creativity (AIGC). Notably, diffusion models, recognized as one paramount materialize ideation into tangible instances across diverse domains, encompassing imagery, text, speech, biology, healthcare. To provide advanced comprehensive insights diffusion, this survey comprehensively elucidates its developmental trajectory future directions from three distinct angles: fundamental formulation algorithmic enhancements, manifold applications diffusion. Each layer is meticulously explored to offer a comprehension evolution. Structured summarized approaches are presented here.

Language: Английский

---

Deep Learning Approaches for Data Augmentation in Medical Imaging: A Review

Journal of Imaging, Journal Year: 2023, Volume and Issue: 9(4), P. 81 - 81

Published: April 13, 2023

Deep learning has become a popular tool for medical image analysis, but the limited availability of training data remains major challenge, particularly in field where acquisition can be costly and subject to privacy regulations. Data augmentation techniques offer solution by artificially increasing number samples, these often produce unconvincing results. To address this issue, growing studies have proposed use deep generative models generate more realistic diverse that conform true distribution data. In review, we focus on three types augmentation: variational autoencoders, adversarial networks, diffusion models. We provide an overview current state art each discuss their potential different downstream tasks imaging, including classification, segmentation, cross-modal translation. also evaluate strengths limitations model suggest directions future research field. Our goal is comprehensive review about highlight improving performance algorithms analysis.

Language: Английский

---

Medical Image Segmentation Review: The Success of U-Net

IEEE Transactions on Pattern Analysis and Machine Intelligence, Journal Year: 2024, Volume and Issue: 46(12), P. 10076 - 10095

Published: Aug. 21, 2024

Automatic medical image segmentation is a crucial topic in the domain and successively critical counterpart computer-aided diagnosis paradigm. U-Net most widespread architecture due to its flexibility, optimized modular design, success all modalities. Over years, model has received tremendous attention from academic industrial researchers who have extended it address scale complexity created by tasks. These extensions are commonly related enhancing U-Net's backbone, bottleneck, or skip connections, including representation learning, combining with Transformer architecture, even addressing probabilistic prediction of map. Having compendium different previously proposed variants makes easier for machine learning identify relevant research questions understand challenges biological tasks that challenge model. In this work, we discuss practical aspects organize each variant into taxonomy. Moreover, measure performance these strategies clinical application, propose fair evaluations some unique famous designs on well-known datasets. Furthermore, provide comprehensive implementation library trained models. addition, ease future studies, an online list papers their possible official implementation.

Language: Английский

---

Opportunities and Challenges of Generative AI in Construction Industry: Focusing on Adoption of Text-Based Models

Buildings, Journal Year: 2024, Volume and Issue: 14(1), P. 220 - 220

Published: Jan. 14, 2024

In the last decade, despite rapid advancements in artificial intelligence (AI) transforming many industry practices, construction largely lags adoption. Recently, emergence and adoption of advanced large language models (LLMs) like OpenAI’s GPT, Google’s PaLM, Meta’s Llama have shown great potential sparked considerable global interest. However, current surge lacks a study investigating opportunities challenges implementing Generative AI (GenAI) sector, creating critical knowledge gap for researchers practitioners. This underlines necessity to explore prospects complexities GenAI integration. Bridging this is fundamental optimizing GenAI’s early stage within sector. Given unprecedented capabilities generate human-like content based on learning from existing content, we reflect two guiding questions: What will future bring industry? are delves into reflected perception literature, analyzes using programming-based word cloud frequency analysis, integrates authors’ opinions answer these questions. paper recommends conceptual implementation framework, provides practical recommendations, summarizes research questions, builds foundational literature foster subsequent expansion its allied architecture engineering domains.

Language: Английский

---

Foundation model for cancer imaging biomarkers

Nature Machine Intelligence, Journal Year: 2024, Volume and Issue: 6(3), P. 354 - 367

Published: March 15, 2024

Abstract Foundation models in deep learning are characterized by a single large-scale model trained on vast amounts of data serving as the foundation for various downstream tasks. generally using self-supervised and excel reducing demand training samples applications. This is especially important medicine, where large labelled datasets often scarce. Here, we developed cancer imaging biomarker discovery convolutional encoder through comprehensive dataset 11,467 radiographic lesions. The was evaluated distinct clinically relevant applications imaging-based biomarkers. We found that it facilitated better more efficient biomarkers yielded task-specific significantly outperformed conventional supervised other state-of-the-art pretrained implementations tasks, when sizes were very limited. Furthermore, stable to input variations showed strong associations with underlying biology. Our results demonstrate tremendous potential discovering new may extend clinical use cases can accelerate widespread translation into settings.

Language: Английский

---

A Revolution of Personalized Healthcare: Enabling Human Digital Twin with Mobile AIGC

IEEE Network, Journal Year: 2024, Volume and Issue: 38(6), P. 234 - 242

Published: Feb. 16, 2024

Mobile artificial intelligence-generated content (AIGC) refers to the adoption of generative intelligence (GAI) algorithms deployed at mobile edge networks automate information creation process while fulfilling requirements end users. AIGC has recently attracted phenomenal attentions and can be a key enabling technology for an emerging application, called human digital twin (HDT). HDT empowered by is expected revolutionize personalized healthcare generating rare disease data, modeling high-fidelity twin, building versatile testbeds, providing 24/7 customized medical services. To promote development this new breed paradigm, in article, we propose system architecture AIGC-driven highlight corresponding design challenges. Moreover, illustrate two use cases, i.e., surgery planning medication. In addition, conduct experimental study prove effectiveness proposed solution, which shows particular application virtual physical therapy teaching platform. Finally, conclude article briefly discussing several open issues future directions.

Language: Английский

---

A survey on training challenges in generative adversarial networks for biomedical image analysis

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(2)

Published: Jan. 29, 2024

Abstract In biomedical image analysis, the applicability of deep learning methods is directly impacted by quantity data available. This due to models requiring large datasets provide high-level performance. Generative Adversarial Networks (GANs) have been widely utilized address limitations through generation synthetic images. GANs consist two models. The generator, a model that learns how produce images based on feedback it receives. discriminator, classifies an as or real and provides generator. Throughout training process, GAN can experience several technical challenges impede suitable imagery. First, mode collapse problem whereby generator either produces identical uniform from distinct input features. Second, non-convergence gradient descent optimizer fails reach Nash equilibrium. Thirdly, vanishing unstable behavior occurs discriminator achieving optimal classification performance resulting in no meaningful being provided These problems result production imagery blurry, unrealistic, less diverse. To date, there has survey article outlining impact these context domain. work presents review taxonomy solutions imaging highlights important outlines future research directions about domain

Language: Английский

---

Multi-channel Optimization Generative Model for Stable Ultra-Sparse-View CT Reconstruction

IEEE Transactions on Medical Imaging, Journal Year: 2024, Volume and Issue: 43(10), P. 3461 - 3475

Published: March 11, 2024

Score-based generative model (SGM) has risen to prominence in sparse-view CT reconstruction due its impressive generation capability. The consistency of data is crucial guiding the process SGM-based methods. However, existing policy exhibits certain limitations. Firstly, it employs partial from reconstructed image iteration for updates, which leads secondary artifacts with compromising quality. Moreover, updates SGM and are considered as distinct stages, disregarding their interdependent relationship. Additionally, reference used compute gradients derived intermediate result rather than ground truth. Motivated by fact that a typical yields outcomes different random noise inputs, we propose Multi-channel Optimization Generative Model (MOGM) stable ultra-sparse-view integrating novel term into stochastic differential equation model. Notably, unique aspect this component exclusive reliance on original effectively confining outcomes. Furthermore, pioneer an inference strategy traces back current truth, enhancing stability through foundational theoretical support. We also establish multi-channel optimization framework, where conventional iterative techniques employed seek solution. Quantitative qualitative assessments 23 views datasets numerical simulation, clinical cardiac sheep's lung underscore superiority MOGM over alternative Reconstructing just 10 7 views, our method consistently demonstrates exceptional performance.

Language: Английский

---