Guillaume Chassagnon’s research while affiliated with Université Paris Cité and other places

... Barat et al. [15] explored the main research initiatives and technological advances shaping the landscape of radiology in France, including new developments in AI applications. ...

Reference:

AI and Interventional Radiology: A Narrative Review of Reviews on Opportunities, Challenges, and Future Directions

... • Anti-U1-RNP antibodies: U1-RNP is a component of the spliceosome, a RNP that is responsible for splicing, a critical process in gene expression where introns (non-coding regions) are removed from a pre-messenger RNA (pre-mRNA) transcript, and exons (coding regions) are joined together. In SSc, they are found in 8 % of patients (14 % in lcSSc, 3 % in dcSSc), especially with Afro-Caribbean genetic background [47], and are associated with lcSSc [48], ILD, muscular and kidney involvement [47]. Anti-U1-RNP positive patients may have a poorer survival than other SSc patients [47]. ...

Reference:

Autoantibodies in Systemic Sclerosis: From Disease Bystanders to Pathogenic Players

... Furthermore, a reduction in the frequency of pulmonary exacerbations and sweat chloride concentration has been well described [13][14][15]20,21]. However, the impact of ETI on pulmonary structural disease assessed using CT lung scores has been addressed in only a limited number of publications to date [16,17,[22][23][24]. To the best of our knowledge, ours is one of the largest series described in the literature on this topic using the modified Bhalla score. ...

Reference:

Evolution of Lung Disease Studied by Computed Tomography in Adults with Cystic Fibrosis Treated with Elexacaftor/Tezacaftor/Ivacaftor

... Augmenting image interpretation AI has rapidly evolved into a powerful tool for enhancing radiologists' diagnostic capabilities ( Fig. 1: imaging interpretation, reducing errors). A study to assess whether AI improves the detection accuracy of abnormalities on chest radiographs by radiologists with varying levels of expertise found that AI assistance led to an increase in sensitivity ranging from 6% to 26% (P < 0.001) for all readers, including thoracic radiologists, general radiologists, and radiology residents 64 . ...

Reference:

AI solutions to the radiology workforce shortage

... However, the incidence of lung cancer in non-smokers still exists and may be increasing in certain regions and populations [6][7][8]. The research by REVEL M and colleagues indicates that in most European countries, it is anticipated that the mortality rate from lung cancer in females will surpass that of breast cancer [9]. Therefore, given these facts, there is a growing focus on cancer risk factors other than smoking, such as viral infections, chronic inflammation, genetic variants, and environmental exposures. ...

Reference:

No genetic causal association between human papillomavirus and lung cancer risk: a bidirectional two-sample Mendelian randomization analysis

... on the test set. 62 Liu et al highlight the practical utility of automatic clot volumetry for estimation of the QI, reporting a measuring time of 12.9 seconds for the AI model compared with 9.8 minutes for manual calculation with a correlation coefficient (r) reaching .825. 58 Lanza et al also trained a CNN to determine blood clot volume, using this parameter as an indicator of PE presence or absence. ...

Reference:

Applications of Artificial Intelligence in Acute Thoracic Imaging

... In the first stage, we build a prediction model for bidimensional SOS slices using an image-based classifier. For building the image-based classifier, we selected slices from the VBP of the cases group that included the breast lesion, and the corresponding slice of the VBP in the controls group to train a multivariate logistic regressor with sequential forward feature selection, following the approach outlined by Decoux et al., 28 which provides a detailed review of different ML and feature selection methods for radiomics analysis and supports the selection of logistic regressor as a strong alternative. We extract a total of 33 radiomic features from each slice. ...

Reference:

Breast cancer detection from ultrasound computed tomography imaging using radiomic analysis: in silico trial

... Similar results were also noted in studies by Holden [39] & Oropesa [40] while Kumar [41] developed an objective framework for robotic surgery training, offering valuable feedback. Chassagnon [98] 2023; Learning from the machine: AI assistance is not an effective learning tool for resident education in chest x-ray interpretation; France To assess whether a computer-aided detection (CADe) system could serve as a learning tool for radiology residents in chest X-ray (CXR) interpretation. GPT-4 also demonstrated higher stability (mean correlation 0.83) and confidence (45% accuracy) compared to GPT-3.5 and PaLM2. ...

Reference:

Modern artificial intelligence and large language models in graduate medical education: a scoping review of attitudes, applications & practice

... p = 0.035), and the implementation of AI-assisted reporting could reduce the number of missed (55). The development of AI methods with good diagnostic capabilities for PE can improve the performance and reproducibility of radiological diagnostics (56)(57)(58). ...

Reference:

Research progress of artificial intelligence and machine learning in pulmonary embolism

... Recently O 3 exposure was linked to increased SSc-ILD severity and faster disease progression (23). Elevated PM 2.5 levels are consistently associated with increased risk of death or lung transplantation in patients with fILD in well-characterized cohorts with robust exposure modelling (24,25). ...

Reference:

Impact of Environmental Exposures on the Development and Progression of Fibrotic Interstitial Lung Disease