The Management of Persistent Distal Occlusions after Mechanical Thrombectomy and Thrombolysis: An Inter- and Intrarater Agreement Study.

BACKGROUND AND PURPOSE: The best management of patients with persistent distal occlusion after mechanical thrombectomy with or without IV thrombolysis remains unknown. We sought to evaluate the variability and agreement in decision-making for persistent distal occlusions. MATERIALS AND METHODS: A portfolio of 60 cases was sent to clinicians with varying backgrounds and experience. Responders were asked whether they considered conservative management or rescue therapy (stent retriever, aspiration, or intra-arterial thrombolytics) a treatment option as well as their willingness to enroll patients in a randomized trial. Agreement was assessed using κ statistics. RESULTS: The electronic survey was answered by 31 physicians (8 vascular neurologists and 23 interventional neuroradiologists). Decisions for rescue therapies were more frequent (n = 1116/1860, 60%) than for conservative management (n = 744/1860, 40%; P < .001). Interrater agreement regarding the final management decision was "slight" (κ = 0.12; 95% CI, 0.09-0.14) and did not improve when subgroups of clinicians were studied according to background, experience, and specialty or when cases were grouped according to the level of occlusion. On delayed re-questioning, 23 of 29 respondents (79.3%) disagreed with themselves on at least 20% of cases. Respondents were willing to offer trial participation in 1295 of 1860 (69.6%) cases. CONCLUSIONS: Individuals did not agree regarding the best management of patients with persistent distal occlusion after mechanical thrombectomy and IV thrombolysis. There is sufficient uncertainty to justify a dedicated randomized trial.

Weakly Supervised Learning using Attention gates for colon cancer histopathological image segmentation.

Recently, Artificial Intelligence namely Deep Learning methods have revolutionized a wide range of domains and applications. Besides, Digital Pathology has so far played a major role in the diagnosis and the prognosis of tumors. However, the characteristics of the Whole Slide Images namely the gigapixel size, high resolution and the shortage of richly labeled samples have hindered the efficiency of classical Machine Learning methods. That goes without saying that traditional methods are poor in generalization to different tasks and data contents. Regarding the success of Deep learning when dealing with Large Scale applications, we have resorted to the use of such models for histopathological image segmentation tasks. First, we review and compare the classical UNet and Att-UNet models for colon cancer WSI segmentation in a sparsely annotated data scenario. Then, we introduce novel enhanced models of the Att-UNet where different schemes are proposed for the skip connections and spatial attention gates positions in the network. In fact, spatial attention gates assist the training process and enable the model to avoid irrelevant feature learning. Alternating the presence of such modules namely in our Alter-AttUNet model adds robustness and ensures better image segmentation results. In order to cope with the lack of richly annotated data in our AiCOLO colon cancer dataset, we suggest the use of a multi-step training strategy that also deals with the WSI sparse annotations and unbalanced class issues. All proposed methods outperform state-of-the-art approaches but Alter-AttUNet generates the best compromise between accurate results and light network. The model achieves 95.88% accuracy with our sparse AiCOLO colon cancer datasets. Finally, to evaluate and validate our proposed architectures we resort to publicly available WSI data: the NCT-CRC-HE-100K, the CRC-5000 and the Warwick colon cancer histopathological dataset. Respective accuracies of 99.65%, 99.73% and 79.03% were reached. A comparison with state-of-art approaches is established to view and compare the key solutions for histopathological image segmentation.

Prospective assessment of aneurysmal rupture risk scores in patients with subarachnoid hemorrhage: a multicentric cohort.

PURPOSE: The natural evolution of unruptured intracranial aneurysms (UIA) is indeed difficult to predict at the individual level. OBJECTIVE: In a large prospective multicentric European cohort, we aimed to evaluate whether the PHASES, UCAS, and ELPASS scores in patients with aneurysmal subarachnoid hemorrhage would have predicted a high risk of aneurysmal rupture or growth. METHODS: Academic centers treating patients with intracranial aneurysms were invited to prospectively collect de-identified data from all patients admitted at their institution for a subarachnoid hemorrhage-related to intracranial aneurysmal rupture between January 1 and March 31, 2021 through a trainee-led research collaborative network. Each responding center was provided with an electronic case record form (CRF) which collected all the elements of the PHASES, ELAPSS, and UCAS scores. RESULTS: A total of 319 patients with aneurysmal subarachnoid hemorrhage were included at 17 centers during a 3-month period. One hundred eighty-three aneurysms (57%) were less than 7 mm. The majority of aneurysms were located on the anterior communicating artery (n = 131, 41%). One hundred eighty-four patients (57%), 103 patients (32%), and 58 (18%) were classified as having a low risk of rupture or growth, according to the PHASES, UCAS, and ELAPSS scores, respectively. CONCLUSION: In a prospective study of European patients with aneurysmal subarachnoid hemorrhage, we showed that 3 common risk-assessment tools designed for patients with unruptured intracranial aneurysms would have not identified most patients to be at high or intermediate risk for rupture, questioning their use for decision-making in the setting of unruptured aneurysms.

Parent Artery Straightening after Flow-Diverter Stenting Improves the Odds of Aneurysm Occlusion.

BACKGROUND AND PURPOSE: Intracranial stents for the treatment of aneurysms can be responsible for parent artery straightening, a phenomenon with potential consequences for aneurysmal occlusion. We aimed to evaluate parent artery straightening following flow-diverter stent placement in patients with intracranial aneurysms and explored the association between parent artery straightening and subsequent aneurysm occlusion. MATERIALS AND METHODS: All patients treated with flow-diverter stents for anterior circulation aneurysms located downstream from the carotid siphon between January 2009 and January 2018 were screened for inclusion. Parent artery straightening was defined as the difference (α-ß) in the parent artery angle at the neck level before (α angle) and after flow-diverter stent deployment (ß angle). We analyzed the procedural and imaging factors associated with parent artery straightening and the associations between parent artery straightening and aneurysmal occlusion. RESULTS: Ninety-five patients met the inclusion criteria (n = 64/95 women, 67.4%; mean age, 54.1 [SD, 11.2] years) with 97 flow-diverter stents deployed for 99 aneurysms. Aneurysms were predominantly located at the MCA bifurcation (n = 44/95, 44.4%). Parent artery straightening was found to be more pronounced in patients treated with cobalt chromium stents than with nitinol stents (P = .02). In multivariate analysis, parent artery straightening (P = .04) was independently associated with aneurysm occlusion after flow-diverter stent deployment. CONCLUSIONS: The use of flow-diverter stents for distal aneurysms induces a measurable parent artery straightening, which is associated with higher occlusion rates. Parent artery straightening, in our sample, appeared to be more prominent with cobalt chromium stents than with nitinol stents. This work highlights the necessary trade-off between navigability and parent artery straightening and may help tailor the selection of flow-diverter stents to aneurysms and parent artery characteristics.

Deep learning for colon cancer histopathological images analysis.

Nowadays, digital pathology plays a major role in the diagnosis and prognosis of tumours. Unfortunately, existing methods remain limited when faced with the high resolution and size of Whole Slide Images (WSIs) coupled with the lack of richly annotated datasets. Regarding the ability of the Deep Learning (DL) methods to cope with the large scale applications, such models seem like an appealing solution for tissue classification and segmentation in histopathological images. This paper focuses on the use of DL architectures to classify and highlight colon cancer regions in a sparsely annotated histopathological data context. First, we review and compare state-of-the-art Convolutional Neural networks (CNN) including the AlexNet, vgg, ResNet, DenseNet and Inception models. To cope with the shortage of rich WSI datasets, we have resorted to the use of transfer learning techniques. This strategy comes with the hallmark of relying on a large size computer vision dataset (ImageNet) to train the network and generate a rich collection of learnt features. The testing and evaluation of such models on our AiCOLO colon cancer dataset ensure accurate patch-level classification results reaching up to 96.98% accuracy rate with ResNet. The CNN models have also been tested and evaluated with the CRC-5000, nct-crc-he-100k and merged datasets. ResNet respectively achieves 96.77%, 99.76% and 99.98% for the three publicly available datasets. Then, we present a pixel-wise segmentation strategy for colon cancer WSIs through the use of both UNet and SegNet models. We introduce a multi-step training strategy as a remedy for the sparse annotation of histopathological images. UNet and SegNet are used and tested in different training scenarios including data augmentation and transfer learning and ensure up to 76.18% and 81.22% accuracy rates. Besides, we test our training strategy and models on the CRC-5000, nct-crc-he-100k and Warwick datasets. Respective accuracy rates of 98.66%, 99.12% and 78.39% were achieved by SegNet. Finally, we analyze the existing models to discover the most suitable network and the most effective training strategy for our colon tumour segmentation case study.1.

Noninvasive Angiographic Results of Clipped or Coiled Intracranial Aneurysms: An Inter- and Intraobserver Reliability Study.

BACKGROUND AND PURPOSE: Noninvasive angiography is commonly used to assess the outcome of surgical or endovascular treatment of intracranial aneurysms in clinical series or randomized trials. We sought to assess whether a standardized 3-grade classification system could be reliably used to compare the CTA and MRA results of both treatments. MATERIALS AND METHODS: An electronic portfolio composed of CTAs of 30 clipped and MRAs of 30 coiled aneurysms was independently evaluated by 24 raters of diverse experience and training backgrounds. Twenty raters performed a second evaluation 1 month later. Raters were asked which angiographic grade and management decision (retreatment; close or long-term follow-up) would be most appropriate for each case. Agreement was analyzed using the Krippendorff α (αK) statistic, and the relationship between angiographic grade and clinical management choice, using the Fisher exact and Cramer V tests. RESULTS: Interrater agreement was substantial (αK = 0.63; 95% CI, 0.55-0.70); results were slightly better for MRA results of coiling (αK = 0.69; 95% CI, 0.56-0.76) than for CTA results of clipping (αK = 0.58; 95% CI, 0.44-0.69). Intrarater agreement was substantial to almost perfect. Interrater agreement regarding clinical management was moderate for both clipped (αK = 0.49; 95% CI, 0.32-0.61) and coiled subgroups (αK = 0.47; 95% CI, 0.34-0.54). The choice of clinical management was strongly associated with the size of the residuum (mean Cramer V = 0.77 [SD, 0.14]), but complete occlusions (grade 1) were followed more closely after coiling than after clipping (P = .01). CONCLUSIONS: A standardized 3-grade scale was found to be a reliable and clinically meaningful tool to compare the results of clipping and coiling of aneurysms using CTA or MRA.

Advanced multiparametric magnetic resonance imaging of multinodular and vacuolating neuronal tumor.

BACKGROUND AND PURPOSE: Multinodular and vacuolating neuronal tumor (MVNT) of the cerebrum is a rare brain lesion with suggestive imaging features. The aim of our study was to report the largest series of MVNTs so far and to evaluate the utility of advanced multiparametric magnetic resonance (MR) techniques. METHODS: This multicenter retrospective study was approved by our institutional research ethics board. From July 2014 to May 2019, two radiologists read in consensus the MR examinations of patients presenting with a lesion suggestive of an MVNT. They analyzed the lesions' MR characteristics on structural images and advanced multiparametric MR imaging. RESULTS: A total of 64 patients (29 women and 35 men, mean age 44.2 ± 15.1 years) from 25 centers were included. Lesions were all hyperintense on fluid-attenuated inversion recovery and T2-weighted imaging without post-contrast enhancement. The median relative apparent diffusion coefficient on diffusion-weighted imaging was 1.13 [interquartile range (IQR), 0.2]. Perfusion-weighted imaging showed no increase in perfusion, with a relative cerebral blood volume of 1.02 (IQR, 0.05) and a relative cerebral blood flow of 1.01 (IQR, 0.08). MR spectroscopy showed no abnormal peaks. Median follow-up was 2 (IQR, 1.2) years, without any changes in size. CONCLUSIONS: A comprehensive characterization protocol including advanced multiparametric magnetic resonance imaging sequences showed no imaging patterns suggestive of malignancy in MVNTs. It might be useful to better characterize MVNTs.

[Do artificial intelligence systems reason in the same way as clinicians when making diagnoses?] / Une intelligence artificielle raisonne-t-elle de la même façon que les cliniciens pour poser des diagnostics ?

Clinical reasoning is at the heart of physicians' competence, as it allows them to make diagnoses. However, diagnostic errors are common, due to the existence of reasoning biases. Artificial intelligence is undergoing unprecedented development in this context. It is increasingly seen as a solution to improve the diagnostic performance of physicians, or even to perform this task for them, in a totally autonomous and more efficient way. In order to understand the challenges associated with the development of artificial intelligence, it is important to understand how the machine works to make diagnoses, what are the similarities and differences with the physician's diagnostic reasoning, and what are the consequences for medical training and practice.

In-silico insights on the prognostic potential of immune cell infiltration patterns in the breast lobular epithelium.

Scattered inflammatory cells are commonly observed in mammary gland tissue, most likely in response to normal cell turnover by proliferation and apoptosis, or as part of immunosurveillance. In contrast, lymphocytic lobulitis (LLO) is a recurrent inflammation pattern, characterized by lymphoid cells infiltrating lobular structures, that has been associated with increased familial breast cancer risk and immune responses to clinically manifest cancer. The mechanisms and pathogenic implications related to the inflammatory microenvironment in breast tissue are still poorly understood. Currently, the definition of inflammation is mainly descriptive, not allowing a clear distinction of LLO from physiological immunological responses and its role in oncogenesis remains unclear. To gain insights into the prognostic potential of inflammation, we developed an agent-based model of immune and epithelial cell interactions in breast lobular epithelium. Physiological parameters were calibrated from breast tissue samples of women who underwent reduction mammoplasty due to orthopedic or cosmetic reasons. The model allowed to investigate the impact of menstrual cycle length and hormone status on inflammatory responses to cell turnover in the breast tissue. Our findings suggested that the immunological context, defined by the immune cell density, functional orientation and spatial distribution, contains prognostic information previously not captured by conventional diagnostic approaches.