Automated surgery planning for an obstructed nose by combining computational fluid dynamics with reinforcement learning.

Septoplasty and turbinectomy are among the most common interventions in the field of rhinology. Their constantly debated success rates and the lack of quantitative flow data of the entire nasal airway for planning the surgery necessitate methodological improvement. Thus, physics-based surgery planning is highly desirable. In this work, a novel and accurate method is developed to enhance surgery planning by physical aspects of respiration, i.e., to plan anti-obstructive surgery, for the first time a reinforcement learning algorithm is combined with large-scale computational fluid dynamics simulations. The method is integrated into an automated pipeline based on computed tomography imaging. The proposed surgical intervention is compared to a surgeon's initial plan, or the maximum possible intervention, which allows the quantitative evaluation of the intended surgery. Two criteria are considered: (i) the capability to supply the nasal airway with air expressed by the pressure loss and (ii) the capability to heat incoming air represented by the temperature increase. For a test patient suffering from a deviated septum near the nostrils and a bony spur further downstream, the method recommends surgical interventions exactly at these locations. For equal weights on the two criteria (i) and (ii), the algorithm proposes a slightly weaker correction of the deviated septum at the first location, compared to the surgeon's plan. At the second location, the algorithm proposes to keep the bony spur. For a larger weight on criterion (i), the algorithm tends to widen the nasal passage by removing the bony spur. For a larger weight on criterion (ii), the algorithm's suggestion approaches the pre-surgical state with narrowed channels that favor heat transfer. A second patient is investigated that suffers from enlarged turbinates in the left nasal passage. For equal weights on the two criteria (i) and (ii), the algorithm proposes a nearly complete removal of the inferior turbinate, and a moderate reduction of the middle turbinate. An increased weight on criterion (i) leads to an additional reduction of the middle turbinate, and a larger weight on criterion (ii) yields a solution with only slight reductions of both turbinates, i.e., focusing on a sufficient heat exchange between incoming air and the air-nose interface. The proposed method has the potential to improve the success rates of the aforementioned surgeries and can be extended to further biomedical flows.

An effective simulation- and measurement-based workflow for enhanced diagnostics in rhinology.

Physics-based analyses have the potential to consolidate and substantiate medical diagnoses in rhinology. Such methods are frequently subject to intense investigations in research. However, they are not used in clinical applications, yet. One issue preventing their direct integration is that these methods are commonly developed as isolated solutions which do not consider the whole chain of data processing from initial medical to higher valued data. This manuscript presents a workflow that incorporates the whole data processing pipeline based on a Jupyter environment. Therefore, medical image data are fully automatically pre-processed by machine learning algorithms. The resulting geometries employed for the simulations on high-performance computing systems reach an accuracy of up to 99.5% compared to manually segmented geometries. Additionally, the user is enabled to upload and visualize 4-phase rhinomanometry data. Subsequent analysis and visualization of the simulation outcome extend the results of standardized diagnostic methods by a physically sound interpretation. Along with a detailed presentation of the methodologies, the capabilities of the workflow are demonstrated by evaluating an exemplary medical case. The pipeline output is compared to 4-phase rhinomanometry data. The comparison underlines the functionality of the pipeline. However, it also illustrates the influence of mucosa swelling on the simulation. Graphical Abstract Workflow for enhanced diagnostics in rhinology.

Deep Learning-Based Automated Thrombolysis in Cerebral Infarction Scoring: A Timely Proof-of-Principle Study.

Background and Purpose: Mechanical thrombectomy is an established procedure for treatment of acute ischemic stroke. Mechanical thrombectomy success is commonly assessed by the Thrombolysis in Cerebral Infarction (TICI) score, assigned by visual inspection of X-ray digital subtraction angiography data. However, expert-based TICI scoring is highly observer-dependent. This represents a major obstacle for mechanical thrombectomy outcome comparison in, for instance, multicentric clinical studies. Focusing on occlusions of the M1 segment of the middle cerebral artery, the present study aimed to develop a deep learning (DL) solution to automated and, therefore, objective TICI scoring, to evaluate the agreement of DL- and expert-based scoring, and to compare corresponding numbers to published scoring variability of clinical experts. Methods: The study comprises 2 independent datasets. For DL system training and initial evaluation, an in-house dataset of 491 digital subtraction angiography series and modified TICI scores of 236 patients with M1 occlusions was collected. To test the model generalization capability, an independent external dataset with 95 digital subtraction angiography series was analyzed. Characteristics of the DL system were modeling TICI scoring as ordinal regression, explicit consideration of the temporal image information, integration of physiological knowledge, and modeling of inherent TICI scoring uncertainties. Results: For the in-house dataset, the DL system yields Cohen's kappa, overall accuracy, and specific agreement values of 0.61, 71%, and 63% to 84%, respectively, compared with the gold standard: the expert rating. Values slightly drop to 0.52/64%/43% to 87% when the model is, without changes, applied to the external dataset. After model updating, they increase to 0.65/74%/60% to 90%. Literature Cohen's kappa values for expert-based TICI scoring agreement are in the order of 0.6. Conclusions: The agreement of DL- and expert-based modified TICI scores in the range of published interobserver variability of clinical experts highlights the potential of the proposed DL solution to automated TICI scoring.

A nanomolar multivalent ligand as entry inhibitor of the hemagglutinin of avian influenza.

Influenza virus attaches itself to sialic acids on the surface of epithelial cells of the upper respiratory tract of the host using its own protein hemagglutinin. Species specificity of influenza virus is determined by the linkages of the sialic acids. Birds and humans have α2-3 and α2-6 linked sialic acids, respectively. Viral hemagglutinin is a homotrimeric receptor, and thus, tri- or oligovalent ligands should have a high binding affinity. We describe the in silico design, chemical synthesis and binding analysis of a trivalent glycopeptide mimetic. This compound binds to hemagglutinin H5 of avian influenza with a dissociation constant of K(D) = 446 nM and an inhibitory constant of K(I) = 15 µM. In silico modeling shows that the ligand should also bind to hemagglutinin H7 of the virus that causes the current influenza outbreak in China. The trivalent glycopeptide mimetic and analogues have the potential to block many different influenza viruses.