Search

To test the hypothesis of equal or even superior applicability and accuracy of a fully integrated, laser-based computed tomography (CT) navigation system compared with conventional CT guidance for percutaneous interventions.CT-guided punctures were first performed in phantoms. Four radiologists with different experience levels (2 residents (L.B., C.D.) and 2 board-certified radiologists (B.M., K.R.) performed 48 punctures using both conventional image-guided and laser-guided approaches. Subsequently, 12 punctures were performed in patients during a clinical pilot trial. Phantom targets required an in-plane or a single-/double-angulated, out-of-plane approach. Planning and intervention time, control scan number, radiation exposure, and accuracy of needle placement (measured by deviation of the needle tip to the designated target) were assessed for each guidance technique and compared (Mann-Whitney U test and t test). Patient interventions were additionally analyzed for applicability in a clinical setting.The application of laser guidance software in the phantom study and in 12 human patients in a clinical setting was both technically and clinically feasible in all cases. The mean planning time (P = .009), intervention time (P = .005), control scan number (P.001), and radiation exposure (P = .013) significantly decreased for laser-navigated punctures compared with those for conventional CT guidance and especially in punctures with out-of-plane-trajectories. The accuracy significantly increased for laser-guided interventions compared with that for conventional CT (P.001).Interventional radiologists with differing levels of experience performed faster and more accurate punctures for out-of-plane trajectories in the phantom models, using a new, fully integrated, laser-guided CT software and demonstrated excellent clinical and technical success in initial clinical experiments.

Cholangiocarcinoma (CCA) are the second most common primary liver tumors and carry a dismal prognosis. Chemosaturation with percutaneous hepatic perfusion (PHP) is a palliative, intra-arterial therapeutic approach that provides a high dose chemotherapy of the liver with reduced systemic exposure. Aim of this retrospective, monocentric study was to analyze PHP as a palliative treatment for unresectable CCA. Toxicity, adverse events and complications were classified using the Common Terminology Criteria for Adverse Events (CTCAE v5.0). Overall response rate (ORR) and disease control rate (DCR) were evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST1.1). Median overall survival (mOS), median progression-free survival (mPFS) and hepatic mPFS (mhPFS) were computed using Kaplan–Meier estimation. In total 17 patients were treated with 42 PHP between 10/2014 and 09/2020. No significant complications occurred during the interventions. mOS was 27.6 (interquartile range (IQR) 16.5–37) months from first diagnosis and 9.9 (IQR 3.8–21) months from first PHP. mPFS was 4 (IQR 2–7) and mhPFS was 4 (IQR 3–10) months. ORR was 25% and DCR 75%. Significant, but transient hematotoxicity was frequent with grade 3/4 thrombopenia after 50%, leukopenia after 26% and anaemia after 21% of the interventions. An increase of transaminases (AST increase after 21% and ALT increase after 14% of the PHP) developed more often than a deterioration of the liver synthesis capacity. Salvage treatment with PHP has the potential to prolong life in selected patients with unresectable, refractory cholangiocarcinoma. The interventional procedure is safe. Post-interventional toxicity is frequent but manageable.

Radiology reports mostly contain free-text, which makes it challenging to obtain structured data. Natural language processing (NLP) techniques transform free-text reports into machine-readable document vectors that are important for creating reliable, scalable methods for data analysis. The aim of this study is to classify unstructured radiograph reports according to fractures of the distal fibula and to find the best text mining method.We established a novel German language report dataset: a designated search engine was used to identify radiographs of the ankle and the reports were manually labeled according to fractures of the distal fibula. This data was used to establish a machine learning pipeline, which implemented the text representation methods bag-of-words (BOW), term frequency-inverse document frequency (TF-IDF), principal component analysis (PCA), non-negative matrix factorization (NMF), latent Dirichlet allocation (LDA), and document embedding (doc2vec). The extracted document vectors were used to train neural networks (NN), support vector machines (SVM), and logistic regression (LR) to recognize distal fibula fractures. The results were compared via cross-tabulations of the accuracy (acc) and area under the curve (AUC).In total, 3268 radiograph reports were included, of which 1076 described a fracture of the distal fibula. Comparison of the text representation methods showed that BOW achieved the best results (AUC = 0.98; acc = 0.97), followed by TF-IDF (AUC = 0.97; acc = 0.96), NMF (AUC = 0.93; acc = 0.92), PCA (AUC = 0.92; acc = 0.9), LDA (AUC = 0.91; acc = 0.89) and doc2vec (AUC = 0.9; acc = 0.88). When comparing the different classifiers, NN (AUC = 0,91) proved to be superior to SVM (AUC = 0,87) and LR (AUC = 0,85).An automated classification of unstructured reports of radiographs of the ankle can reliably detect findings of fractures of the distal fibula. A particularly suitable feature extraction method is the BOW model. Key Points: Citation Format

Purpose To retrospectively evaluate the technical and clinical success of interventional treatments employed in three University medical centers and to develop work-flow recommendations for intra-arterial embolizations in patients with life-threatening spontaneous retroperitoneal and rectus sheath hemorrhage (SRRSH). Materials and methods Retrospective evaluation of all patients with contrast-enhanced CT and digital subtraction angiography (DSA) for SRRSH from 01/2018 to 12/2022, amounted to 91 interventions in 83 patients (45f, 38m) with a mean age of 68.1 ± 13.2 years. Analysis of the amount of bleeding and embolized vessels, choice of embolization material, technical success, and 30-day mortality was performed. Results Pre-interventional contrast-enhanced CT demonstrated active contrast extravasation in 79 cases (87%). DSA identified a mean of 1.4 ± 0.88 active bleeds in all but two interventions (98%), consisting of 60 cases with a singular and 39 cases of >1 bleeding artery, which were consecutively embolized. The majority of patients underwent embolization with either n-butyl-2-cyanoacrylate (NBCA; n=38), coils (n=21), or a combination of embolic agents (n=23). While the technical success rate was documented at 97.8%, 25 patients (30%) died within 30 days after the initial procedure, with mortality rates ranging from 25 to 86% between the centers, each following different diagnostic algorithms. Conclusion Embolotherapy is a safe therapy option with high technical success rates in patients with life-threatening SRRSH. To maximize clinical success and survival rates, we propose a standardized approach to angiography as well as a low threshold for re-angiography.

Background The artery of Adamkiewicz (AKA) or arteria radicularis anterior magna, is the name of the thoracolumbar segment of the dominant radiculomedullary artery, supplying the anterior lower spinal cord in addition to the anterior spinal artery. Results of a meta-analysis by Taterra et al. [1] showed its prevalence in 84.6% of people, a predominantly single appearance, most commonly originating from the left intercostal or lumbar arteries (87.4%) between the levels of T8-L1 (89%) [1], and a mean diameter of 1 mm [2, 3]. The artery most commonly arises from the spinal branch of the posterior ramus of the segmental artery (subcostal, posterior intercostal, or lumbar artery), which in turn arises from the descending aorta. The distal portion of the AKA, together with the anterior spinal artery, forms a characteristic “hairpin” turn [4]. Its location and anatomical variability cause its vulnerability to injury, especially during surgical but also interventional procedures. Anatomical knowledge paired with accurate localization of the AKA is therefore an integral part in planning orthopedic/neurosurgical spine, aortic vascular surgery repair procedures and endovascular selective arterial embolizations in this area. Case presentation In this case of a 73-year-old female patient with progression of a large, myelon-compressing vertebral metastasis of L1, protective, blood-flow-controlling occlusion of the proximal-most points of the artery of Adamkiewicz artery (AKA) and segmental spinal arteries was safely performed prior to surgical decompression and tumor debulking. Conclusions Current evidence indicates that knowledge of AKA location and the amount of collateralization may help estimate the risk of post-interventional spinal cord injury.

Background To evaluate effectivity of a 3D-motion correction algorithm in C-Arm CTs (CACT) with limited image quality (IQ) during transarterial chemoembolization (TACE). Methods From 1/2015–5/2021, 644 CACTs were performed in patients during TACE. Of these, 27 CACTs in 26 patients (18 m, 8f; 69.7 years ± 10.7 SD) of limited IQ were included. Post-processing of the original raw-data sets (CACTOrg) included application of a 3D-motion correction algorithm and bone segmentation (CACTMC_no_bone). Four radiologists (R1-4) compared the images by choosing their preferred dataset and recommending repeat acquisition in case of severe IQ-impairment. R1,2 performed additional grading of intrahepatic vessel visualization, presence/extent of movement artifacts, and overall IQ. Results R1,2 demonstrated excellent interobserver agreement for overall IQ (ICC 0.79,p p Org) to 1.39 (CACTMC_no_bone;p p Org:1.31 ± 1.67, CACTMC_no_bone:1.00 ± 1.34, p MC_no_bone were preferred, with identical datasets chosen by the readers to show benefit from the algorithm. Conclusion Application of a 3D-motion correction algorithm significantly improved IQ in diagnostically limited CACTs during TACE, with the potential to decrease repeat acquisitions.