Large language models (LLMs) in radiology exams for medical students: Performance and consequences.

The evolving field of medical education is being shaped by technological advancements, including the integration of Large Language Models (LLMs) like ChatGPT. These models could be invaluable resources for medical students, by simplifying complex concepts and enhancing interactive learning by providing personalized support. LLMs have shown impressive performance in professional examinations, even without specific domain training, making them particularly relevant in the medical field. This study aims to assess the performance of LLMs in radiology examinations for medical students, thereby shedding light on their current capabilities and implications.This study was conducted using 151 multiple-choice questions, which were used for radiology exams for medical students. The questions were categorized by type and topic and were then processed using OpenAI's GPT-3.5 and GPT- 4 via their API, or manually put into Perplexity AI with GPT-3.5 and Bing. LLM performance was evaluated overall, by question type and by topic.GPT-3.5 achieved a 67.6% overall accuracy on all 151 questions, while GPT-4 outperformed it significantly with an 88.1% overall accuracy (p<0.001). GPT-4 demonstrated superior performance in both lower-order and higher-order questions compared to GPT-3.5, Perplexity AI, and medical students, with GPT-4 particularly excelling in higher-order questions. All GPT models would have successfully passed the radiology exam for medical students at our university.In conclusion, our study highlights the potential of LLMs as accessible knowledge resources for medical students. GPT-4 performed well on lower-order as well as higher-order questions, making ChatGPT-4 a potentially very useful tool for reviewing radiology exam questions. Radiologists should be aware of ChatGPT's limitations, including its tendency to confidently provide incorrect responses. · ChatGPT demonstrated remarkable performance, achieving a passing grade on a radiology examination for medical students that did not include image questions.. · GPT-4 exhibits significantly improved performance compared to its predecessors GPT-3.5 and Perplexity AI with 88% of questions answered correctly.. · Radiologists as well as medical students should be aware of ChatGPT's limitations, including its tendency to confidently provide incorrect responses.. · Gotta J, Le Hong QA, Koch V et al. Large language models (LLMs) in radiology exams for medical students: Performance and consequences. Fortschr Röntgenstr 2024; DOI 10.1055/a-2437-2067.

Dual-energy CT late arterial phase iodine maps for the diagnosis of acute non-occlusive mesenteric ischemia.

PURPOSE: To evaluate the diagnostic accuracy of dual-energy CT (DECT) iodine maps in comparison to conventional CT series for the assessment of non-occlusive mesenteric ischemia (NOMI). MATERIAL AND METHODS: We evaluated data from 142 patients (72 men; 50.7%) who underwent DECT between 2018 and 2022, with surgically confirmed diagnosis of NOMI. One board-certified radiologist performed region of interest (ROI) measurements in bowel segments on late arterial (LA) and portal venous (PV) phase DECT iodine maps as well as LA conventional series, in both ischemic and non-ischemic bowel loops, using surgical reports as reference standard, and in a control group of 97 patients. Intra- and inter-reader agreement with a second board-certified radiologist was also evaluated. Receiver operating characteristic (ROC) curve analysis was performed to calculate the optimal threshold for discriminating ischemic from non-ischemic bowel segments. Subjective image rating of LA and PV iodine maps was performed. RESULTS: DECT-based iodine concentration (IC) measurements showed significant differences in LA phase iodine maps between ischemic (median:0.72; IQR 0.52-0.91 mg/mL) and non-ischemic bowel loops (5.16; IQR 3.45-6.31 mg/ml) (P <.0001). IC quantification on LA phase revealed an AUC of 0.966 for the assessment of acute bowel ischemia, significantly higher compared to both IC quantification based on PV phase (0.951) and attenuation values evaluated on LA conventional CT series (0.828). Excellent intra-observer and strong inter-observer agreements were observed for the quantification of iodine concentration. Conversely, weak inter-observer agreement was noted for conventional HU assessments. The optimal LA phase-based IC threshold for assessing bowel ischemia was 1.34 mg/mL, yielding a sensitivity of 100% and specificity of 96.48%. CONCLUSION: Iodine maps based on LA phase significantly improve the diagnostic accuracy for the assessment of NOMI compared to conventional CT series and PV phase iodine maps.

Malignancy predictors and treatment strategies for adult intestinal intussusception.

Intussusception in adult patients is a rare medical finding, which is accompanied by an underlying tumor in some cases. However, no accepted method has been established to identify patients at risk for tumor-related intussusception. This study aimed to identify imaging features as predictors for tumor-related intussusception.CT images of patients with confirmed intussusception were retrospectively acquired between 01/2008 and 12/2022. Available follow-up images and medical health records were evaluated to identify various imaging features, the cause of intussusception, and treatment strategies. Imaging interpretation was conducted by two blinded radiologists. A third radiologist was consulted in cases of disagreement.A total of 71 consecutive patients were included in this study (42 males, 29 females) with a median age of 56 years (interquartile range: 40.5-73.8 years). Enteroenteric intussusceptions in the small bowel were the most common type observed in adult patients. In contrast, colocolic intussusception was more frequently associated with malignancy, and this association was statistically significant (p < 0.05). Among the malignant tumors, adenocarcinoma was the most common, followed by metastases and lymphoma. Additionally, bowel obstruction and wall thickening were significantly correlated with malignancy (p < 0.05). The high negative predictive values (NPVs) and high specificities for ileus (NPV 88.5%, specificity 82.1%), bowel wall thickening (NPV 90.9%, specificity 71.4%), and acute abdomen (NPV 84.6%, specificity 78.8%) suggest that the absence of these features strongly predicts a low probability of malignancy in cases of adult intussusception.Active surveillance with follow-up exams is suitable for asymptomatic and transient intussusception when imaging features suggest a low likelihood of a neoplasm. Additionally, malignancy predictors such as ileus and thickening of the bowel wall in the affected segment could guide tailored treatment. Surgical interventions are essential for symptomatic cases, with adenocarcinoma being the most common malignancy found in colocolic intussusceptions.Intussusception in adults is rare and is often associated with underlying tumors, particularly in colocolic intussusceptions. Key imaging predictors for malignancy include bowel obstruction, wall thickening in the affected segment, and the presence of acute abdomen, with high NPVs and specificities indicating low malignancy risk when these features are absent. Active surveillance is recommended for asymptomatic cases with low neoplasm probability, while surgical intervention is the method of choice for symptomatic patients. · Reschke P, Le Hong QA, Gruenewald LD et al. Malignancy predictors and treatment strategies for adult intestinal intussusception. Fortschr Röntgenstr 2024; DOI 10.1055/a-2434-7932.

Standardization of Dual-Energy CT Iodine Uptake of the Abdomen and Pelvis: Defining Reference Values in a Big Data Cohort.

Background: To establish dual-energy-derived iodine density reference values in abdominopelvic organs in a large cohort of healthy subjects. Methods: 597 patients who underwent portal venous phase dual-energy CT scans of the abdomen were retrospectively enrolled. Iodine distribution maps were reconstructed, and regions of interest measurements were placed in abdominal and pelvic structures to obtain absolute iodine values. Subsequently, normalization of the abdominal aorta was conducted to obtain normalized iodine ratios. The values obtained were subsequently analyzed and differences were investigated in subgroups defined by sex, age and BMI. Results: Overall mean iodine uptake values and normalized iodine ratios ranged between 0.31 and 6.08 mg/mL and 0.06 and 1.20, respectively. Women exhibited higher absolute iodine concentration across all organs. With increasing age, normalized iodine ratios mostly tend to decrease, being most significant in the uterus, prostate, and kidneys (p < 0.015). BMI was the parameter less responsible for variations in iodine concentrations; normal weighted patients demonstrated higher values of both absolute and normalized iodine. Conclusions: Iodine concentration values and normalized iodine ratios of abdominal and pelvic organs reveal significant gender-, age-, and BMI-related differences, underscoring the necessity to integrate these variables into clinical practice.

Carotid artery assessment in dual-source photon-counting CT: impact of low-energy virtual monoenergetic imaging on image quality, vascular contrast and diagnostic assessability.

PURPOSE: Preliminary dual-energy CT studies have shown that low-energy virtual monoenergetic (VMI) + reconstructions can provide superior image quality compared to standard 120 kV CTA series. The purpose of this study is to evaluate the impact of low-energy VMI reconstructions on quantitative and qualitative image quality, vascular contrast, and diagnostic assessability of the carotid artery in patients undergoing photon-counting CTA examinations. MATERIALS AND METHODS: A total of 122 patients (67 male) who had undergone dual-source photon-counting CTA scans of the carotid artery were retrospectively analyzed in this study. Standard 120 kV CT images and low-keV VMI series from 40 to 100 keV with an interval of 15 keV were reconstructed. Quantitative analyses included the evaluation of vascular CT numbers, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). CT number measurements were performed in the common, external, and internal carotid arteries. Qualitative analyses were performed by three board-certified radiologists independently using five-point scales to evaluate image quality, vascular contrast, and diagnostic assessability of the carotid artery. RESULTS: Mean attenuation, CNR and SNR values were highest in 40 keV VMI reconstructions (HU, 1362.32 ± 457.81; CNR, 33.19 ± 12.86; SNR, 34.37 ± 12.89) followed by 55-keV VMI reconstructions (HU, 736.94 ± 150.09; CNR, 24.49 ± 7.11; SNR, 26.25 ± 7.34); all three mean values at these keV levels were significantly higher compared with the remaining VMI series and standard 120 kV CT series (HU, 154.43 ± 23.69; CNR, 16.34 ± 5.47; SNR, 24.44 ± 7.14) (p < 0.0001). The qualitative analysis showed the highest rating scores for 55 keV VMI reconstructions followed by 40 keV and 70 keV VMI series with a significant difference compared to standard 120 kV CT images series regarding image quality, vascular contrast, and diagnostic assessability of the carotid artery (all comparisons, p < 0.01). CONCLUSIONS: Low-keV VMI reconstructions at a level of 40-55 keV significantly improve image quality, vascular contrast, and the diagnostic assessability of the carotid artery compared with standard CT series in photon-counting CTA.

Accuracy and time efficiency of a novel deep learning algorithm for Intracranial Hemorrhage detection in CT Scans.

PURPOSE: To evaluate a deep learning-based pipeline using a Dense-UNet architecture for the assessment of acute intracranial hemorrhage (ICH) on non-contrast computed tomography (NCCT) head scans after traumatic brain injury (TBI). MATERIALS AND METHODS: This retrospective study was conducted using a prototype algorithm that evaluated 502 NCCT head scans with ICH in context of TBI. Four board-certified radiologists evaluated in consensus the CT scans to establish the standard of reference for hemorrhage presence and type of ICH. Consequently, all CT scans were independently analyzed by the algorithm and a board-certified radiologist to assess the presence and type of ICH. Additionally, the time to diagnosis was measured for both methods. RESULTS: A total of 405/502 patients presented ICH classified in the following types: intraparenchymal (n = 172); intraventricular (n = 26); subarachnoid (n = 163); subdural (n = 178); and epidural (n = 15). The algorithm showed high diagnostic accuracy (91.24%) for the assessment of ICH with a sensitivity of 90.37% and specificity of 94.85%. To distinguish the different ICH types, the algorithm had a sensitivity of 93.47% and a specificity of 99.79%, with an accuracy of 98.54%. To detect midline shift, the algorithm had a sensitivity of 100%. In terms of processing time, the algorithm was significantly faster compared to the radiologist's time to first diagnosis (15.37 ± 1.85 vs 277 ± 14 s, p < 0.001). CONCLUSION: A novel deep learning algorithm can provide high diagnostic accuracy for the identification and classification of ICH from unenhanced CT scans, combined with short processing times. This has the potential to assist and improve radiologists' ICH assessment in NCCT scans, especially in emergency scenarios, when time efficiency is needed.

Biomechanical Assessment of Liver Integrity: Prospective Evaluation of Mechanical Versus Acoustic MR Elastography.

BACKGROUND: Magnetic resonance elastography (MRE) can quantify tissue biomechanics noninvasively, including pathological hepatic states like metabolic dysfunction-associated steatohepatitis. PURPOSE: To compare the performance of 2D/3D-MRE using the gravitational (GT) transducer concept with the current commercial acoustic (AC) solution utilizing a 2D-MRE approach. Additionally, quality index markers (QIs) were proposed to identify image pixels with sufficient quality for reliably estimating tissue biomechanics. STUDY TYPE: Prospective. POPULATION: One hundred seventy participants with suspected or confirmed liver disease (median age, 57 years [interquartile range (IQR), 46-65]; 66 females), and 11 healthy volunteers (median age, 31 years [IQR, 27-34]; 5 females). FIELD STRENGTH/SEQUENCE: Participants were scanned twice at 1.5 T and 60 Hz vibration frequency: first, using AC-MRE (2D-MRE, spin-echo EPI sequence, 11 seconds breath-hold), and second, using GT-MRE (2D- and 3D-MRE, gradient-echo sequence, 14 seconds breath-hold). ASSESSMENT: Image analysis was performed by four independent radiologists and one biomedical engineer. Additionally, superimposed analytic plane shear waves of known wavelength and attenuation at fixed shear modulus were used to propose pertinent QIs. STATISTICAL TESTS: Spearman's correlation coefficient (r) was applied to assess the correlation between modalities. Interreader reproducibility was evaluated using Bland-Altman bias and reproducibility coefficients. P-values <0.05 were considered statistically significant. RESULTS: Liver stiffness quantified via GT-2D/3D correlated well with AC-2D (r ≥ 0.89 [95% CI: 0.85-0.92]) and histopathological grading (r ≥ 0.84 [95% CI: 0.72-0.91]), demonstrating excellent agreement in Bland-Altman plots and between readers (κ ≥ 0.86 [95% CI: 0.81-0.91]). However, GT-2D showed a bias in overestimating stiffness compared to GT-3D. Proposed QIs enabled the identification of pixels deviating beyond 10% from true stiffness based on a combination of total wave amplitude, temporal sinusoidal nonlinearity, and wave signal-to-noise ratio for GT-3D. CONCLUSION: GT-MRE represents an alternative to AC-MRE for noninvasive liver tissue characterization. Both GT-2D and 3D approaches correlated strongly with the established commercial approach, offering advanced capabilities in abdominal imaging compared to AC-MRE. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 2.

Diagnostic Accuracy of Dual-Energy CT-Derived Metrics for the Prediction of Osteoporosis-Associated Fractures.

RATIONALE AND OBJECTIVES: This study aimed to compare the diagnostic value of dual-energy CT (DECT)-based volumetric material decomposition with that of Hounsfield units (HU)-based values and cortical thickness ratio for predicting the 2-year risk of osteoporosis-associated fractures. METHODS: The L1 vertebrae of 111 patients (55 men, 56 women; median age, 62 years) who underwent DECT between 01/2015 and 12/2018 were retrospectively analyzed. For phantomless bone mineral density (BMD) assessment, a specialized DECT postprocessing software employing material decomposition was utilized. The digital records of all patients were monitored for two years after the DECT scans to track the incidence of osteoporotic fractures. Diagnostic accuracy parameters were calculated for all metrics using receiver-operating characteristic (ROC) and precision-recall (PR) curves. Logistic regression models were used to determine associations of various predictive metrics with the occurrence of osteoporotic fractures. RESULTS: Patients who sustained one or more osteoporosis-associated fractures in a 2-year interval were significantly older (median age 74.5 years [IQR 57-83 years]) compared those without such fractures (median age 50.5 years [IQR 38.5-69.5 years]). According to logistic regression models, DECT-derived BMD was the sole predictive parameter significantly associated with osteoporotic fracture occurrence across all age groups. ROC and PR curve analyses confirmed the highest diagnostic accuracy for DECT-based BMD, with an area under the curve (AUC) of 0.95 [95% CI: 0.89-0.98] for the ROC curve and an AUC of 0.96 [95% CI: 0.85-0.99] for the PR curve. CONCLUSION: The diagnostic performance of DECT-based BMD in predicting the 2-year risk of osteoporotic fractures is greater than that of HU-based metrics and the cortical thickness ratio. DECT-based BMD values are highly valuable in identifying patients at risk for osteoporotic fractures.

Optimization of window settings for coronary arteries assessment using spectral CT-derived virtual monoenergetic imaging.

PURPOSE: To determine the optimal window setting for virtual monoenergetic images (VMI) reconstructed from dual-layer spectral coronary computed tomography angiography (DE-CCTA) datasets. MATERIAL AND METHODS: 50 patients (30 males; mean age 61.1 ± 12.4 years who underwent DE-CCTA from May 2021 to June 2022 for suspected coronary artery disease, were retrospectively included. Image quality assessment was performed on conventional images and VMI reconstructions at 70 and 40 keV. Objective image quality was assessed using contrast-to-noise ratio (CNR). Two independent observers manually identified the best window settings (B-W/L) for VMI 70 and VMI 40 visualization. B-W/L were then normalized with aortic attenuation using linear regression analysis to obtain the optimized W/L (O-W/L) settings. Additionally, subjective image quality was evaluated using a 5-point Likert scale, and vessel diameters were measured to examine any potential impact of different W/L settings. RESULTS: VMI 40 demonstrated higher CNR values compared to conventional and VMI 70. B-W/L settings identified were 1180/280 HU for VMI 70 and 3290/900 HU for VMI 40. Subsequent linear regression analysis yielded O-W/L settings of 1155/270 HU for VMI 70 and 3230/880 HU for VMI 40. VMI 40 O-W/L received the highest scores for each parameter compared to conventional (all p < 0.0027). Using O-W/L settings for VMI 70 and VMI 40 did not result in significant differences in vessel measurements compared to conventional images. CONCLUSION: Optimization of VMI requires adjustments in W/L settings. Our results recommend W/L settings of 1155/270 HU for VMI 70 and 3230/880 HU for VMI 40.

Prostatic Artery Embolization: Lessons From 551 Procedures at a Single-Center.

RATIONALE AND OBJECTIVES: This retrospective study evaluates the efficacy and safety of Prostatic Artery Embolization (PAE) for the treatment of benign prostatic hyperplasia (BPH) over five years at a single center, conducted by an experienced interventional radiologist. MATERIALS AND METHODS: We analyzed 551 PAE interventions from January 2019 to July 2023. Key metrics included patient demographics, procedural details (radiation exposure, particle size), complication rates, pre- and post-interventional prostatic volume (PV), Prostate-specific Antigen (PSA) levels, International Prostate Symptom Score (IPSS), Quality of Life (QoL) scores and International Index of Erectile Function (IIEF) scores. We assessed data normality, performed group and paired sample comparisons, and evaluated correlations. RESULTS: For 551 men, the average patient age was 68.81 ± 8.61 years undergoing bilateral embolization. The particle size predominantly used was 100-300 µm (n = 441). PAE lead to significant (p < .001) reduction of both PV (-9.67 ± 14.52 mL) and PSA level (-2,65 ± 1.56 ng/mL) between pre- and three months after PAE. Substantial improvement were observed for IPSS (-9 points) and QoL scores (-2 points), with stable IIEF scores. Only minor complications (n = 16) were reported, and no major complications were observed. Between the first PAE in 2019 and the routinely performed PAE in 2023 significant (p < .0001) reductions in fluoroscopy (-25.2%), and procedural times (-26.1%) were observed. CONCLUSION: In conclusion, PAE is a safe and effective treatment for BPH, offering significant improvements in lower urinary tract symptoms (LUTS) and QoL while maintaining sexual function.

Optimizing resource allocation: Cost-effectiveness of specified D-dimer cut-offs in cancer patients with suspected venous thromboembolism.

An accurate diagnosis of venous thromboembolism (VTE) is crucial, given the potential for high mortality in undetected cases. Strategic D-dimer testing may aid in identifying low-risk patients, preventing overdiagnosis and reducing imaging costs. We conducted a retrospective, comparative analysis to assess the potential cost savings that could be achieved by adopting different approaches to determine the most effective D-dimer cut-off value in cancer patients with suspected VTE, compared to the commonly used rule-out cut-off level of 0.5 mg/L. The study included 526 patients (median age 65, IQR 55-75) with a confirmed cancer diagnosis who underwent D-dimer testing. Among these patients, the VTE prevalence was 29% (n = 152). Each diagnostic strategy's sensitivity, specificity, negative likelihood ratio (NLR), as well as positive likelihood ratio (PLR), and the proportion of patients exhibiting a negative D-dimer test result, were calculated. The diagnostic strategy that demonstrated the best balance between specificity, sensitivity, NLR, and PLR, utilized an inverse age-specific cut-off level for D-dimer [0.5 + (66-age) × 0.01 mg/L]. This method yielded a PLR of 2.9 at a very low NLR for the exclusion of VTE. We observed a significant cost reduction of 4.6% and 1.0% for PE and DVT, respectively. The utilization of an age-adjusted cut-off [patient's age × 0.01 mg/L] resulted in the highest cost savings, reaching 8.1% for PE and 3.4% for DVT. Using specified D-dimer cut-offs in the diagnosis of VTE could improve economics, considering the limited occurrence of confirmed cases among patients with suspected VTE.

Predictive Value of Dual-Energy CT-Derived Metrics for the Use of Bone Substitutes in Distal Radius Fracture Surgery.

(1) Background: Low bone mineral density (BMD) is a significant risk factor for complicated surgery and leads to the increased use of bone substitutes in patients with distal radius fractures (DRFs). No accepted model has yet been established to predict the use of bone substitutes to facilitate preoperative planning. (2) Methods: Unenhanced dual-energy CT (DECT) images of DRFs were retrospectively acquired between March 2016 and September 2020 using the internal PACS system. Available follow-up imaging and medical health records were reviewed to determine the use of bone substitutes. DECT-based BMD, trabecular Hounsfield units (HU), cortical HU, and cortical thickness ratio were measured in non-fractured segments of the distal radius. Diagnostic accuracy parameters were calculated for all metrics using receiver-operating characteristic (ROC) curves and associations of all metrics with the use of bone substitutes were evaluated using logistic regression models. (3) The final study population comprised 262 patients (median age 55 years [IQR 43-67 years]; 159 females, 103 males). According to logistic regression analysis, DECT-based BMD was the only metric significantly associated with the use of bone substitutes (odds ratio 0.96, p = 0.003). However, no significant associations were found for cortical HU (p = 0.06), trabecular HU (p = 0.33), or cortical thickness ratio (p = 0.21). ROC-curve analysis revealed that a combined model of all four metrics had the highest diagnostic accuracy with an area under the curve (AUC) of 0.76. (4) Conclusions: DECT-based BMD measurements performed better than HU-based measurements and cortical thickness ratio. The diagnostic performance of all four metrics combined was superior to that of the individual parameters.

Advancing Differentiation of Hepatic Metastases in Malignant Melanoma through Dual-Energy Computed Tomography Rho/Z Maps.

OBJECTIVES: The aim of this study is to evaluate the diagnostic accuracy of dual-energy computed tomography (DECT)-based Rho/Z maps in differentiating between metastases and benign liver lesions in patients diagnosed with malignant melanoma compared to conventional CT value measurements. METHODS: This retrospective study included 73 patients (mean age, 70 ± 13 years; 43 m/30 w) suffering from malignant melanoma who had undergone third-generation DECT as part of tumor staging between December 2017 and December 2021. For this study, we measured Rho (electron density) and Z (effective atomic number) values as well as Hounsfield units (HUs) in hypodense liver lesions. Values were compared, and diagnostic accuracy for differentiation was computed using receiver operating characteristic (ROC) curve analyses. Additional performed MRI or biopsies served as a standard of reference. RESULTS: A total of 136 lesions (51 metastases, 71 cysts, and 14 hemangiomas) in contrast-enhanced DECT images were evaluated. The most notable discrepancy (p < 0.001) between measured values and the highest diagnostic accuracy for distinguishing melanoma metastases from benign cysts was observed for the Z (0.992; 95% CI, 0.956-1) parameters, followed by Rho (0.908; 95% CI, 0.842-0.953) and finally HU120kV (0.829; 95% CI, 0.751-0.891). Conversely, when discriminating between liver metastases and hemangiomas, the HU120kV parameters showed the most significant difference (p < 0.001) and yielded the highest values for diagnostic accuracy (0.859; 95% CI, 0.740-0.937), followed by the Z parameters (0.790; 95% CI, 0.681-0.876) and finally the Rho values (0.621; 95% CI, 0.501-0.730). CONCLUSIONS: Rho and Z measurements derived from DECT allow for improved differentiation of liver metastases and benign liver cysts in patients with malignant melanoma compared to conventional CT value measurements. In contrast, in differentiation between liver hemangiomas and metastases, Rho/Z maps show inferior diagnostic accuracy. Therefore, differentiation between these two lesions remains a challenge for CT imaging.

Dual-Energy CT Material Decomposition: The Value in the Detection of Lymph Node Metastasis from Breast Cancer.

PURPOSE: To evaluate the diagnostic performance of a dual-energy computed tomography (DECT)-based material decomposition algorithm for iodine quantification and fat fraction analysis to detect lymph node metastases in breast cancer patients. MATERIALS AND METHODS: 30 female patients (mean age, 63.12 ± 14.2 years) diagnosed with breast cancer who underwent pre-operative chest DECT were included. To establish a reference standard, the study correlated histologic repots after lymphadenectomy or confirming metastasis in previous/follow-up examinations. Iodine concentration and fat fraction were determined through region-of-interest measurements on venous DECT iodine maps. Receiver operating characteristic curve analysis was conducted to identify the optimal threshold for differentiating between metastatic and non-metastatic lymph nodes. RESULTS: A total of 168 lymph nodes were evaluated, divided into axillary (metastatic: 46, normal: 101) and intramammary (metastatic: 10, normal: 11). DECT-based fat fraction values exhibited significant differences between metastatic (9.56 ± 6.20%) and non-metastatic lymph nodes (41.52 ± 19.97%) (p < 0.0001). Absolute iodine concentrations showed no significant differences (2.25 ± 0.97 mg/mL vs. 2.08 ± 0.97 mg/mL) (p = 0.7999). The optimal fat fraction threshold for diagnosing metastatic lymph nodes was determined to be 17.75%, offering a sensitivity of 98% and a specificity of 94%. CONCLUSIONS: DECT fat fraction analysis emerges as a promising method for identifying metastatic lymph nodes, overcoming the morpho-volumetric limitations of conventional CT regarding lymph node assessment. This innovative approach holds potential for improving pre-operative lymph node evaluation in breast cancer patients, offering enhanced diagnostic accuracy.

Potential of Dual-Energy CT-Based Collagen Maps for the Assessment of Disk Degeneration in the Lumbar Spine.

RATIONALE AND OBJECTIVES: Lumbar disk degeneration is a common condition contributing significantly to back pain. The objective of the study was to evaluate the potential of dual-energy CT (DECT)-derived collagen maps for the assessment of lumbar disk degeneration. PATIENTS AND METHODS: We conducted a retrospective analysis of 127 patients who underwent dual-source DECT and MRI of the lumbar spine between 07/2019 and 10/2022. The level of lumbar disk degeneration was categorized by three radiologists as follows: no/mild (Pfirrmann 1&2), moderate (Pfirrmann 3&4), and severe (Pfirrmann 5). Recall (sensitivity) and accuracy of DECT collagen maps were calculated. Intraclass correlation coefficient (ICC) was used to evaluate inter-reader reliability. Subjective evaluations were performed using 5-point Likert scales for diagnostic confidence and image quality. RESULTS: We evaluated a total of 762 intervertebral disks from 127 patients (median age, 69.7 (range, 23.0-93.7), female, 56). MRI identified 230 non/mildly degenerated disks (30.2%), 484 moderately degenerated disks (63.5%), and 48 severely degenerated disks (6.3%). DECT collagen maps yielded an overall accuracy of 85.5% (1955/2286). Recall (sensitivity) was 79.3% (547/690) for the detection of no/mild lumbar disk degeneration, 88.7% (1288/1452) for the detection of moderate disk degeneration, and 83.3% (120/144) for the detection of severe disk degeneration (ICC=0.9). Subjective evaluations of DECT collagen maps showed high diagnostic confidence (median 4) and good image quality (median 4). CONCLUSION: The use of DECT collagen maps to distinguish different stages of lumbar disk degeneration may have clinical significance in the early diagnosis of disk-related pathologies in patients with contraindications for MRI or in cases of unavailability of MRI.

Value of Dual-Energy CT-Derived Metrics for the Prediction of Bone Non-union in Distal Radius Fractures.

RATIONALE AND OBJECTIVES: Bone non-union is a serious complication of distal radius fractures (DRF) that can result in functional limitations and persistent pain. However, no accepted method has been established to identify patients at risk of developing bone non-union yet. This study aimed to compare various CT-derived metrics for bone mineral density (BMD) assessment to identify predictive values for the development of bone non-union. MATERIALS AND METHODS: CT images of 192 patients with DRFs who underwent unenhanced dual-energy CT (DECT) of the distal radius between 03/2016 and 12/2020 were retrospectively identified. Available follow-up imaging and medical health records were evaluated to determine the occurrence of bone non-union. DECT-based BMD, trabecular Hounsfield unit (HU), cortical HU and cortical thickness ratio were measured in normalized non-fractured segments of the distal radius. RESULTS: Patients who developed bone non-union were significantly older (median age 72 years vs. 54 years) and had a significantly lower DECT-based BMD (median 68.1 mg/cm3 vs. 94.6 mg/cm3, p < 0.001). Other metrics (cortical thickness ratio, cortical HU, trabecular HU) showed no significant differences. ROC and PR curve analyses confirmed the highest diagnostic accuracy for DECT-based BMD with an area under the curve (AUC) of 0.83 for the ROC curve and an AUC of 0.46 for the PR curve. In logistic regression models, DECT-based BMD was the sole metric significantly associated with bone non-union. CONCLUSION: DECT-derived metrics can accurately predict bone non-union in patients who sustained DRF. The diagnostic performance of DECT-based BMD is superior to that of HU-based metrics and cortical thickness ratio.

From pixels to prognosis: Imaging biomarkers for discrimination and outcome prediction of pulmonary embolism : Original Research Article.

PURPOSE: Recent advancements in medical imaging have transformed diagnostic assessments, offering exciting possibilities for extracting biomarker-based information. This study aims to investigate the capabilities of a machine learning classifier that incorporates dual-energy computed tomography (DECT) radiomics. The primary focus is on discerning and predicting outcomes related to pulmonary embolism (PE). METHODS: The study included 131 participants who underwent pulmonary artery DECT angiography between January 2015 and March 2022. Among them, 104 patients received the final diagnosis of PE and 27 patients served as a control group. A total of 107 radiomic features were extracted for every case based on DECT imaging. The dataset was divided into training and test sets for model development and validation. Stepwise feature reduction identified the most relevant features, which were used to train a gradient-boosted tree model. Receiver operating characteristics analysis and Cox regression tests assessed the association of texture features with overall survival. RESULTS: The trained machine learning classifier achieved a classification accuracy of 0.94 for identifying patients with acute PE with an area under the receiver operating characteristic curve of 0.91. Radiomics features could be valuable for predicting outcomes in patients with PE, demonstrating strong prognostic capabilities in survival prediction (c-index, 0.991 [0.979-1.00], p = 0.0001) with a median follow-up of 130 days (IQR, 38-720). Notably, the inclusion of clinical or DECT parameters did not enhance predictive performance. CONCLUSION: In conclusion, our study underscores the promising potential of leveraging radiomics on DECT imaging for the identification of patients with acute PE and predicting their outcomes. This approach has the potential to improve clinical decision-making and patient management, offering efficiencies in time and resources by utilizing existing DECT imaging without the need for an additional scoring system.

Incident fractures of the distal radius: Dual-energy CT-derived metrics for opportunistic risk stratification.

BACKGROUND: Dual-energy CT (DECT)-derived bone mineral density (BMD) of the distal radius and other CT-derived metrics related to bone health have been suggested for opportunistic osteoporosis screening and risk evaluation for sustaining distal radius fractures (DRFs). METHODS: The distal radius of patients who underwent DECT between 01/2016 and 08/2021 was retrospectively analyzed. Cortical Hounsfield Unit (HU), trabecular HU, cortical thickness, and DECT-based BMD were acquired from a non-fractured, metaphyseal area in all examinations. Receiver-operating characteristic (ROC) analysis was conducted to determine the area under the curve (AUC) values for predicting DRFs based on DECT-derived BMD, HU values, and cortical thickness. Logistic regression models were then employed to assess the associations of these parameters with the occurrence of DRFs. RESULTS: In this study, 263 patients (median age: 52 years; interquartile range: 36-64; 132 women; 192 fractures) were included. ROC curve analysis revealed a higher area under the curve (AUC) value for DECT-derived BMD compared to cortical HU, trabecular HU, and cortical thickness (0.91 vs. 0.61, 0.64, and 0.69, respectively; p <.001). Logistic regression models confirmed the association between lower DECT-derived BMD and the occurrence of DRFs (Odds Ratio, 0.83; p <.001); however, no influence was observed for cortical HU, trabecular HU, or cortical thickness. CONCLUSIONS: DECT can be used to assess the BMD of the distal radius without dedicated equipment such as calibration phantoms to increase the detection rates of osteoporosis and stratify the individual risk to sustain DRFs. In contrast, assessing HU-based values and cortical thickness does not provide clinical benefit.

Indicators for Hospitalization in Acute Pulmonary Embolism: Uncover the Association Between D-dimer Levels, Thrombus Volume and Radiomics.

BACKGROUND: The advent of advanced computed tomography (CT) technology and the field of radiomics has opened up new avenues in diagnostic assessments. Increasingly, there is substantial evidence advocating for the incorporation of quantitative imaging biomarkers in the clinical decision-making process. This study aimed to examine the correlation between D-dimer levels and thrombus size in acute pulmonary embolism (PE) combining dual-energy CT (DECT) and radiomics and to investigate the diagnostic utility of a machine learning classifier based on dual-energy computed tomography (DECT) radiomics for identifying patients with a complicated course, defined as at least hospitalization at IMC. METHODS: The study was conducted including 136 participants who underwent pulmonary artery CT angiography from January 2015 to March 2022. Based on DECT imaging, 107 radiomic features were extracted for each patient using standardized image processing. After dividing the dataset into training and test sets, stepwise feature reduction based on reproducibility, variable importance and correlation analyses were performed to select the most relevant features; these were used to train and validate the gradient-boosted tree models.Receiver operating characteristics (ROC) analysis was utilized to evaluate the association between volumetric, laboratory data and adverse outcomes. RESULTS: In the central PE group, we observed a significant correlation between thrombus volumetrics and D-dimer levels (p = 0.0037), as well as between thrombus volumetrics and hospitalization at the Intermediate Care Unit (IMC) (p = 0.0001). In contrast, no statistically significant differences were identified in thrombus sizes between patients who experienced complications and those who had a favorable course (p = 0.3162). The trained machine learning classifier achieved an accuracy of 61% and 55% in identifying patients with a complicated course, as indicated by an area under the ROC curve of 0.63 and 0.58. CONCLUSION: In conclusion, our findings indicate a positive correlation between D-dimer levels and central PE's pulmonary embolic burden. Thrombus volumetrics may serve as an indicator for complications and outcomes in acute PE patients. Thus, thrombus volumetrics, as opposed to D-dimers, could be an additional marker for evaluating embolic disease severity. Moreover, DECT-derived radiomic feature models show promise in identifying patients with a complicated course, such as hospitalization at IMC.

Imaging-based risk stratification of patients with pulmonary embolism based on dual-energy CT-derived radiomics.

BACKGROUND: Technological progress in the acquisition of medical images and the extraction of underlying quantitative imaging data has introduced exciting prospects for the diagnostic assessment of a wide range of conditions. This study aims to investigate the diagnostic utility of a machine learning classifier based on dual-energy computed tomography (DECT) radiomics for classifying pulmonary embolism (PE) severity and assessing the risk for early death. METHODS: Patients who underwent CT pulmonary angiogram (CTPA) between January 2015 and March 2022 were considered for inclusion in this study. Based on DECT imaging, 107 radiomic features were extracted for each patient using standardized image processing. After dividing the dataset into training and test sets, stepwise feature reduction based on reproducibility, variable importance and correlation analyses were performed to select the most relevant features; these were used to train and validate the gradient-boosted tree models. RESULTS: The trained machine learning classifier achieved a classification accuracy of .90 for identifying high-risk PE patients with an area under the receiver operating characteristic curve of .59. This CT-based radiomics signature showed good diagnostic accuracy for risk stratification in individuals presenting with central PE, particularly within higher risk groups. CONCLUSION: Models utilizing DECT-derived radiomics features can accurately stratify patients with pulmonary embolism into established clinical risk scores. This approach holds the potential to enhance patient management and optimize patient flow by assisting in the clinical decision-making process. It also offers the advantage of saving time and resources by leveraging existing imaging to eliminate the necessity for manual clinical scoring.