Evaluation and mitigation of the limitations of large language models in clinical decision-making.

Clinical decision-making is one of the most impactful parts of a physician's responsibilities and stands to benefit greatly from artificial intelligence solutions and large language models (LLMs) in particular. However, while LLMs have achieved excellent performance on medical licensing exams, these tests fail to assess many skills necessary for deployment in a realistic clinical decision-making environment, including gathering information, adhering to guidelines, and integrating into clinical workflows. Here we have created a curated dataset based on the Medical Information Mart for Intensive Care database spanning 2,400 real patient cases and four common abdominal pathologies as well as a framework to simulate a realistic clinical setting. We show that current state-of-the-art LLMs do not accurately diagnose patients across all pathologies (performing significantly worse than physicians), follow neither diagnostic nor treatment guidelines, and cannot interpret laboratory results, thus posing a serious risk to the health of patients. Furthermore, we move beyond diagnostic accuracy and demonstrate that they cannot be easily integrated into existing workflows because they often fail to follow instructions and are sensitive to both the quantity and order of information. Overall, our analysis reveals that LLMs are currently not ready for autonomous clinical decision-making while providing a dataset and framework to guide future studies.

Open Access Data and Deep Learning for Cardiac Device Identification on Standard DICOM and Smartphone-based Chest Radiographs.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop and evaluate a publicly available deep learning model for segmenting and classifying cardiac implantable electronic devices (CIEDs) on Digital Imaging and Communications in Medicine (DICOM) and smartphone-based chest radiograph (CXR) images. Materials and Methods This institutional review board-approved retrospective study included patients with implantable pacemakers, cardioverter defibrillators, cardiac resynchronization therapy devices, and cardiac monitors who underwent chest radiography between January 2012 and January 2022. A U-Net model with a ResNet-50 backbone was created to classify CIEDs on DICOM and smartphone images. Using 2,321 CXRs from 897 patients (median age, 76 years (range 18-96 years); 625 male, 272 female), CIEDs were categorized into four manufacturers, 27 models, and one 'other' category. Five smartphones were used to acquire 11,072 images. Performance was reported using the Dice coefficient on the validation set for segmentation or balanced accuracy on the test set for manufacturer and model classification, respectively. Results The segmentation tool achieved a mean Dice coefficient of 0.936 (IQR: 0.890-0.958). The model had an accuracy of 94.36% (95% CI: 90.93%-96.84%; n = 251/266) for CIED manufacturer classification and 84.21% (95% CI: 79.31%-88.30%; n = 224/266) for CIED model classification. Conclusion The proposed deep learning model, trained on both traditional DICOM and smartphone images, showed high accuracy for segmentation and classification of CIEDs on CXRs. ©RSNA, 2024.

Impact of baseline adipose tissue characteristics on change in adipose tissue volume during a low calorie diet in people with obesity-results from the LION study.

BACKGROUND/OBJECTIVES: Weight loss outcomes vary individually. Magnetic resonance imaging (MRI)-based evaluation of adipose tissue (AT) might help to identify AT characteristics that predict AT loss. This study aimed to assess the impact of an 8-week low-calorie diet (LCD) on different AT depots and to identify predictors of short-term AT loss using MRI in adults with obesity. METHODS: Eighty-one adults with obesity (mean BMI 34.08 ± 2.75 kg/m², mean age 46.3 ± 10.97 years, 49 females) prospectively underwent baseline MRI (liver dome to femoral head) and anthropometric measurements (BMI, waist-to-hip-ratio, body fat), followed by a post-LCD-examination. Visceral and subcutaneous AT (VAT and SAT) volumes and AT fat fraction were extracted from the MRI data. Apparent lipid volumes based on MRI were calculated as approximation for the lipid contained in the AT. SAT and VAT volumes were subdivided into equidistant thirds along the craniocaudal axis and normalized by length of the segmentation. T-tests compared baseline and follow-up measurements and sex differences. Effect sizes on subdivided AT volumes were compared. Spearman Rank correlation explored associations between baseline parameters and AT loss. Multiple regression analysis identified baseline predictors for AT loss. RESULTS: Following the LCD, participants exhibited significant weight loss (11.61 ± 3.07 kg, p < 0.01) and reductions in all MRI-based AT parameters (p < 0.01). Absolute SAT loss exceeded VAT loss, while relative apparent lipid loss was higher in VAT (both p < 0.01). The lower abdominopelvic third showed the most significant SAT and VAT reduction. The predictor of most AT and apparent lipid losses was the normalized baseline SAT volume in the lower abdominopelvic third, with smaller volumes favoring greater AT loss (p < 0.01 for SAT and VAT loss and SAT apparent lipid volume loss). CONCLUSIONS: The LCD primarily reduces lower abdominopelvic SAT and VAT. Furthermore, lower abdominopelvic SAT volume was detected as a potential predictor for short-term AT loss in persons with obesity.

Silicone implant and fibrous capsule assessment based on water-fat-silicone chemical shift encoding-based species separation in breast MRI.

BACKGROUND: With rising breast augmentations worldwide, there is an increasing clinical need for an early and accurate detection of implant complications. PURPOSE: To compare the quality of chemical shift encoding-based (CSE) water-fat-silicone separation compared to double inversion recovery (DIR) silicone-only imaging in breast magnetic resonance imaging (MRI). MATERIAL AND METHODS: This retrospective, single-center study included women with silicone implants subjected to 3-T MRI between January 2021 and March 2022. MRI included (i) two-dimensional silicone-only T2-weighted turbo spin echo DIR acquisition and (ii) three-dimensional CSE imaging based on multi-echo gradient-echo sequence enabling water-, fat-, and silicone-image separation. Images were evaluated and compared by three independent radiologists using a clinically established rating including differentiability of the silicone implant, visibility and contouring of the adjacent fibrous capsule, and accuracy of intralesional folds in a ranking of 1-5. The apparent contrast-to-noise (aCNR) was calculated. RESULTS: In 71 women, the average quality of water-fat-silicone images from CSE imaging was assessed as "good" (assessment 4 ± 0.9). In 68 (96%) patients, CSE imaging achieved a concise delineation of the silicone implant and precise visualization of the fibrous capsule that was not distinguishable in DIR imaging. Implant ruptures were more easily detected in CSE imaging. The aCNR was higher in CSE compared to DIR imaging (18.43 ± 9.8 vs. 14.73 ± 2.5; P = 0.002). CONCLUSION: Intrinsically co-registered water-fat-silicone-separated CSE-based images enable a reliable assessment of silicone implants. The simultaneously improved differentiability of the implant and fibrous capsule may provide clinicians with a valuable tool for an accurate evaluation of implant integrity and early detection of potential complications.

Partial Fourier in the presence of respiratory motion in prostate diffusion-weighted echo planar imaging.

PURPOSE: To investigate the effect of respiratory motion in terms of signal loss in prostate diffusion-weighted imaging (DWI), and to evaluate the usage of partial Fourier in a free-breathing protocol in a clinically relevant b-value range using both single-shot and multi-shot acquisitions. METHODS: A controlled breathing DWI acquisition was first employed at 3 T to measure signal loss from deep breathing patterns. Single-shot and multi-shot (2-shot) acquisitions without partial Fourier (no pF) and with partial Fourier (pF) factors of 0.75 and 0.65 were employed in a free-breathing protocol. The apparent SNR and ADC values were evaluated in 10 healthy subjects to measure if low pF factors caused low apparent SNR or overestimated ADC. RESULTS: Controlled breathing experiments showed a difference in signal coefficient of variation between shallow and deep breathing. In free-breathing single-shot acquisitions, the pF 0.65 scan showed a significantly (p < 0.05) higher apparent SNR than pF 0.75 and no pF in the peripheral zone (PZ) of the prostate. In the multi-shot acquisitions in the PZ, pF 0.75 had a significantly higher apparent SNR than 0.65 pF and no pF. The single-shot pF 0.65 scan had a significantly lower ADC than single-shot no pF. CONCLUSION: Deep breathing patterns can cause intravoxel dephasing in prostate DWI. For single-shot acquisitions at a b-value of 800 s/mm2, any potential risks of motion-related artefacts at low pF factors (pF 0.65) were outweighed by the increase in signal from a lower TE, as shown by the increase in apparent SNR. In multi-shot acquisitions however, the minimum pF factor should be larger, as shown by the lower apparent SNR at low pF factors.

Integrating Text and Image Analysis: Exploring GPT-4V's Capabilities in Advanced Radiological Applications Across Subspecialties.

This study demonstrates that GPT-4V outperforms GPT-4 across radiology subspecialties in analyzing 207 cases with 1312 images from the Radiological Society of North America Case Collection.

Risk-adjusted Screening for Prostate Cancer-Defining the Low-risk Group by Data from the PROBASE Trial.

BACKGROUND: Risk-adjusted screening for prostate cancer (PCa) aims to reduce harms by less frequent retesting, especially in men at a low risk of PCa. Definitions of low risk are based mainly on studies in men starting screening at age 55-60 yr. OBJECTIVE: To identify men at age 45 yr with a low risk of PCa. DESIGN, SETTING, AND PARTICIPANTS: A population-based, risk-adjusted PCa screening trial was conducted in Germany using baseline prostate-specific antigen (PSA) starting in young men (PROBASE). INTERVENTION: PSA measurements starting at the age of 45 yr. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The incidence of PCa within 5 yr was assessed in men with screen-negative baseline PSA <1.5 ng/ml compared with those with PSA 1.5-≤3.0 ng/ml. RESULTS AND LIMITATIONS: Of 23301 men who received a first PSA test at age 45 yr, 0.79% had a screen-positive PSA value of ≥3 ng/ml. Among the 89% of men who had a screen-negative baseline PSA value of <1.5 ng/ml, only 0.45% received a positive PSA test ≥3 ng/ml upon retesting after 5 yr. By contrast, for those with a screen-negative baseline PSA value of 1.5-3 ng/ml, 13% surpassed 3 ng/ml upon biennial testing within the next 4 yr. The incidence of PCa in subsequent screening rounds increased with increasing baseline PSA levels, from 0.13 per 1000 person-years for men with initial PSA level of <1.5 ng/ml to 8.0 per 1000 person-years for those with PSA levels of 1.5-3.0 ng/ml. A limitation is a follow-up time of only 5 yr, so far. CONCLUSIONS: Men with baseline PSA <1.5 ng/ml at age 45 yr are at a very low risk of PCa over the next 5 yr. PATIENT SUMMARY: The PROBASE study showed that men with baseline prostate-specific antigen (PSA) <1.5 ng/ml at age 45 yr have a very low prostate cancer detection rate over 5 yr and do not need PSA retesting during this time.

Sensitivity of magnetic resonance elastography to extracellular matrix and cell motility in human prostate cancer cell line-derived xenograft models.

Prostate cancer (PCa) is a significant health problem in the male population of the Western world. Magnetic resonance elastography (MRE), an emerging medical imaging technique sensitive to mechanical properties of biological tissues, detects PCa based on abnormally high stiffness and viscosity values. Yet, the origin of these changes in tissue properties and how they correlate with histopathological markers and tumor aggressiveness are largely unknown, hindering the use of tumor biomechanical properties for establishing a noninvasive PCa staging system. To infer the contributions of extracellular matrix (ECM) components and cell motility, we investigated fresh tissue specimens from two PCa xenograft mouse models, PC3 and LNCaP, using magnetic resonance elastography (MRE), diffusion-weighted imaging (DWI), quantitative histology, and nuclear shape analysis. Increased tumor stiffness and impaired water diffusion were observed to be associated with collagen and elastin accumulation and decreased cell motility. Overall, LNCaP, while more representative of clinical PCa than PC3, accumulated fewer ECM components, induced less restriction of water diffusion, and exhibited increased cell motility, resulting in overall softer and less viscous properties. Taken together, our results suggest that prostate tumor stiffness increases with ECM accumulation and cell adhesion - characteristics that influence critical biological processes of cancer development. MRE paired with DWI provides a powerful set of imaging markers that can potentially predict prostate tumor development from benign masses to aggressive malignancies in patients. STATEMENT OF SIGNIFICANCE: Xenograft models of human prostate tumor cell lines, allowing correlation of microstructure-sensitive biophysical imaging parameters with quantitative histological methods, can be investigated to identify hallmarks of cancer.

Motion-induced phase-corrected homodyne reconstruction for partial Fourier single-shot diffusion-weighted echo planar imaging of the liver.

Partial Fourier encoding is popular in single-shot (ss) diffusion-weighted (DW) echo planar imaging (EPI) because it enables a shorter echo time (TE) and, hence, improves the signal-to-noise-ratio. Motion during diffusion encoding causes k-space shifting and dispersion, which compromises the quality of the homodyne reconstruction. This work provides a comprehensive understanding of the artifacts in homodyne reconstruction of partial Fourier ss-DW-EPI data in the presence of motion-induced phase and proposes the motion-induced phase-corrected homodyne (mpc-hdyne) reconstruction method to ameliorate these artifacts. Simulations with different types of motion-induced phase were performed to provide an understanding of the potential artifacts that occur in the homodyne reconstruction of partial Fourier ss-DW-EPI data. To correct for the artifacts, the mpc-hdyne reconstruction is proposed. The algorithm recenters k-space, updates the partial Fourier factor according to detected global k-space shifts, and removes low-resolution nonlinear phase before the conventional homodyne reconstruction. The mpc-hdyne reconstruction is tested on both simulation and in vivo data. Motion-induced phase can cause signal overestimation, worm artifacts, and signal loss in partial Fourier ss-DW-EPI data with the conventional homodyne reconstruction. Simulation and in vivo data showed that the proposed mpc-hdyne reconstruction ameliorated artifacts, yielding higher quality DW images compared with conventional homodyne reconstruction. Based on the understanding of the artifacts in homodyne reconstruction of partial Fourier ss-DW-EPI data, the mpc-hdyne reconstruction was proposed and showed superior performance compared with the conventional homodyne reconstruction on both simulation and in vivo data.

Comparison of Virtual Non-Contrast and True Non-Contrast CT Images Obtained by Dual-Layer Spectral CT in COPD Patients.

Chronic obstructive pulmonary disease (COPD) is one of the leading causes of death. Recent studies have underlined the importance of non-contrast-enhanced chest CT scans not only for emphysema progression quantification, but for correlation with clinical outcomes as well. As about 40 percent of the 300 million CT scans per year are contrast-enhanced, no proper emphysema quantification is available in a one-stop-shop approach for patients with known or newly diagnosed COPD. Since the introduction of spectral imaging (e.g., dual-energy CT scanners), it has been possible to create virtual non-contrast-enhanced images (VNC) from contrast-enhanced images, making it theoretically possible to offer proper COPD imaging despite contrast enhancing. This study is aimed towards investigating whether these VNC images are comparable to true non-contrast-enhanced images (TNC), thereby reducing the radiation exposure of patients and usage of resources in hospitals. In total, 100 COPD patients with two scans, one with (VNC) and one without contrast media (TNC), within 8 weeks or less obtained by a spectral CT using dual-layer technology, were included in this retrospective study. TNC and VNC were compared according to their voxel-density histograms. While the comparison showed significant differences in the low attenuated volumes (LAVs) of TNC and VNC regarding the emphysema threshold of -950 Houndsfield Units (HU), the 15th and 10th percentiles of the LAVs used as a proxy for pre-emphysema were comparable. Upon further investigation, the threshold-based LAVs (-950 HU) of TNC and VNC were comparable in patients with a water equivalent diameter (DW) below 270 mm. The study concludes that VNC imaging may be a viable option for assessing emphysema progression in COPD patients, particularly those with a normal body mass index (BMI). Further, pre-emphysema was generally comparable between TNC and VNC. This approach could potentially reduce radiation exposure and hospital resources by making additional TNC scans obsolete.

Multimaterial decomposition in dual-energy CT for characterization of clots from acute ischemic stroke patients.

BACKGROUND: Nowadays, there is no method to quantitatively characterize the material composition of acute ischemic stroke thrombi prior to intervention, but dual-energy CT (DE-CT) offers imaging-based multimaterial decomposition. We retrospectively investigated the material composition of thrombi ex vivo using DE-CT with histological analysis as a reference. METHODS: Clots of 70 patients with acute ischemic stroke were extracted by mechanical thrombectomy and scanned ex vivo in formalin-filled tubes with DE-CT. Multimaterial decomposition in the three components, i.e., red blood cells (RBC), white blood cells (WBC), and fibrin/platelets (F/P), was performed and compared to histology (hematoxylin/eosin staining) as reference. Attenuation and effective Z values were assessed, and histological composition was compared to stroke etiology according to the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) criteria. RESULTS: Histological and imaging analysis showed the following correlation coefficients for RBC (r = 0.527, p < 0.001), WBC (r = 0.305, p = 0.020), and F/P (r = 0.525, p < 0.001). RBC-rich thrombi presented higher clot attenuation in Hounsfield units than F/P-rich thrombi (51 HU versus 42 HU, p < 0.01). In histological analysis, cardioembolic clots showed less RBC (40% versus 56%, p = 0.053) and more F/P (53% versus 36%, p = 0.024), similar to cryptogenic clots containing less RBC (34% versus 56%, p = 0.006) and more F/P (58% versus 36%, p = 0.003) than non-cardioembolic strokes. No difference was assessed for the mean WBC portions in all TOAST groups. CONCLUSIONS: DE-CT has the potential to quantitatively characterize the material composition of ischemic stroke thrombi. RELEVANCE STATEMENT: Using DE-CT, the composition of ischemic stroke thrombi can be determined. Knowledge of histological composition prior to intervention offers the opportunity to define personalized treatment strategies for each patient to accomplish faster recanalization and better clinical outcomes. KEY POINTS: • Acute ischemic stroke clots present different recanalization success according to histological composition. • Currently, no method can determine clot composition prior to intervention. • DE-CT allows quantitative material decomposition of thrombi ex vivo in red blood cells, white blood cells, and fibrin/platelets. • Histological clot composition differs between stroke etiology. • Insights into the histological composition in situ offer personalized treatment strategies.

Deep Learning to Differentiate Benign and Malignant Vertebral Fractures at Multidetector CT.

Background Differentiating between benign and malignant vertebral fractures poses diagnostic challenges. Purpose To investigate the reliability of CT-based deep learning models to differentiate between benign and malignant vertebral fractures. Materials and Methods CT scans acquired in patients with benign or malignant vertebral fractures from June 2005 to December 2022 at two university hospitals were retrospectively identified based on a composite reference standard that included histopathologic and radiologic information. An internal test set was randomly selected, and an external test set was obtained from an additional hospital. Models used a three-dimensional U-Net encoder-classifier architecture and applied data augmentation during training. Performance was evaluated using the area under the receiver operating characteristic curve (AUC) and compared with that of two residents and one fellowship-trained radiologist using the DeLong test. Results The training set included 381 patients (mean age, 69.9 years ± 11.4 [SD]; 193 male) with 1307 vertebrae (378 benign fractures, 447 malignant fractures, 482 malignant lesions). Internal and external test sets included 86 (mean age, 66.9 years ± 12; 45 male) and 65 (mean age, 68.8 years ± 12.5; 39 female) patients, respectively. The better-performing model of two training approaches achieved AUCs of 0.85 (95% CI: 0.77, 0.92) in the internal and 0.75 (95% CI: 0.64, 0.85) in the external test sets. Including an uncertainty category further improved performance to AUCs of 0.91 (95% CI: 0.83, 0.97) in the internal test set and 0.76 (95% CI: 0.64, 0.88) in the external test set. The AUC values of residents were lower than that of the best-performing model in the internal test set (AUC, 0.69 [95% CI: 0.59, 0.78] and 0.71 [95% CI: 0.61, 0.80]) and external test set (AUC, 0.70 [95% CI: 0.58, 0.80] and 0.71 [95% CI: 0.60, 0.82]), with significant differences only for the internal test set (P < .001). The AUCs of the fellowship-trained radiologist were similar to those of the best-performing model (internal test set, 0.86 [95% CI: 0.78, 0.93; P = .39]; external test set, 0.71 [95% CI: 0.60, 0.82; P = .46]). Conclusion Developed models showed a high discriminatory power to differentiate between benign and malignant vertebral fractures, surpassing or matching the performance of radiology residents and matching that of a fellowship-trained radiologist. © RSNA, 2024 See also the editorial by Booz and D'Angelo in this issue.

Evaluation of Conventional MR Imaging of the Shoulder in the Diagnosis of Lesions of the Biceps Pulley.

RATIONALE AND OBJECTIVES: To determine the diagnostic accuracy and reproducibility of conventional MR imaging (MRI) of the shoulder in evaluating biceps pulley lesions using arthroscopy as the standard of reference. METHODS: In a retrospective study, MR examinations of 68 patients with arthroscopically proven torn or intact biceps pulley were assessed for the presence of pulley lesions by three radiologists. The following criteria were evaluated: displacement of the long head of the biceps tendon (LHBT) relative to the subscapularis tendon (displacement sign), subluxation/dislocation of the LHBT, the integrity of the superior glenohumeral ligament (SGHL) and the coracohumeral ligament (CHL), lesions of the supraspinatus (SSP) and subscapularis (SSC) tendons adjacent to the rotator interval, presence of biceps tendinopathy and subacromial bursitis. RESULTS: There were 42 patients with pulley lesions in the study group. Conventional MR imaging showed an overall sensitivity of 95.2%, 88.1% and 92.9%, a specificity of 61.5%, 73.1%, and 80.8% and an accuracy of 82.4%, 82.4% and 88.2% in the diagnosis of pulley lesions. Interobserver agreement was substantial (multirater k = 0.75). Biceps tendinopathy (97.6%, 95.2%, 97.6%), defects of the SGHL (86.3%, 81.0%, 88.1%) and the displacement sign (88.1%, 81.0%, 85.7%) were the most sensitive diagnostic criteria. Subluxation/dislocation of the LHBT was insensitive (78.6%, 42.9%, 33.3%), but specific (69.2%, 100,0%, 96.2%). CONCLUSION: In the diagnosis of pulley lesions, conventional MR imaging is reproducible and shows high sensitivity and accuracy but moderate specificity.

Preserving fairness and diagnostic accuracy in private large-scale AI models for medical imaging.

BACKGROUND: Artificial intelligence (AI) models are increasingly used in the medical domain. However, as medical data is highly sensitive, special precautions to ensure its protection are required. The gold standard for privacy preservation is the introduction of differential privacy (DP) to model training. Prior work indicates that DP has negative implications on model accuracy and fairness, which are unacceptable in medicine and represent a main barrier to the widespread use of privacy-preserving techniques. In this work, we evaluated the effect of privacy-preserving training of AI models regarding accuracy and fairness compared to non-private training. METHODS: We used two datasets: (1) A large dataset (N = 193,311) of high quality clinical chest radiographs, and (2) a dataset (N = 1625) of 3D abdominal computed tomography (CT) images, with the task of classifying the presence of pancreatic ductal adenocarcinoma (PDAC). Both were retrospectively collected and manually labeled by experienced radiologists. We then compared non-private deep convolutional neural networks (CNNs) and privacy-preserving (DP) models with respect to privacy-utility trade-offs measured as area under the receiver operating characteristic curve (AUROC), and privacy-fairness trade-offs, measured as Pearson's r or Statistical Parity Difference. RESULTS: We find that, while the privacy-preserving training yields lower accuracy, it largely does not amplify discrimination against age, sex or co-morbidity. However, we find an indication that difficult diagnoses and subgroups suffer stronger performance hits in private training. CONCLUSIONS: Our study shows that - under the challenging realistic circumstances of a real-life clinical dataset - the privacy-preserving training of diagnostic deep learning models is possible with excellent diagnostic accuracy and fairness.

Artificial intelligence (AI), in which computers can learn to do tasks that normally require human intelligence, is particularly useful in medical imaging. However, AI should be used in a way that preserves patient privacy. We explored the balance between maintaining patient data privacy and AI performance in medical imaging. We use an approach called differential privacy to protect the privacy of patients' images. We show that, although training AI with differential privacy leads to a slight decrease in accuracy, it does not substantially increase bias against different age groups, genders, or patients with multiple health conditions. However, we notice that AI faces more challenges in accurately diagnosing complex cases and specific subgroups when trained under these privacy constraints. These findings highlight the importance of designing AI systems that are both privacy-conscious and capable of reliable diagnoses across patient groups.

Artificial intelligence support in MR imaging of incidental renal masses: an early health technology assessment.

OBJECTIVE: This study analyzes the potential cost-effectiveness of integrating an artificial intelligence (AI)-assisted system into the differentiation of incidental renal lesions as benign or malignant on MR images during follow-up. MATERIALS AND METHODS: For estimation of quality-adjusted life years (QALYs) and lifetime costs, a decision model was created, including the MRI strategy and MRI + AI strategy. Model input parameters were derived from recent literature. Willingness to pay (WTP) was set to $100,000/QALY. Costs of $0 for the AI were assumed in the base-case scenario. Model uncertainty and costs of the AI system were assessed using deterministic and probabilistic sensitivity analysis. RESULTS: Average total costs were at $8054 for the MRI strategy and $7939 for additional use of an AI-based algorithm. The model yielded a cumulative effectiveness of 8.76 QALYs for the MRI strategy and of 8.77 for the MRI + AI strategy. The economically dominant strategy was MRI + AI. Deterministic and probabilistic sensitivity analysis showed high robustness of the model with the incremental cost-effectiveness ratio (ICER), which represents the incremental cost associated with one additional QALY gained, remaining below the WTP for variation of the input parameters. If increasing costs for the algorithm, the ICER of $0/QALY was exceeded at $115, and the defined WTP was exceeded at $667 for the use of the AI. CONCLUSIONS: This analysis, rooted in assumptions, suggests that the additional use of an AI-based algorithm may be a potentially cost-effective alternative in the differentiation of incidental renal lesions using MRI and needs to be confirmed in the future. CLINICAL RELEVANCE STATEMENT: These results hint at AI's the potential impact on diagnosing renal masses. While the current study urges careful interpretation, ongoing research is essential to confirm and seamlessly integrate AI into clinical practice, ensuring its efficacy in routine diagnostics. KEY POINTS: • This is a model-based study using data from literature where AI has been applied in the diagnostic workup of incidental renal lesions. • MRI + AI has the potential to be a cost-effective alternative in the differentiation of incidental renal lesions. • The additional use of AI can reduce costs in the diagnostic workup of incidental renal lesions.

Spectral computed tomography angiography using a gadolinium-based contrast agent for imaging of pathologies of the aorta.

OBJECTIVES: Especially patients with aortic aneurysms and multiple computed tomography angiographies (CTA) might show medical conditions which oppose the use of iodine-based contrast agents. CTA using monoenergetic reconstructions from dual layer CT and gadolinium (Gd-)based contrast agents might be a feasible alternative in these patients. Therefore, the purpose of this study was to evaluate the feasibility of clinical spectral CTA with a Gd-based contrast agent in patients with aortic aneurysms. METHODS: Twenty-one consecutive scans in 15 patients with and without endovascular aneurysm repair showing contraindications for iodine-based contrast agents were examined using clinical routine doses (0.2 mmol/kg) of Gd-based contrast agent with spectral CT. Monoenergetic reconstructions of the spectral data set were computed. RESULTS: There was a significant increase in the intravascular attenuation of the aorta between pre- and post-contrast images for the MonoE40 images in the thoracic and the abdominal aorta (p < 0.001 for both). Additionally, the ratio between pre- and post-contrast images was significantly higher in the MonoE40 images as compared to the conventional images with a factor of 6.5 ± 4.5 vs. 2.4 ± 0.5 in the thoracic aorta (p = 0.003) and 4.1 ± 1.8 vs. 1.9 ± 0.5 in the abdominal aorta (p < 0.001). CONCLUSIONS: To conclude, our study showed that Gd-CTA is a valid and reliable alternative for diagnostic imaging of the aorta for clinical applications. Monoenergetic reconstructions of computed tomography angiographies using gadolinium based contrast agents may be a useful alternative in patients with aortic aneurysms and contraindications for iodine based contrast agents.

Characterization of bone marrow lesions in axial spondyloarthritis using quantitative T1 mapping MRI.

OBJECTIVE: Conventional magnetic resonance imaging (MRI) uses T1-weighted and short-tau inversion recovery (STIR) sequences to characterize bone marrow in axial spondyloarthritis. However, quantification is restricted to estimating the extent of lesions because signal intensities are highly variable both within individuals and across patients and MRI scanners. This study evaluates the performance of quantitative T1 mapping for distinguishing different types of bone marrow lesions of the sacroiliac joints. MATERIALS AND METHODS: In this prospective study, 62 patients underwent computed tomography (CT) and MRI of the sacroiliac joints including T1, STIR, and T1 mapping. Bone marrow lesions were characterized by three readers and assigned to one of four groups: sclerosis, osteitis, fat lesions, and mixed marrow lesions. Relaxation times on T1 maps were compared using generalized estimating equations and receiver operating characteristics (ROC) analysis. RESULTS: A total of 119 lesions were selected (sclerosis: 38, osteitis: 27, fat lesions: 40; mixed lesions: 14). T1 maps showed highly significant differences between the lesions with the lowest values for sclerosis (1516±220 ms), followed by osteitis (1909±75 ms), and fat lesions (2391±200 ms); p<0.001. T1 mapping differentiated lesions with areas under the ROC curve of 99% (sclerosis vs. osteitis) and 100% (other comparisons). CONCLUSION: T1 mapping allows accurate characterization of sclerosis, osteitis, and fat lesions at the sacroiliac joint but only for homogeneous, non-mixed lesions. Thus, further sequence development is needed before implementation in clinical routine.

Accelerated liver water T1 mapping using single-shot continuous inversion-recovery spiral imaging.

PURPOSE: Liver T1 mapping techniques typically require long breath holds or long scan time in free-breathing, need correction for B 1 + inhomogeneities and process composite (water and fat) signals. The purpose of this work is to accelerate the multi-slice acquisition of liver water selective T1 (wT1) mapping in a single breath hold, improving the k-space sampling efficiency. METHODS: The proposed continuous inversion-recovery (IR) Look-Locker methodology combines a single-shot gradient echo spiral readout, Dixon processing and a dictionary-based analysis for liver wT1 mapping at 3 T. The sequence parameters were adapted to obtain short scan times. The influence of fat, B 1 + inhomogeneities and TE on the estimation of T1 was first assessed using simulations. The proposed method was then validated in a phantom and in 10 volunteers, comparing it with MRS and the modified Look-Locker inversion-recovery (MOLLI) method. Finally, the clinical feasibility was investigated by comparing wT1 maps with clinical scans in nine patients. RESULTS: The phantom results are in good agreement with MRS. The proposed method encodes the IR-curve for the liver wT1 estimation, is minimally sensitive to B 1 + inhomogeneities and acquires one slice in 1.2 s. The volunteer results confirmed the multi-slice capability of the proposed method, acquiring nine slices in a breath hold of 11 s. The present work shows robustness to B 1 + inhomogeneities ( wT 1 , No B 1 + = 1.07 wT 1 , B 1 + - 45.63 , R 2 = 0.99 ) , good repeatability ( wT 1 , 2 ° = 1 . 0 wT 1 , 1 ° - 2.14 , R 2 = 0.96 ) and is in better agreement with MRS ( wT 1 = 0.92 wT 1 MRS + 103.28 , R 2 = 0.38 ) than is MOLLI ( wT 1 MOLLI = 0.76 wT 1 MRS + 254.43 , R 2 = 0.44 ) . The wT1 maps in patients captured diverse lesions, thus showing their clinical feasibility. CONCLUSION: A single-shot spiral acquisition can be combined with a continuous IR Look-Locker method to perform rapid repeatable multi-slice liver water T1 mapping at a rate of 1.2 s per slice without a B 1 + map. The proposed method is suitable for nine-slice liver clinical applications acquired in a single breath hold of 11 s.

Evaluating Treatment Response in GEJ Adenocarcinoma: The Role of Pretherapeutic and Posttherapeutic Iodine Mapping.

BACKGROUND: Neoadjuvant therapy regimens have significantly improved the prognosis of GEJ (gastroesophageal junction) cancer; however, there are a significant percentage of patients who benefit from earlier resection or adapted therapy regimens, and the true response rate can only be determined histopathologically. Methods that allow preoperative assessment of response are lacking. PURPOSE: The purpose of this retrospective study is to assess the potential of pretherapeutic and posttherapeutic spectral CT iodine density (IoD) in predicting histopathological response to neoadjuvant chemotherapy in patients diagnosed with adenocarcinoma of the GEJ. METHODS: In this retrospective cohort study, a total of 62 patients with GEJ carcinoma were studied. Patients received a multiphasic CT scan at diagnosis and preoperatively. Iodine-density maps were generated based on spectral CT data. All tumors were histopathologically analyzed, and the tumor regression grade (TRG) according to Becker et al ( Cancer . 2003;98:1521-1530) was determined. Two experienced radiologists blindly placed 5 defined ROIs in the tumor region of highest density, and the maximum value was used for further analysis. Iodine density was normalized to the aortic iodine uptake. In addition, tumor response was assessed according to standard RECIST measurement. After assessing interrater reliability, the correlation of IoD values with treatment response and with histopathologic TRG was evaluated. RESULTS: The normalized ΔIoD (IoD at diagnosis - IoD after neoadjuvant treatment) and the normalized IoD after neoadjuvant treatment correlated significantly with the TRG. For the detection of responders and nonresponders, the receiver operating characteristic (ROC) curve for normalized ΔIoD yielded the highest area under the curve of 0.95 and achieved a sensitivity and specificity of 92.3% and 92.1%, respectively. Iodine density after neoadjuvant treatment achieved an area under the curve of 0.88 and a sensitivity and specificity of 86.8% and 84.6%, respectively (cutoff, 0.266). Iodine density at diagnosis and RECIST did not provide information to distinguish responders from nonresponders. Using the cutoff value for IoD after neoadjuvant treatment, a reliable classification of responders and nonresponders was achieved for both readers in a test set of 11 patients. Intraclass correlation coefficient revealed excellent interrater reliability (intraclass correlation coefficient, >0.9). Lastly, using the cutoff value for normalized ΔIoD as a definition for treatment response, a significantly longer survival of responders was shown. CONCLUSIONS: Changes in IoD after neoadjuvant treatment of GEJ cancer may be a potential surrogate for therapy response.

Development of an image-based Random Forest classifier for prediction of surgery duration of laparoscopic sigmoid resections.

PURPOSE: Sigmoid diverticulitis is a disease with a high socioeconomic burden, accounting for a high number of left-sided colonic resections worldwide. Modern surgical scheduling relies on accurate prediction of operation times to enhance patient care and optimize healthcare resources. This study aims to develop a predictive model for surgery duration in laparoscopic sigmoid resections, based on preoperative CT biometric and demographic patient data. METHODS: This retrospective single-center cohort study included 85 patients who underwent laparoscopic sigmoid resection for diverticular disease. Potentially relevant procedure-specific anatomical parameters recommended by a surgical expert were measured in preoperative CT imaging. After random split into training and test set (75% / 25%) multiclass logistic regression was performed and a Random Forest classifier was trained on CT imaging parameters, patient age, and sex in the training cohort to predict categorized surgery duration. The models were evaluated in the test cohort using established performance metrics including receiver operating characteristics area under the curve (AUROC). RESULTS: The Random Forest model achieved a good average AUROC of 0.78. It allowed a very good prediction of long (AUROC = 0.89; specificity 0.71; sensitivity 1.0) and short (AUROC = 0.81; specificity 0.77; sensitivity 0.56) procedures. It clearly outperformed the multiclass logistic regression model (AUROC: average = 0.33; short = 0.31; long = 0.22). CONCLUSION: A Random Forest classifier trained on demographic and CT imaging biometric patient data could predict procedure duration outliers of laparoscopic sigmoid resections. Pending validation in a multicenter study, this approach could potentially improve procedure scheduling in visceral surgery and be scaled to other procedures.