Survey of Explainable AI Techniques in Healthcare

Sensors, Journal Year: 2023, Volume and Issue: 23(2), P. 634 - 634

Published: Jan. 5, 2023

Artificial intelligence (AI) with deep learning models has been widely applied in numerous domains, including medical imaging and healthcare tasks. In the field, any judgment or decision is fraught risk. A doctor will carefully judge whether a patient sick before forming reasonable explanation based on patient's symptoms and/or an examination. Therefore, to be viable accepted tool, AI needs mimic human interpretation skills. Specifically, explainable (XAI) aims explain information behind black-box model of that reveals how decisions are made. This paper provides survey most recent XAI techniques used related applications. We summarize categorize types, highlight algorithms increase interpretability topics. addition, we focus challenging problems applications provide guidelines develop better interpretations using concepts image text analysis. Furthermore, this future directions guide developers researchers for prospective investigations clinical topics, particularly imaging.

Language: Английский

---

Deep Learning With Radiomics for Disease Diagnosis and Treatment: Challenges and Potential

Frontiers in Oncology, Journal Year: 2022, Volume and Issue: 12

Published: Feb. 17, 2022

The high-throughput extraction of quantitative imaging features from medical images for the purpose radiomic analysis, i.e., radiomics in a broad sense, is rapidly developing and emerging research field that has been attracting increasing interest, particularly multimodality multi-omics studies. In this context, analysis multidimensional data plays an essential role assessing spatio-temporal characteristics different tissues organs their microenvironment. Herein, recent developments method, including manually defined features, acquisition preprocessing, lesion segmentation, feature extraction, selection dimension reduction, statistical model construction, are reviewed. addition, deep learning-based techniques automatic segmentation being analyzed to address limitations such as rigorous workflow, manual/semi-automatic annotation, inadequate criteria, multicenter validation. Furthermore, summary current state-of-the-art applications technology disease diagnosis, treatment response, prognosis prediction perspective radiology images, histopathology three-dimensional dose distribution data, oncology, presented. potential value diagnostic therapeutic strategies also further analyzed, first time, advances challenges associated with dosiomics radiotherapy summarized, highlighting latest progress radiomics. Finally, robust framework presented recommendations future development discussed, but not limited factors affect stability (medical big multitype expert knowledge medical), data-driven processes (reproducibility interpretability studies, alternatives various institutions, prospective researches clinical trials), thoughts on directions (the capability achieve open platform analysis).

Language: Английский

---

A Fully Automated Multimodal MRI-Based Multi-Task Learning for Glioma Segmentation and IDH Genotyping

IEEE Transactions on Medical Imaging, Journal Year: 2022, Volume and Issue: 41(6), P. 1520 - 1532

Published: Jan. 26, 2022

The accurate prediction of isocitrate dehydrogenase (IDH) mutation and glioma segmentation are important tasks for computer-aided diagnosis using preoperative multimodal magnetic resonance imaging (MRI). two ongoing challenges due to the significant inter-tumor intra-tumor heterogeneity. existing methods address them mostly based on single-task approaches without considering correlation between tasks. In addition, acquisition IDH genetic labels is expensive costly, resulting in a limited number data modeling. To comprehensively these problems, we propose fully automated MRI-based multi-task learning framework simultaneous genotyping. Specifically, task heterogeneity tackled with hybrid CNN-Transformer encoder that consists convolutional neural network transformer extract shared spatial global information learned from decoder multi-scale classifier Then, loss designed balance by combining classification functions uncertain weights. Finally, an uncertainty-aware pseudo-label selection proposed generate pseudo-labels larger unlabeled improving accuracy genotyping semi-supervised learning. We evaluate our method multi-institutional public dataset. Experimental results show achieves promising performance outperforms counterparts other state-of-the-art methods. With introduction data, further improves source codes publicly available at https://github.com/miacsu/MTTU-Net.git .

Language: Английский

---

Role of Artificial Intelligence in Radiogenomics for Cancers in the Era of Precision Medicine

Cancers, Journal Year: 2022, Volume and Issue: 14(12), P. 2860 - 2860

Published: June 9, 2022

Radiogenomics, a combination of “Radiomics” and “Genomics,” using Artificial Intelligence (AI) has recently emerged as the state-of-the-art science in precision medicine, especially oncology care. Radiogenomics syndicates large-scale quantifiable data extracted from radiological medical images enveloped with personalized genomic phenotypes. It fabricates prediction model through various AI methods to stratify risk patients, monitor therapeutic approaches, assess clinical outcomes. shown tremendous achievements prognosis, treatment planning, survival prediction, heterogeneity analysis, reoccurrence, progression-free for human cancer study. Although immense performance care aspects, it several challenges limitations. The proposed review provides an overview radiogenomics viewpoints on role terms its promises computational well oncological aspects offers opportunities era medicine. also presents recommendations diminish these obstacles.

Language: Английский

---

Artificial Intelligence in Brain Tumor Imaging: A Step toward Personalized Medicine

Current Oncology, Journal Year: 2023, Volume and Issue: 30(3), P. 2673 - 2701

Published: Feb. 22, 2023

The application of artificial intelligence (AI) is accelerating the paradigm shift towards patient-tailored brain tumor management, achieving optimal onco-functional balance for each individual. AI-based models can positively impact different stages diagnostic and therapeutic process. Although histological investigation will remain difficult to replace, in near future radiomic approach allow a complementary, repeatable non-invasive characterization lesion, assisting oncologists neurosurgeons selecting best option correct molecular target chemotherapy. AI-driven tools are already playing an important role surgical planning, delimiting extent lesion (segmentation) its relationships with structures, thus allowing precision surgery as radical reasonably acceptable preserve quality life. Finally, AI-assisted prediction complications, recurrences response, suggesting most appropriate follow-up. Looking future, AI-powered promise integrate biochemical clinical data stratify risk direct patients personalized screening protocols.

Language: Английский

---

Multimodal data integration for oncology in the era of deep neural networks: a review

Frontiers in Artificial Intelligence, Journal Year: 2024, Volume and Issue: 7

Published: July 25, 2024

Cancer research encompasses data across various scales, modalities, and resolutions, from screening diagnostic imaging to digitized histopathology slides types of molecular clinical records. The integration these diverse for personalized cancer care predictive modeling holds the promise enhancing accuracy reliability screening, diagnosis, treatment. Traditional analytical methods, which often focus on isolated or unimodal information, fall short capturing complex heterogeneous nature data. advent deep neural networks has spurred development sophisticated multimodal fusion techniques capable extracting synthesizing information disparate sources. Among these, Graph Neural Networks (GNNs) Transformers have emerged as powerful tools learning, demonstrating significant success. This review presents foundational principles learning including oncology taxonomy strategies. We delve into recent advancements in GNNs oncology, spotlighting key studies their pivotal findings. discuss unique challenges such heterogeneity complexities, alongside opportunities it a more nuanced comprehensive understanding cancer. Finally, we present some latest pan-cancer By surveying landscape our goal is underline transformative potential Transformers. Through technological methodological innovations presented this review, aim chart course future promising field. may be first that highlights current state applications using transformers, sources, sets stage evolution, encouraging further exploration care.

Language: Английский

---

Clustering Functional Magnetic Resonance Imaging Time Series in Glioblastoma Characterization: A Review of the Evolution, Applications, and Potentials

Brain Sciences, Journal Year: 2024, Volume and Issue: 14(3), P. 296 - 296

Published: March 20, 2024

In this paper, we discuss how the clustering analysis technique can be applied to analyze functional magnetic resonance imaging (fMRI) time-series data in context of glioblastoma (GBM), a highly heterogeneous brain tumor. The precise characterization GBM is challenging and requires advanced analytical approaches. We have synthesized existing literature provide an overview algorithms help identify unique patterns within dynamics GBM. Our review shows that fMRI time series has great potential for improving differentiation between various subtypes GBM, which pivotal developing personalized therapeutic strategies. Moreover, method proves effective capturing temporal changes occurring enhancing monitoring disease progression response treatment. By thoroughly examining consolidating current research, paper contributes understanding techniques refine This article emphasizes importance incorporating cutting-edge into neuroimaging neuro-oncology research. providing detailed perspective, approach may guide future investigations boost development tailored strategies

Language: Английский

---

Clinical management and survival outcomes of patients with different molecular subtypes of diffuse gliomas in China (2011–2017): a multicenter retrospective study from CGGA

Cancer Biology and Medicine, Journal Year: 2022, Volume and Issue: 19(10), P. 1460 - 1476

Published: Nov. 1, 2022

Objective: We aimed to summarize the clinicopathological characteristics and prognostic features of various molecular subtypes diffuse gliomas (DGs) in Chinese population. Methods: In total, 1,418 patients diagnosed with DG between 2011 2017 were classified into 5 according 2016 WHO classification central nervous system tumors. The IDH mutation status was determined by immunohistochemistry and/or DNA sequencing, 1p/19q codeletion detected fluorescence situ hybridization. median clinical follow-up time 1,076 days. T-tests chi-square tests used compare characteristics. Kaplan-Meier Cox regression methods evaluate factors. Results: Our cohort included 15.5% lower-grade gliomas, IDH-mutant 1p/19q-codeleted (LGG-IDHm-1p/19q); 18.1% (LGG-IDHm); 13.1% IDH-wildtype (LGG-IDHwt); 36.1% glioblastoma, (GBM-IDHwt); 17.2% (GBM-IDHm). Approximately 63.3% enrolled primary overall survival times for LGG-IDHm, LGG-IDHwt, GBM-IDHwt, GBM-IDHm 75.97, 34.47, 11.57, 15.17 months, respectively. 5-year rate LGG-IDHm-1p/19q 76.54%. observed a significant association high resection favorable outcomes across all also role chemotherapy prolonging GBM-IDHwt GBM-IDHm, post-relapse 2 recurrent GBM subtypes. Conclusions: By controlling subtypes, we found that factors associated DG.

Language: Английский

---

Brain Tumor Characterization Using Radiogenomics in Artificial Intelligence Framework

Cancers, Journal Year: 2022, Volume and Issue: 14(16), P. 4052 - 4052

Published: Aug. 22, 2022

Brain tumor characterization (BTC) is the process of knowing underlying cause brain tumors and their characteristics through various approaches such as segmentation, classification, detection, risk analysis. The substantial includes identification molecular signature useful genomes whose alteration causes tumor. radiomics approach uses radiological image for disease by extracting quantitative features in artificial intelligence (AI) environment. However, when considering a higher level genetic information mutation status, combined study “radiomics genomics” has been considered under umbrella “radiogenomics”. Furthermore, AI radiogenomics’ environment offers benefits/advantages finalized outcome personalized treatment individualized medicine. proposed summarizes tumor’s prospect an emerging field research, i.e., radiogenomics environment, with help statistical observation risk-of-bias (RoB) PRISMA search was used to find 121 relevant studies review using IEEE, Google Scholar, PubMed, MDPI, Scopus. Our findings indicate that both have successfully applied aggressively several oncology applications numerous advantages. paradigm, conventional deep made impact on favorable outcomes BTC. analysis better understanding architectures stronger benefits providing bias involved them.

Language: Английский

---

Artificial Intelligence and Precision Medicine: A New Frontier for the Treatment of Brain Tumors

Life, Journal Year: 2022, Volume and Issue: 13(1), P. 24 - 24

Published: Dec. 22, 2022

Brain tumors are a widespread and serious neurological phenomenon that can be life- threatening. The computing field has allowed for the development of artificial intelligence (AI), which mimic neural network human brain. One use this technology been to help researchers capture hidden, high-dimensional images brain tumors. These provide new insights into nature improve treatment options. AI precision medicine (PM) converging revolutionize healthcare. potential cancer imaging interpretation in several ways, including more accurate tumor genotyping, precise delineation volume, better prediction clinical outcomes. AI-assisted surgery an effective safe option treating This review discusses various PM techniques used treatment. tumors, i.e., genomic profiling, microRNA panels, quantitative imaging, radiomics, hold great promise future. However, there challenges must overcome these technologies reach their full

Language: Английский

---