TY - JOUR A1 - Huflage, Henner A1 - Fieber, Tabea A1 - Färber, Christian A1 - Knarr, Jonas A1 - Veldhoen, Simon A1 - Jordan, Martin C. A1 - Gilbert, Fabian A1 - Bley, Thorsten Alexander A1 - Meffert, Rainer H. A1 - Grunz, Jan-Peter A1 - Schmalzl, Jonas T1 - Interobserver reliability of scapula fracture classifications in intra- and extra-articular injury patterns JF - BMC Musculoskeletal Disorders N2 - Background Morphology and glenoid involvement determine the necessity of surgical management in scapula fractures. While being present in only a small share of patients with shoulder trauma, numerous classification systems have been in use over the years for categorization of scapula fractures. The purpose of this study was to evaluate the established AO/OTA classification in comparison to the classification system of Euler and Rüedi (ER) with regard to interobserver reliability and confidence in clinical practice. Methods Based on CT imaging, 149 patients with scapula fractures were retrospectively categorized by two trauma surgeons and two radiologists using the classification systems of ER and AO/OTA. To measure the interrater reliability, Fleiss kappa (κ) was calculated independently for both fracture classifications. Rater confidence was stated subjectively on a five-point scale and compared with Wilcoxon signed rank tests. Additionally, we computed the intraclass correlation coefficient (ICC) based on absolute agreement in a two-way random effects model to assess the diagnostic confidence agreement between observers. Results In scapula fractures involving the glenoid fossa, interrater reliability was substantial (κ = 0.722; 95% confidence interval [CI] 0.676–0.769) for the AO/OTA classification in contrast to moderate agreement (κ = 0.579; 95% CI 0.525–0.634) for the ER classification system. Diagnostic confidence for intra-articular fracture patterns was superior using the AO/OTA classification compared to ER (p < 0.001) with higher confidence agreement (ICC: 0.882 versus 0.831). For extra-articular fractures, ER (κ = 0.817; 95% CI 0.771–0.863) provided better interrater reliability compared to AO/OTA (κ = 0.734; 95% CI 0.692–0.776) with higher diagnostic confidence (p < 0.001) and superior agreement between confidence ratings (ICC: 0.881 versus 0.912). Conclusions The AO/OTA classification is most suitable to categorize intra-articular scapula fractures with glenoid involvement, whereas the classification system of Euler and Rüedi appears to be superior in extra-articular injury patterns with fractures involving only the scapula body, spine, acromion and coracoid process. KW - confidence KW - scapula KW - glenoid KW - fracture KW - classification KW - reliability Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:20-opus-299795 VL - 23 IS - 1 ER - TY - JOUR A1 - Vollmer, Andreas A1 - Vollmer, Michael A1 - Lang, Gernot A1 - Straub, Anton A1 - Kübler, Alexander A1 - Gubik, Sebastian A1 - Brands, Roman C. A1 - Hartmann, Stefan A1 - Saravi, Babak T1 - Performance analysis of supervised machine learning algorithms for automatized radiographical classification of maxillary third molar impaction JF - Applied Sciences N2 - Background: Oro-antral communication (OAC) is a common complication following the extraction of upper molar teeth. The Archer and the Root Sinus (RS) systems can be used to classify impacted teeth in panoramic radiographs. The Archer classes B-D and the Root Sinus classes III, IV have been associated with an increased risk of OAC following tooth extraction in the upper molar region. In our previous study, we found that panoramic radiographs are not reliable for predicting OAC. This study aimed to (1) determine the feasibility of automating the classification (Archer/RS classes) of impacted teeth from panoramic radiographs, (2) determine the distribution of OAC stratified by classification system classes for the purposes of decision tree construction, and (3) determine the feasibility of automating the prediction of OAC utilizing the mentioned classification systems. Methods: We utilized multiple supervised pre-trained machine learning models (VGG16, ResNet50, Inceptionv3, EfficientNet, MobileNetV2), one custom-made convolutional neural network (CNN) model, and a Bag of Visual Words (BoVW) technique to evaluate the performance to predict the clinical classification systems RS and Archer from panoramic radiographs (Aim 1). We then used Chi-square Automatic Interaction Detectors (CHAID) to determine the distribution of OAC stratified by the Archer/RS classes to introduce a decision tree for simple use in clinics (Aim 2). Lastly, we tested the ability of a multilayer perceptron artificial neural network (MLP) and a radial basis function neural network (RBNN) to predict OAC based on the high-risk classes RS III, IV, and Archer B-D (Aim 3). Results: We achieved accuracies of up to 0.771 for EfficientNet and MobileNetV2 when examining the Archer classification. For the AUC, we obtained values of up to 0.902 for our custom-made CNN. In comparison, the detection of the RS classification achieved accuracies of up to 0.792 for the BoVW and an AUC of up to 0.716 for our custom-made CNN. Overall, the Archer classification was detected more reliably than the RS classification when considering all algorithms. CHAID predicted 77.4% correctness for the Archer classification and 81.4% for the RS classification. MLP (AUC: 0.590) and RBNN (AUC: 0.590) for the Archer classification as well as MLP 0.638) and RBNN (0.630) for the RS classification did not show sufficient predictive capability for OAC. Conclusions: The results reveal that impacted teeth can be classified using panoramic radiographs (best AUC: 0.902), and the classification systems can be stratified according to their relationship to OAC (81.4% correct for RS classification). However, the Archer and RS classes did not achieve satisfactory AUCs for predicting OAC (best AUC: 0.638). Additional research is needed to validate the results externally and to develop a reliable risk stratification tool based on the present findings. KW - oro-antral communication KW - oro-antral fistula KW - prediction KW - machine learning KW - teeth extraction KW - complications KW - classification KW - artificial intelligence Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:20-opus-281662 SN - 2076-3417 VL - 12 IS - 13 ER - TY - JOUR A1 - Merkel, Helena A1 - Lindner, Dirk A1 - Gaber, Khaled A1 - Ziganshyna, Svitlana A1 - Jentzsch, Jennifer A1 - Mucha, Simone A1 - Gerhards, Thilo A1 - Sari, Sabine A1 - Stock, Annika A1 - Vothel, Felicitas A1 - Falter, Lea A1 - Quäschling, Ulf A1 - Hoffmann, Karl-Titus A1 - Meixensberger, Jürgen A1 - Halama, Dirk A1 - Richter, Cindy T1 - Standardized classification of cerebral vasospasm after subarachnoid hemorrhage by digital subtraction angiography JF - Journal of Clinical Medicine N2 - Background: During the last decade, cerebral vasospasm after aneurysmal subarachnoid hemorrhage (SAH) was a current research focus without a standardized classification in digital subtraction angiography (DSA). This study was performed to investigate a device-independent visual cerebral vasospasm classification for endovascular treatment. Methods: The analyses are DSA based rather than multimodal. Ten defined points of intracranial arteries were measured in 45 patients suffering from cerebral vasospasm after SAH at three time points (hospitalization, before spasmolysis, control after six months). Mathematical clustering of vessel diameters was performed to generate four objective grades for comparison. Six interventional neuroradiologists in two groups scored 237 DSAs after a new visual classification (grade 0–3) developed on a segmental pattern of vessel contraction. For the second group, a threshold-based criterion was amended. Results: The raters had a reproducibility of 68.4% in the first group and 75.2% in the second group. The complementary threshold-based criterion increased the reproducibility by about 6.8%, while the rating deviated more from the mathematical clustering in all grades. Conclusions: The proposed visual classification scheme of cerebral vasospasm is suitable as a standard grading procedure for endovascular treatment. There is no advantage of a threshold-based criterion that compensates for the effort involved. Automated vessel analysis is superior to compare inter-group results in research settings. KW - cerebral vasospasm KW - subarachnoid hemorrhage KW - classification KW - vessel diameter Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:20-opus-270638 SN - 2077-0383 VL - 11 IS - 7 ER - TY - JOUR A1 - Mertens, Griet A1 - Andries, Ellen A1 - Kurz, Anja A1 - Tȧvora-Vieira, Dayse A1 - Calvino, Miryam A1 - Amann, Edda A1 - Anderson, Ilona A1 - Lorens, Artur T1 - Towards a consensus on an ICF-based classification system for horizontal sound-source localization JF - Journal of Personalized Medicine N2 - The study aimed to develop a consensus classification system for the reporting of sound localization testing results, especially in the field of cochlear implantation. Against the background of an overview of the wide variations present in localization testing procedures and reporting metrics, a novel classification system was proposed to report localization errors according to the widely accepted International Classification of Functioning, Disability and Health (ICF) framework. The obtained HEARRING_LOC_ICF scale includes the ICF graded scale: 0 (no impairment), 1 (mild impairment), 2 (moderate impairment), 3 (severe impairment), and 4 (complete impairment). Improvement of comparability of localization results across institutes, localization testing setups, and listeners was demonstrated by applying the classification system retrospectively to data obtained from cohorts of normal-hearing and cochlear implant listeners at our institutes. The application of our classification system will help to facilitate multi-center studies, as well as allowing better meta-analyses of data, resulting in improved evidence-based practice in the field. KW - sound localization KW - binaural hearing KW - cochlear implant KW - interaural level difference KW - interaural time difference KW - single-sided deafness KW - unilateral hearing loss KW - testing-method consensus KW - classification KW - ICF Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:20-opus-297319 SN - 2075-4426 VL - 12 IS - 12 ER -