Machine Learning Empowering Personalized Medicine: A Comprehensive Review of Medical Image Analysis Methods

Electronics, Journal Year: 2023, Volume and Issue: 12(21), P. 4411 - 4411

Published: Oct. 25, 2023

Artificial intelligence (AI) advancements, especially deep learning, have significantly improved medical image processing and analysis in various tasks such as disease detection, classification, anatomical structure segmentation. This work overviews fundamental concepts, state-of-the-art models, publicly available datasets the field of imaging. First, we introduce types learning problems commonly employed then proceed to present an overview used methods, including convolutional neural networks (CNNs), recurrent (RNNs), generative adversarial (GANs), with a focus on task they are solving, object detection/localization, segmentation, generation, registration. Further, highlight studies conducted application areas, encompassing neurology, brain imaging, retinal analysis, pulmonary digital pathology, breast cardiac bone abdominal musculoskeletal The strengths limitations each method carefully examined, paper identifies pertinent challenges that still require attention, limited availability annotated data, variability images, interpretability issues. Finally, discuss future research directions particular developing explainable methods integrating multi-modal data.

Language: Английский

---

Retinal Imaging-Based Oculomics: Artificial Intelligence as a Tool in the Diagnosis of Cardiovascular and Metabolic Diseases

Biomedicines, Journal Year: 2024, Volume and Issue: 12(9), P. 2150 - 2150

Published: Sept. 23, 2024

Cardiovascular diseases (CVDs) are a major cause of mortality globally, emphasizing the need for early detection and effective risk assessment to improve patient outcomes. Advances in oculomics, which utilize relationship between retinal microvascular changes systemic vascular health, offer promising non-invasive approach assessing CVD risk. Retinal fundus imaging optical coherence tomography/angiography (OCT/OCTA) provides critical information diagnosis, with parameters such as vessel caliber, tortuosity, branching patterns identified key biomarkers. Given large volume data generated during routine eye exams, there is growing automated tools aid diagnosis prediction. The study demonstrates that AI-driven analysis images can accurately predict cardiovascular factors, events, metabolic diseases, surpassing traditional diagnostic methods some cases. These models achieved area under curve (AUC) values ranging from 0.71 0.87, sensitivity 71% 89%, specificity 40% 70%, This highlights potential component personalized medicine, enabling more precise earlier intervention. It not only aids detecting abnormalities may precede events but also offers scalable, non-invasive, cost-effective solution widespread screening. However, article emphasizes further research standardize protocols validate clinical utility these biomarkers across different populations. By integrating oculomics into practice, healthcare providers could significantly enhance management ultimately improving Fundus image thus represents valuable tool future precision medicine health management.

Language: Английский

---

Oculomics: A Crusade Against the Four Horsemen of Chronic Disease

Ophthalmology and Therapy, Journal Year: 2024, Volume and Issue: 13(6), P. 1427 - 1451

Published: April 17, 2024

Chronic, non-communicable diseases present a major barrier to living long and healthy life. In many cases, early diagnosis can facilitate prevention, monitoring, treatment efforts, improving patient outcomes. There is therefore critical need make screening techniques as accessible, unintimidating, cost-effective possible. The association between ocular biomarkers systemic health disease (oculomics) presents an attractive opportunity for detection of diseases, ophthalmic are often relatively low-cost, fast, non-invasive. this review, we highlight the key associations structural in eye four globally leading causes morbidity mortality: cardiovascular disease, cancer, neurodegenerative metabolic disease. We observe that particularly promising target oculomics, with detected multiple structures. Cardiovascular choroid, retinal vasculature, nerve fiber layer, eyelid, tear fluid, lens, vasculature. contrast, only fluid emerged cancer. retina rich source oculomics data, analysis which has been enhanced by artificial intelligence-based tools. Although not all disease-specific, limiting their current diagnostic utility, future research will likely benefit from combining data various structures improve specificity, well active design, development, optimization instruments specific signatures, thus facilitating differential diagnoses. Long-term stop people lives. help prevent, monitor, treat patients' health. order diagnose tools easy patients access, painless, low-cost. may provide solution. discuss link changes types long-term that, together, kill most population: (1) (affecting heart and/or blood). (2) Cancer (abnormal growth cells). (3) Neurodegenerative brain nervous system). (4) Metabolic (problems storing, accessing, using body's fuel). show leaves tell-tale signs lots different parts eye. Signs mostly found back eye, cancer be fluid. seen them tell us what is. believe understand more about how detect it if combine information within develop new these

Language: Английский

---

Multi-omic spatial effects on high-resolution AI-derived retinal thickness

Nature Communications, Journal Year: 2025, Volume and Issue: 16(1)

Published: Feb. 4, 2025

Abstract Retinal thickness is a marker of retinal health and more broadly, seen as promising biomarker for many systemic diseases. measurements are procured from optical coherence tomography (OCT) part routine clinical eyecare. We processed the UK Biobank OCT images using convolutional neural network to produce fine-scale across > 29,000 points in macula, retina responsible human central vision. The macula disproportionately affected by high disease burden disorders such age-related macular degeneration diabetic retinopathy, which both involve metabolic dysregulation. Analysis common genomic variants, metabolomic, blood immune biomarkers, PheCodes genetic scores grid, reveals multiple novel loci including four on X chromosome; thinning associated with sclerosis; associations correlated metabolites that cluster spatially retina. highlight parafoveal be particularly susceptible insults. These results demonstrate gains discovery power resolution achievable AI-leveraged analysis. Results accessible bespoke web interface gives full control pursue findings.

Language: Английский

---

Artificial intelligence-enhanced retinal imaging as a biomarker for systemic diseases

Theranostics, Journal Year: 2025, Volume and Issue: 15(8), P. 3223 - 3233

Published: Feb. 18, 2025

Retinal images provide a non-invasive and accessible means to directly visualize human blood vessels nerve fibers. Growing studies have investigated the intricate microvascular neural circuitry within retina, its interactions with other systemic vascular nervous systems, link between retinal biomarkers various diseases. Using eye study health, based on these connections, has been given term as oculomics. Advancements in artificial intelligence (AI) technologies, particularly deep learning, further increased potential impact of this study. Leveraging analysis demonstrated potentials detecting numerous diseases, including cardiovascular central system chronic kidney metabolic endocrine disorders, hepatobiliary AI-based imaging, which incorporates established modalities such digital color fundus photographs, optical coherence tomography (OCT) OCT angiography, well emerging technologies like ultra-wide field shows great promises predicting This provides valuable opportunity for diseases screening, early detection, prediction, risk stratification, personalized prognostication. As AI big data research grows, mission transforming healthcare, they also face challenges limitations both technology. The application natural language processing framework, large model, generative techniques presents opportunities concerns that require careful consideration. In review, we not only summarize key AI-enhanced imaging but underscore significance advancements healthcare. By highlighting remarkable progress made thus far, comprehensive overview state-of-the-art explore rapidly evolving field. review aims serve resource researchers clinicians, guiding future fostering integration clinical practice.

Language: Английский

---

High-resolution fundus images for ophthalmomics and early cardiovascular disease prediction

Scientific Data, Journal Year: 2025, Volume and Issue: 12(1)

Published: April 3, 2025

Cardiovascular diseases (CVDs) remain the foremost cause of mortality globally, emphasizing imperative for early detection to improve patient outcomes and mitigate healthcare burdens. Carotid intima-media thickness (CIMT) serves as a well-established predictive marker atherosclerosis cardiovascular risk assessment. Fundus imaging offers non-invasive modality investigate microvascular pathology systemic vascular health. However, paucity high-quality, publicly available datasets linking fundus images with CIMT measurements has hindered progression AI-driven models CVDs. Addressing this gap, we introduce China-Fundus-CIMT dataset, comprising bilateral high-resolution images, measurements, clinical data-including age gender-from 2,903 patients. Our experiments multimodal reveal that integrating information substantially enhances performance, yielding AUC-ROC increases 3.22% 7.83% on validation test sets, respectively, compared unimodal models. This dataset constitutes vital resource developing validating AI-based screening CVDs using is now accessible research community.

Language: Английский

---

Retina Fundus Photograph-Based Artificial Intelligence Algorithms in Medicine: A Systematic Review

Ophthalmology and Therapy, Journal Year: 2024, Volume and Issue: 13(8), P. 2125 - 2149

Published: June 24, 2024

We conducted a systematic review of research in artificial intelligence (AI) for retinal fundus photographic images. highlighted the use various AI algorithms, including deep learning (DL) models, application ophthalmic and non-ophthalmic (i.e., systemic) disorders. found that algorithms interpretation images, compared to clinical data physician experts, represents an innovative solution with demonstrated superior accuracy identifying many (e.g., diabetic retinopathy (DR), age-related macular degeneration (AMD), optic nerve disorders), disorders dementia, cardiovascular disease). There has been significant amount imaging this research, leading potential incorporation DL automated analysis. transform healthcare by improving accuracy, speed, workflow, lowering cost, increasing access, reducing mistakes, transforming worker education training.

Language: Английский

---

A Systematic Review and Meta-Analysis of Applying Deep Learning in the Prediction of the Risk of Cardiovascular Diseases From Retinal Images

Translational Vision Science & Technology, Journal Year: 2023, Volume and Issue: 12(7), P. 14 - 14

Published: July 13, 2023

Purpose: The purpose of this study was to perform a systematic review and meta-analysis synthesize evidence from studies using deep learning (DL) predict cardiovascular disease (CVD) risk retinal images. Methods: A literature search performed in MEDLINE, Scopus, Web Science up June 2022. We extracted data pertaining predicted outcomes, model development, validation performance metrics. Included were graded the Quality Assessment Diagnostic Accuracies Studies 2 tool. Model pooled across eligible random-effects model. Results: total 26 included analysis. There 42 CVD risk-related outcomes images identified, including 33 factors, 4 cardiac imaging biomarkers, scores, presence CVD, incident CVD. Three that aimed development future events reported an area under receiver operating curve (AUROC) between 0.68 0.81. Models used as input had mean absolute error 3.19 years (95% confidence interval [CI] = 2.95–3.43) for age prediction; AUROC 0.96 CI 0.95–0.97) gender classification; 0.80 0.73–0.86) diabetes detection; 0.86 0.81–0.92) detection chronic kidney disease. observed high level heterogeneity variation designs. Conclusions: Although DL models appear have reasonably good when it comes predicting risk, further work is necessary evaluate real-world applicability predictive accuracy. Translational Relevance: DL-based assessment holds great promise be translated clinical practice novel approach assessment, given its simple, quick, noninvasive nature.

Language: Английский

---

Retinal Vein Occlusion–Background Knowledge and Foreground Knowledge Prospects—A Review

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(13), P. 3950 - 3950

Published: July 5, 2024

Thrombosis of retinal veins is one the most common vascular diseases that may lead to blindness. The latest epidemiological data leave no illusions burden on healthcare system, as impacted by patients with this diagnosis, will increase worldwide. This obliges scientists search for new therapeutic and diagnostic options. In 21st century, there has been tremendous progress in imaging techniques, which facilitated a better understanding mechanisms related development vein occlusion (RVO) its complications, consequently enabled introduction treatment methods. Moreover, artificial intelligence (AI) likely assist selecting best option near future. aim comprehensive review re-evaluate old but still relevant RVO confront them studies. paper provide detailed overview current treatment, prevention, future possibilities regarding RVO, well clarifying mechanism macular edema disease entity.

Language: Английский

---

Automatic artery/vein classification methods for retinal blood vessel: A review

Computerized Medical Imaging and Graphics, Journal Year: 2024, Volume and Issue: 113, P. 102355 - 102355

Published: Feb. 16, 2024

Language: Английский

---