High-resolution computed tomography (HRCT) is important for diagnosing interstitial lung disease (ILD) in inflammatory rheumatic disease (IRD) patients. However, visual ILD assessment via HRCT often has high inter-reader variability. Artificial intelligence (AI)-based techniques for quantitative image analysis promise more accurate diagnostic and prognostic information. This study evaluated the reliability of artificial intelligence-based quantification of pulmonary HRCT (AIqpHRCT) in IRD-ILD patients and verified IRD-ILD quantification using AIqpHRCT in the clinical setting. Reproducibility of AIqpHRCT was verified for each typical HRCT pattern (ground-glass opacity [GGO], non-specific interstitial pneumonia [NSIP], usual interstitial pneumonia [UIP], granuloma). Additional, 50 HRCT datasets from 50 IRD-ILD patients using AIqpHRCT were analysed and correlated with clinical data and pulmonary lung function parameters. AIqpHRCT presented 100% agreement (coefficient of variation = 0.00%, intraclass correlation coefficient = 1.000) regarding the detection of the different HRCT pattern. Furthermore, AIqpHRCT data showed an increase of ILD from 10.7 ± 28.3% (median = 1.3%) in GGO to 18.9 ± 12.4% (median = 18.0%) in UIP pattern. The extent of fibrosis negatively correlated with FVC (ρ=-0.501), TLC (ρ=-0.622), and DLCO (ρ=-0.693) (p < 0.001). GGO measured by AIqpHRCT also significant negatively correlated with DLCO (ρ=-0.699), TLC (ρ=-0.580) and FVC (ρ=-0.423). For the first time, the study demonstrates that AIpqHRCT provides a highly reliable method for quantifying lung parenchymal changes in HRCT images of IRD-ILD patients. Further, the AIqpHRCT method revealed significant correlations between the extent of ILD and lung function parameters. This highlights the potential of AIpqHRCT in enhancing the accuracy of ILD diagnosis and prognosis in clinical settings, ultimately improving patient management and outcomes.
Interstitial lung disease (ILD) is a common and serious organ manifestation in patients with connective tissue disease (CTD), but it is uncertain whether there is a difference in ILD between symptomatic and asymptomatic patients. Therefore, we conducted a study to evaluate differences in the extent of ILD based on radiological findings between symptomatic/asymptomatic patients, using an artificial intelligence (AI)-based quantification of pulmonary high-resolution computed tomography (AIpqHRCT). Within the study, 67 cross-sectional HRCT datasets and clinical data (including pulmonary function test) of consecutively patients (mean age: 57.1 ± 14.7 years, woman n = 45; 67.2%) with both, initial diagnosis of CTD, with systemic sclerosis being the most frequent (n = 21, 31.3%), and ILD (all without immunosuppressive therapy), were analysed using AIqpHRCT. 25.4% (n = 17) of the patients with ILD at initial diagnosis of CTD had no pulmonary symptoms. Regarding the baseline characteristics (age, gender, disease), there were no significant difference between the symptomatic and asymptomatic group. The pulmonary function test (PFT) revealed the following mean values (%predicted) in the symptomatic and asymptomatic group, respectively: Forced vital capacity (FVC) 69.4 ± 17.4% versus 86.1 ± 15.8% (p = 0.001), and diffusing capacity of the lung for carbon monoxide (DLCO) 49.7 ± 17.9% versus 60.0 ± 15.8% (p = 0.043). AIqpHRCT data showed a significant higher amount of high attenuated volume (HAV) (14.8 ± 11.0% versus 8.9 ± 3.9%; p = 0.021) and reticulations (5.4 ± 8.7% versus 1.4 ± 1.5%; p = 0.035) in symptomatic patients. A quarter of patients with ILD at the time of initial CTD diagnosis had no pulmonary symptoms, showing DLCO were reduced in both groups. Also, AIqpHRCT demonstrated clinically relevant ILD in asymptomatic patients. These results underline the importance of an early risk adapted screening for ILD also in asymptomatic CTD patients, as ILD is associated with increased mortality.
When finger joints become immobile due to an accident during sports or a widespread disease such as rheumatoid arthritis, customised finger joint implants are to be created. In an automated process chain, implants will be produced from ceramic or metallic materials. Artificial intelligence-supported software is used to calculate three-dimensional models of the finger bones from two-dimensional X-ray images. Then, the individual implant design is derived from the finger model and 3D printed. The 3D printing process and the structures used are evaluated via model tests and the final implant design via a reliability calculation in a way to ensure that this is also possible via an AI process in the future. Using additive manufacturing with silicon nitride-based ceramics, model specimens and implants are produced via the lithography-based ceramic vat photopolymerisation process with full geometry or elements of triple periodic minimal surfaces structure. The model specimens are tested experimentally, and the loads are matched with a characteristic strength assuming a Weibull distribution of defects in the volume to generate and match failure probabilities. Calculated fracture forces of the silicon nitride-based ceramic structure was validated by comparison of simulation and tests, and the calculation can be used as a quality index for training of artificial intelligence in the future. The proposed method for individualized finger implant design and manufacturing may allow for correction of potential malpositions of the fingers in the future.
15 March 2019 in IEEE Transactions on Biomedical Engineering
Objective: The ASTM F2213 standard describes the measurement of the magnetically induced torques by means of a torsional spring method. As this method is highly complex, it has been rarely ever employed. The purpose was to develop an easy-to-handle measuring setup for the reproducible, fast and precise measurement of magnetically induced torques, following ASTM F2213, at a static magnetic field $(B_0)$ of 1.5 T and 3 T. Methods: An MR-safe measuring platform was developed according to the ASTM F2213 standard and combined with a precision balance. The evaluation was performed with 10 test objects, e.g. a steel screw (length 60 mm) and a neurostimulator, at 1.5 T and 3 T. The torque was measured at 10-degree increments as the test object was rotated relative to $B_0$ . Results: The measurement setup allows fast and reproducible measurements. The maximum magnetically induced torque measured was $\rm 397 \pm 2 N\cdot mm$ for the steel screw and $\rm 63 \pm 1 N\cdot mm$ for the neurostimulator. The maximum torque on a test object is insensitive to $B_0$ if the magnetization of the object has reached saturation. Conclusion: Ferromagnetic implants may exhibit strong magnetically induced torques during MRI examination. For analyzing these magnetically induced torques, we developed a digital measuring device that allows simplified and accelerated application of the ASTM F2213 standard test method. Significance: In a $B_0$ , ferromagnetic implants may exert considerable torques on the surrounding tissue. These forces may affect the functionality of the implant and inflict severe injuries on a patient.
Purpose: During the coronavirus disease 2019 (COVID-19) pandemic, hospitals still face the challenge of timely identification of infected individuals before inpatient admission. An artificial intelligence approach based on an established clinical network may improve prospective pandemic preparedness.Method: Supervised machine learning was used to construct diagnostic models to predict COVID-19. A pooled database was retrospectively generated from 4437 participant data that were collected between January 2017 and October 2020 at 12 German centers that belong to the radiological cooperative network of the COVID-19 (RACOON) consortium. A total of 692 (15.6 %) participants were COVID-19 positive according to the reference of the reverse transcription-polymerase chain reaction test. The diagnostic models included chest CT features (model R), clinical examination and laboratory test features (model CL), or all three feature categories (model RCL). Performance outcomes included accuracy, sensitivity, specificity, negative and positive predictive value, and area under the receiver operating curve (AUC).Results: Performance of predictive models improved significantly by adding chest CT features to clinical evaluation and laboratory test features. Without (model CL) and with inclusion of chest CT (model RCL), sensitivity was 0.82 and 0.89 (p < 0.0001), specificity was 0.84 and 0.89 (p < 0.0001), negative predictive value was 0.96 and 0.97 (p < 0.0001), AUC was 0.92 and 0.95 (p < 0.0001), and proportion of false negative classifications was 2.6 % and 1.7 % (p < 0.0001), respectively.Conclusions: Addition of chest CT features to machine learning-based predictive models improves the effectiveness in ruling out COVID-19 before inpatient admission to regular wards.
27 December 2011 in Magnetic Resonance in Medicine
The in vivo pain treatment was successfully performed with the patient in a prone position. The PD-weighted TSE with echo time = 10 ms rendered contrast-to-noise-ratio values of 27 ± 10 for needle/fat, 1.6 ± 5 for needle/muscle, and 4 ± 4.7 for needle/nerve tissue. The mean diameter of the needle artifact was 1.2 ± 0.2 mm. In the T1-weighted gradient echo, the needle's artifact diameter was 6 ± 2 mm; the needle's contrast-to-noise ratio relative to muscle tissue was 4 ± 2, 7.6 ± 1.5 for needle/fat, and 5 ± 1 for needle/nerve tissue. With the PD-weighted TSE (echo time = 10 ms) and the T1-weighted gradient echo, the needle was imaged reliably throughout the intervention. The butterfly surface coil is feasible for the guidance of spinal interventions in a prone patient. Magn Reson Med, 2012.