Nonaneurysmal perimesencephalic subarachnoid hemorrhage on noncontrast head CT: An accuracy, inter-rater, and intra-rater reliability study.

BACKGROUND AND PURPOSE: To evaluate the reliability and accuracy of nonaneurysmal perimesencephalic subarachnoid hemorrhage (NAPSAH) on Noncontrast Head CT (NCCT) between numerous raters. MATERIALS AND METHODS: 45 NCCT of adult patients with SAH who also had a catheter angiography (CA) were independently evaluated by 48 diverse raters; 45 raters performed a second assessment one month later. For each case, raters were asked: 1) whether they judged the bleeding pattern to be perimesencephalic; 2) whether there was blood anterior to brainstem; 3) complete filling of the anterior interhemispheric fissure (AIF); 4) extension to the lateral part of the sylvian fissure (LSF); 5) frank intraventricular hemorrhage; 6) whether in the hypothetical presence of a negative CT angiogram they would still recommend CA. An automatic NAPSAH diagnosis was also generated by combining responses to questions 2-5. Reliability was estimated using Gwet's AC1 (κG), and the relationship between the NCCT diagnosis of NAPSAH and the recommendation to perform CA using Cramer's V test. Multi-rater accuracy of NCCT in predicting negative CA was explored. RESULTS: Inter-rater reliability for the presence of NAPSAH was moderate (κG = 0.58; 95%CI: 0.47, 0.69), but improved to substantial when automatically generated (κG = 0.70; 95%CI: 0.59, 0.81). The most reliable criteria were the absence of AIF filling (κG = 0.79) and extension to LSF (κG = 0.79). Mean intra-rater reliability was substantial (κG = 0.65). NAPSAH weakly correlated with CA decision (V = 0.50). Mean sensitivity and specificity were 58% (95%CI: 44%, 71%) and 83 % (95%CI: 72 %, 94%), respectively. CONCLUSION: NAPSAH remains a diagnosis of exclusion. The NCCT diagnosis was moderately reliable and its impact on clinical decisions modest.

Assessment of Radiology Artificial Intelligence Software: A Validation and Evaluation Framework.

Artificial intelligence (AI) software in radiology is becoming increasingly prevalent and performance is improving rapidly with new applications for given use cases being developed continuously, oftentimes with development and validation occurring in parallel. Several guidelines have provided reporting standards for publications of AI-based research in medicine and radiology. Yet, there is an unmet need for recommendations on the assessment of AI software before adoption and after commercialization. As the radiology AI ecosystem continues to grow and mature, a formalization of system assessment and evaluation is paramount to ensure patient safety, relevance and support to clinical workflows, and optimal allocation of limited AI development and validation resources before broader implementation into clinical practice. To fulfil these needs, we provide a glossary for AI software types, use cases and roles within the clinical workflow; list healthcare needs, key performance indicators and required information about software prior to assessment; and lay out examples of software performance metrics per software category. This conceptual framework is intended to streamline communication with the AI software industry and provide healthcare decision makers and radiologists with tools to assess the potential use of these software. The proposed software evaluation framework lays the foundation for a radiologist-led prospective validation network of radiology AI software. Learning Points: The rapid expansion of AI applications in radiology requires standardization of AI software specification, classification, and evaluation. The Canadian Association of Radiologists' AI Tech & Apps Working Group Proposes an AI Specification document format and supports the implementation of a clinical expert evaluation process for Radiology AI software.

Neonatal 6-OHDA lesion of the SNc induces striatal compensatory sprouting from surviving SNc dopaminergic neurons without VTA contribution.

Dopamine (DA) neurons of the substantia nigra pars compacta (SNc) are uniquely vulnerable to neurodegeneration in Parkinson's disease (PD). We hypothesize that their large axonal arbor is a key factor underlying their vulnerability, due to increased bioenergetic, proteostatic and oxidative stress. In keeping with this model, other DAergic populations with smaller axonal arbors are mostly spared during the course of PD and are more resistant to experimental lesions in animal models. Aiming to improve mouse PD models, we examined if neonatal partial SNc lesions could lead to adult mice with fewer SNc DA neurons that are endowed with larger axonal arbors because of compensatory mechanisms. We injected 6-hydroxydopamine (6-OHDA) unilaterally in the SNc at an early postnatal stage at a dose selected to induce loss of approximately 50% of SNc DA neurons. We find that at 10 and 90 days after the lesion, the axons of SNc DA neurons show massive compensatory sprouting, as revealed by the proportionally smaller decrease in tyrosine hydroxylase (TH) in the striatum compared with the loss of SNc DA neuron cell bodies. The extent and origin of this axonal sprouting was further investigated by AAV-mediated expression of eYFP in SNc or ventral tegmental area (VTA) DA neurons of adult mice. Our results reveal that SNc DA neurons have the capacity to substantially increase their axonal arbor size and suggest that mice designed to have reduced numbers of SNc DA neurons could potentially be used to develop better mouse models of PD, with elevated neuronal vulnerability.

Predictive online 3D target tracking with population-based generative networks for image-guided radiotherapy.

PURPOSE: Respiratory motion of thoracic organs poses a severe challenge for the administration of image-guided radiotherapy treatments. Providing online and up-to-date volumetric information during free breathing can improve target tracking, ultimately increasing treatment efficiency and reducing toxicity to surrounding healthy tissue. In this work, a novel population-based generative network is proposed to address the problem of 3D target location prediction from 2D image-based surrogates during radiotherapy, thus enabling out-of-plane tracking of treatment targets using images acquired in real time. METHODS: The proposed model is trained to simultaneously create a low-dimensional manifold representation of 3D non-rigid deformations and to predict, ahead of time, the motion of the treatment target. The predictive capabilities of the model allow correcting target location errors that can arise due to system latency, using only a baseline volume of the patient anatomy. Importantly, the method does not require supervised information such as ground-truth registration fields, organ segmentation, or anatomical landmarks. RESULTS: The proposed architecture was evaluated on both free-breathing 4D MRI and ultrasound datasets. Potential challenges present in a realistic therapy, like different acquisition protocols, were taken into account by using an independent hold-out test set. Our approach enables 3D target tracking from single-view slices with a mean landmark error of 1.8 mm, 2.4 mm and 5.2 mm in volunteer MRI, patient MRI and US datasets, respectively, without requiring any prior subject-specific 4D acquisition. CONCLUSIONS: This model presents several advantages over state-of-the-art approaches. Namely, it benefits from an explainable latent space with explicit respiratory phase discrimination. Thanks to the strong generalization capabilities of neural networks, it does not require establishing inter-subject correspondences. Once trained, it can be quickly deployed with an inference time of only 8 ms. The results show the capability of the network to predict future anatomical changes and track tumors in real time, yielding statistically significant improvements over related methods.