Meta-Radiology最新文献

Radiation therapy (RT) is currently the main clinical treatment of tumors. Before treatment initiation, precise delineation of the planned target volume (PTV) and organs at risk (OAR) is essential. This segmentation, together with the dose prediction algorithm, aids in the calculation and evaluation of the dose distribution, and ultimately in the refinement of the treatment plan. To provide a comprehensive view of the current landscape of research on dose prediction methods, we meticulously collected and summarized papers published between 2017 and 2023. First, we present our rigorous literature search approach, providing a statistical analysis of the pooled papers and an elaborate overview of the evaluation metrics that are commonly and consistently employed in this domain. Then, we focus on a detailed survey of the evolutionary trajectories of dose prediction methods. This comprehensive investigation covers a spectrum ranging from traditional Knowledge-Based Planning (KBP) methods to emerging deep learning-based methods, which include input improvement methods, U-Net-based methods, GAN-based methods, and other deep learning-based methods. Throughout this exposition, we have carefully outlined the strengths and limitations inherent in these various approaches. Finally, we conclude with a summary of the primary challenges facing the field and propose several prospective research directions to effectively address them.

The high incidence of mild traumatic brain injury (mTBI) and the associated post-concussion symptoms, such as headache and cognitive deficits, have captured the significant attention from researchers globally. Magnetic resonance spectroscopy (MRS), a non-invasively technique derived from Magnetic Resonance Imaging (MRI), provides a complement approach to investigating brain metabolites as biomarkers for in vivo pathophysiological changes following mTBI, which are not evident in traditional MRI or CT scans. However, the separate review of MRS in mTBI patients has been limited, given the myriad factors involved and wide spectrum of TBI severity. In this review, we first delve into metabolite changes after mTBI, highlighting a reduction in N-acetyl-aspartate (NAA) as a relatively stable marker associated with neuronal loss or disfunction following mTBI. We then discuss the varying results observed for different metabolites and enumerate possible factors contributing to these inconsistent findings. These factors include variations in experimental methods, such as scanner types, acquisition methods, and region of interest. Additionally, we address subjects-specific factors, such as occupation, cause of injury, control group selection, injury stage, severity, the number of traumatic events, and the assessment of clinical features. Finally, we discuss the trend for future research in this field.

Alzheimer's disease (AD) is a common genetically related cognitive disorder. Studies have shown that ferroptosis, N⁶-Methyladenosine (M6A) and immune checkpoint are related to the development of AD. However, the effects of these three gene pathways on AD progression are still unclear. Here, we used genes expressed in the middle temporal gyrus (MTG) to study the differences in ferroptosis, M6A and immune checkpoint-related gene in 97 Alzheimer's disease and 98 normal controls (NC). We then conducted correlation analysis between ferroptosis, M6A and immune checkpoint-related gene expression levels to investigate the relationship between these genes and AD. Compared to the NC, the gene expression from MTG in AD are as follows: (1) in ferroptosis related genes, the expression of CARS, CDKN1A, HSPB1, MT1G, EMC2, SAT1 and SLC1A5 was increased, while the expression of ACSL4, ATP5MC3, CSID1, CS, DPP4, GLS2 and GPX4 was decreased; (2) in M6A-related genes, the expression of HNRNPA2B1, IGF2BP2, RBM15B and YTHDC1 was increased, while the expression of FTO, YTHDC2 and YTHDF2 was decreased; (3) the expression of immune checkpoint-related genes (including CTLA4, HAVCR2 and LAG3) was increased. Further, we determined related gene pathways among these genes by conducting a literature review. By verifying the dataset, we can well verify our results and prove that our results have good robustness. We concluded of gene expression that a complete set of ferroptosis, M6A and immune checkpoint regulatory mechanisms is activated in the MTG during AD development.

Missing data are a common problem for large cohort or longitudinal research and have been handled through data imputation. Based on simplified models such as linear or nonlinear interpolations, current imputation methods may not be accurate for real-life data such as biological and behavioral data. The purpose of this work was to explore the capability of ChatGPT, a powerful Large Language Model (LLM) developed by OpenAI, for biological and psychological data imputation. We tested the feasibility using data from the Human Connectome Project. Performance was evaluated by comparing the imputed data against known ground truth (GT) and measured with metrics like Pearson correlation coefficient (r), relative accuracy (MP), and mean absolute error (MAE). Comparative analyses with traditional imputation techniques are also conducted to demonstrate the superior efficacy of the ChatGPT as a data imputer. In summary, through customized data-to-text prompting engineering, ChatGPT can successfully capture intricate patterns and dependencies within biological data, resulting in precise imputations. Fine-tuning ChatGPT with domain-specific biological vocabulary with human in-loop as an interpreter enhances the accuracy and relevance of the imputations.

Large Language Models (LLMs) have consistently showcased remarkable generalization capa-bilities when applied to various language tasks. Nonetheless, harnessing the full potential of LLMs for Radiology Report Generation (R2Gen) still presents a challenge, stemming from the inherent disparity in modality between LLMs and the R2Gen task. To bridge this gap effectively, we propose R2GenGPT, which is a novel solution that aligns visual features with the word embedding space of LLMs using an efficient visual alignment module. This innovative approach empowers the previously static LLM to seamlessly integrate and process image information, marking a step forward in optimizing R2Gen performance. R2GenGPT offers the following benefits. First, it attains state-of-the-art (SOTA) performance by training only the lightweight visual alignment module while freezing all the parameters of LLM. Second, it exhibits high training efficiency, as it requires the training of an exceptionally minimal number of parameters while achieving rapid convergence. By employing delta tuning, our model only trains 5 ​M parameters (which constitute just 0.07 ​% of the total parameter count) to achieve performance close to the SOTA levels. Our code is available at https://github.com/wang-zhanyu/R2GenGPT.

Visual prompt engineering is a fundamental methodology in the field of visual and image artificial general intelligence. As the development of large vision models progresses, the importance of prompt engineering becomes increasingly evident. Designing suitable prompts for specific visual tasks has emerged as a meaningful research direction. This review aims to summarize the methods employed in the computer vision domain for large vision models and visual prompt engineering, exploring the latest advancements in visual prompt engineering. We present influential large models in the visual domain and a range of prompt engineering methods employed on these models. It is our hope that this review provides a comprehensive and systematic description of prompt engineering methods based on large visual models, offering valuable insights for future researchers in their exploration of this field.

The emergence of artificial general intelligence (AGI) is transforming radiation oncology. As prominent vanguards of AGI, large language models (LLMs) such as GPT-4 and PaLM 2 can process extensive texts and large vision models (LVMs) such as the Segment Anything Model (SAM) can process extensive imaging data to enhance the efficiency and precision of radiation therapy. This paper explores full-spectrum applications of AGI across radiation oncology including initial consultation, simulation, treatment planning, treatment delivery, treatment verification, and patient follow-up. The fusion of vision data with LLMs also creates powerful multimodal models that elucidate nuanced clinical patterns. Together, AGI promises to catalyze a shift towards data-driven, personalized radiation therapy. However, these models should complement human expertise and care. This paper provides an overview of how AGI can transform radiation oncology to elevate the standard of patient care in radiation oncology, with the key insight being AGI's ability to exploit multimodal clinical data at scale.

Malignant tumors have risen to prominence as the leading threat to both life and the overall health of individuals. Precision medicine relies heavily on precise imaging. Among the plethora of imaging techniques, the advantages of dual-energy CT in tumor diagnosis and treatment are becoming increasingly pronounced. Accurate imaging evaluation of tumors involves various aspects, including diagnosis and differential diagnosis, staging and classification, assessment of treatment efficacy, and prediction of prognosis. Notably, dual-energy CT has showcased its unique advantages across these domains. In this review, we commence by offering a succinct overview of the implementation techniques and postprocessing of dual-energy CT. Then, we focus on providing a comprehensive survey of the current application of dual-energy CT in common cancers, such as central nervous system tumors, head and neck tumors, lung tumors, breast tumors, abdominal tumors, and bone tumors. Finally, we discuss the present technical constraints and prospective avenues of dual-energy CT application. As the clinical integration of dual-energy CT increases, its future outlook is poised to be expansive, potentially paving the way for routine application within clinical settings.

Background

It has been revealed that brain gray matter volume (GMV) abnormalities are present in patients with vascular cognitive impairment (VCI). However, the GMV alterations that have been uncovered are highly inconsistent, and their correlation with gene expression profiles is still largely unknown.

Purpose

To establish a correlation between VCI-related GMV alterations and gene expression patterns and uncover potential genetic profiles underlying GMV abnormalities in VCI.

Materials and methods

Here, a quantitative meta-analysis that compared voxel-based GMV between VCI patients and healthy controls (HCs) was carried out on 11 datasets (10 from previous studies and 1 newly collected), comprising 385 VCI individuals and 334 ​HCs, to investigate GMV alterations in VCI patients. Partial least squares regression analysis was then conducted to investigate the relationship between the GMV alterations in VCI and gene expression profiles obtained from Allen Human Brain Atlas database.

Results

Compared with healthy controls, patients with VCI showed consistent decreased GMV which predominantly included the right insula, right Rolandic operculum, right putamen, right superior temporal gyrus, left medial superior frontal gyrus, and right median cingulate and paracingulate gyri. Meta-regression analysis revealed that decreased GMV in left medial superior frontal gyrus was negatively correlated with Mini-Mental State Examination score in VCI. Furthermore, 2835 genes were identified whose expression patterns were correlated with VCI-related GMV changes, and these genes were enriched in distinct biological processes, brain cell types and lifespan windows across brain regions.

Conclusion

Together, these findings could provide the potential neurobiological underpinnings and the genetic substrates underlying GMV abnormalities of VCI.

Objectives

Previous studies showed the clinical effectiveness of elexacaftor-tezacaftor-ivacaftor (ETI) in cystic fibrosis (CF) patients and a recently published study evaluated twelve CF patients that performed chest and sinus computed tomography (CT) examinations and showed that ETI decreased pulmonary and sinus morphological abnormalities after one year of treatment. The aim of the present study was to evaluate the role of CFTR modulator therapy in improving radiological and clinical scores one year after ETI therapy initiation in a wider CF patient population.

Materials and methods

Between January 2020 and December 2022, 44 CF adult patients received elexacaftor-tezacaftor-ivacaftor (ETI) therapy for at least one year and underwent a chest CT examination at our hospital before and one year after ETI therapy initiation. Experienced radiologists who were blinded to the treatment assessed the images in consensus. The Brody-II score (BSII), the Lund-Mackay score (LM score) and the Sheikh-Lind CT sinus disease severity scoring system (SL score) were evaluated. Clinical scores such as cystic fibrosis clinical score (CFCS), Cystic Fibrosis Questionnaire-Revised (CFQ-R) score, the 22-item SinoNasal Outcome Test (SNOT-22) questionnaire and the CF-specific 28-modal abdominal symptom score (CFAbd-Score) were evaluated. Forced expiratory volume in 1 ​s (FEV1) and forced vital capacity (FVC) were also assessed. Paired samples t-tests were used to compare differences before and after one year of ETI therapy initiation, and Pearson's correlation coefficient was used to evaluate changes in FEV1 and total BSII and in FVC and total BSII.

Results

Total BIIS one year after ETI initiation showed statistically significant lower scores (−6.0 p, p ​< ​0.0001). In particular, mucous plugging (−15.8 p, p ​< ​0.0001), peribronchial thickening (−16.2 p, p ​< ​0.0001) and parenchyma (−0.3 p, p ​= ​0.0397) showed statistically significant lower scores. LM score, SL score, FEV1, FVC, CFCS, CFQ-R, SNOT-22 and CFAbd-Score showed statistically significant lower scores one year after ETI initiation (p ​< ​0.0001). The correlation between ΔFEV1 and Δtotal BSII was statistically significant and moderate (r ​= ​−0.5188, p ​= ​0.0003), and the correlation between ΔFVC and Δtotal BSII was statistically significant and weak (r ​= ​−0.3160, p ​= ​0.0367).

Conclusion

Evolution of imaging findings on CT during follow-up closely correlate with improved clinical scores and functional data one year after ETI therapy initiation, indicating that CT may be a useful adjunct during follow-up of CF patients under this treatment as an objective measure of disease improvement.