Background/Objectives: Large language models (LLMs), such as ChatGPT, have emerged as potential clinical support tools to enhance precision in personalized patient care, but their reliability in radiological image interpretation remains uncertain. The primary aim of our study was to evaluate the diagnostic accuracy of ChatGPT-4o in interpreting chest X-rays (CXRs) and abdominal X-rays (AXRs) by comparing its performance to expert radiology findings, whilst secondary aims were diagnostic confidence and patient safety. Methods: A total of 500 X-rays, including 257 CXR (51.4%) and 243 AXR (48.5%), were analyzed. Diagnoses made by ChatGPT-4o were compared to expert interpretations. Confidence scores (1–4) were assigned and responses were evaluated for patient safety. Results: ChatGPT-4o correctly identified 345 of 500 (69%) pathologies (95% CI: 64.81–72.9). For AXRs 175 of 243 (72.02%) pathologies were correctly diagnosed (95% CI: 66.06–77.28), while for CXRs 170 of 257 (66.15%) were accurate (95% CI: 60.16–71.66). The highest detection rates among CXRs were observed for pulmonary edema, tumor, pneumonia, pleural effusion, cardiomegaly, and emphysema, and lower rates were observed for pneumothorax, rib fractures, and enlarged mediastinum. AXR performance was highest for intestinal obstruction and foreign bodies, and weaker for pneumoperitoneum, renal calculi, and diverticulitis. Confidence scores were higher for AXRs (mean 3.45 ± 1.1) than CXRs (mean 2.48 ± 1.45). All responses (100%) were considered to be safe for the patient. Interobserver agreement was high (kappa = 0.920), and reliability (second prompt) was moderate (kappa = 0.750). Conclusions: ChatGPT-4o demonstrated moderate accuracy for the interpretation of X-rays, being higher for AXRs compared to CXRs. Improvements are required for its use as efficient clinical support tool.

Background: Robotic-based guidance systems are becoming increasingly capable of assisting in needle placement during interventional procedures. Despite these technical advances, less sophisticated low-cost guidance devices promise to enhance puncture accuracy compared with the traditional freehand technique. Purpose: To compare the in vitro accuracy and feasibility of two different aiming devices for computed-tomography (CT)-guided punctures. Methods: A total of 560 CT-guided punctures were performed by using either a robotic (Perfint Healthcare: Maxio) or a novel low-cost patient-mounted system (Medical Templates AG: Puncture Cube System [PCS]) for the placement of Kirschner wires in a plexiglass phantom with different slice thicknesses. Needle placement accuracy as well as procedural time were assessed. The Euclidean (ED) and normal distances (ND) were calculated at the entry and target point. Results: Using the robotic device, the ND at the target for 1.25 mm, 2.5 mm, 3.75 mm and 5 mm slice thickness were 1.28 mm (SD ± 0.79), 1.25 mm (SD ± 0.81), 1.35 mm (SD ± 1.00) and 1.35 mm (SD ± 1.03). Using the PCS, the ND at the target for 1 mm, 3 mm and 5 mm slices were 3.84 mm (SD ± 1.75), 4.41 mm (SD ± 2.31) and 4.41 mm (SD ± 2.11), respectively. With all comparable slice thicknesses, the robotic device was significantly more accurate compared to the low-cost device (p < 0.001). Needle placement with the PCS resulted in lower intervention time (mean, 158.83 s [SD ± 23.38] vs. 225.67 s [SD ± 17.2]). Conclusion: Although the robotic device provided more accurate results, both guidance systems showed acceptable results and may be helpful for interventions in difficult anatomical regions and for those requiring complex multi-angle trajectories.

Objectives: The purpose of this study was to assess the performance of an optically tracked robot for computed-tomography (CT)-guided needle placements in a phantom study. Methods: In total, 240 needle punctures were carried out with the help of an optically tracked robotic device (Micromate) based on CT image datasets at three different slice thicknesses (1, 3, and 5 mm). Conically shaped targets inside a gelatin-filled plexiglass phantom were punctured. The target positioning error between the planned and actual needle trajectory was assessed by measuring the lateral positioning error (ND) between the target and the puncture needle and the Euclidean distance (ED) between the needle tip and target in control CTs. Results: The mean ND and ED for the thinnest CT slice thickness were 1.34 mm (SD ± 0.82) and 2.1 mm (SD ± 0.75), respectively. There was no significant impact of target depth on targeting accuracy for ND (p = 0.094) or ED (p = 0.187). The mean duration for the planning of one trajectory and for needle positioning were 42 s (SD ± 4) and 64 s (SD ± 7), respectively. Conclusions: In this ex vivo study, the robotic targeting device yielded satisfactory accuracy results at CT slice thicknesses of 1 and 3 mm. This technology may be particularly useful in interventions where the accurate placement of needle-like instruments is required.

Objective: To evaluate two different dual-energy computed tomography (DECT) post-processing protocols for the detection of MSU deposits in foot tendons of cadavers with verification by polarizing light microscopy as the gold standard. Material and Methods: A total of 40 embalmed cadavers (15 male; 25 female; median age, 82 years; mean, 80 years; range, 52–99; SD ± 10.9) underwent DECT to assess MSU deposits in foot tendons. Two postprocessing DECT protocols with different Hounsfield unit (HU) thresholds, 150/500 (=established) versus 120/500 (=modified). HU were applied to dual source acquisition with 80 kV for tube A and 140 kV for tube B. Six fresh cadavers (4 male; 2 female; median age, 78; mean, 78.5; range 61–95) were examined by DECT. Tendon dissection of 2/6 fresh cadavers with positive DECT 120 and negative DECT 150 studies were used to verify MSU deposits by polarizing light microscopy. Results: The tibialis anterior tendon was found positive in 57.5%/100% (DECT 150/120), the peroneus tendon in 35%/100%, the achilles tendon in 25%/90%, the flexor halluces longus tendon in 10%/100%, and the tibialis posterior tendon in 12.5%/97.5%. DECT 120 resulted in increased tendon MSU deposit detection, when DECT 150 was negative, with an overall agreement between DECT 150 and DECT 120 of 80% (p = 0.013). Polarizing light microscope confirmed MSU deposits detected only by DECT 120 in the tibialis anterior, the achilles, the flexor halluces longus, and the peroneal tendons. Conclusion: The DECT 120 protocol showed a higher sensitivity when compared to DECT 150.

Purpose: To compare the results of a novice with those of experienced interventional radiologists (IRs) for stereotactic radiofrequency ablation (SRFA) of malignant liver tumors in terms of safety, technical success, and local tumor control. Methods: A database, including all SRFA procedures performed in a single center between January 2011 and December 2018 was retrospectively analyzed. A total of 39 ablation sessions performed by a novice IR were compared to the results of three more experienced IRs. Comparative SRFA sessions were selected using propensity score matching considering tumor type, age, sex, tumor size, and tumor number as matching variables. Overall, 549 target tumors were treated in 273 sessions. Median tumor size was 2.2 cm (1.0–8.5 cm) for 178 hepatocellular carcinomas (HCCs) and 3.0 cm (0.5–13.0 cm) for 371 metastases. A median of 2 (1–11) tumors were treated per session. Results: No significant differences were observed when comparing the results of more experienced IRs with those of a novice IR regarding the rates of major complications (6.8% [16/234] vs. 5.1% [2/39]; p = 0.477), mortality (1.3% [2/234] vs. 0% [0/39]; p = 0.690), primary technical efficacy (98.5% [525/533] vs. 98.9% [94/95]; p = 0.735), and local recurrence (5.6% [30/533] vs. 5.3% [5/95]; p = 0.886). However, the median planning/placement time was significantly shorter for the experienced IRs (92 min vs. 119 min; p = 0.002). Conclusions: SRFA is a safe, effective, and reliable treatment option for malignant liver tumors and favorable outcomes can be achieved even by inexperienced operators with minimal supervision.

Objectives: To compare the volumes and shapes of the coagulation zone (CZ) of a multi-probe RFA system (three RFA electrodes) and a single-probe MWA system from the same vendor in an ex vivo bovine liver model. Material & Methods: A total of 48 CZs were obtained in bovine liver specimens with three different ablation system configurations (single-probe MWA vs. multi-probe RFA with 20 mm inter-probe distance [confluent CZ] vs. multi-probe RFA with 50 mm inter-probe distance [three individual CZs]) at 4, 6, 8, and 10 min ablation time using a fixed ablation protocol. Ablation diameters were measured and ellipticity indices (EIs) and volumes calculated. Calculations for all systems/configurations were compared. Results: Volumes and diameters increased with ablation time for all configurations. At 4 and 6 min ablation time volumes obtained with the RFA 50 mm setup, and at 8 and 10 min with the RFA 20 mm setup were the largest at 26.5 ± 4.1 mL, 38.1 ± 5.8 mL, 46.3 ± 4.9 mL, 48.4 ± 7.3 mL, respectively. The single-probe MWA could not reach the volumes of the RFA setups for any of the ablation times evaluated. EI were very similar and almost round for RFA 20 mm and single-probe MWA, and differed significantly to the more ovoid ones for the RFA 50 mm configuration. Conclusions: The multi-probe RFA system employing three electrodes achieved significantly larger ablation volumes in both configurations (confluent CZ and three individual CZs) per time as compared with a single-probe MWA system in this ex vivo bovine liver model.