Difference In Diagnostic Performance Research Articles

Caries Detection and Classification in Photographs Using an Artificial Intelligence-Based Model-An External Validation Study.

This ex vivo diagnostic study aimed to externally validate a freely accessible AI-based model for caries detection, classification, localisation and segmentation using an independent image dataset. It was hypothesised that there would be no difference in diagnostic performance compared to previously published internal validation data. For the independent dataset, 718 dental images representing different stages of carious (n = 535) and noncarious teeth (n = 183) were retrieved from the internet. All photographs were evaluated by the dental team (reference standard) and the AI-based model (test method). Diagnostic performance was statistically determined using cross-tabulations to calculate accuracy (ACC), sensitivity (SE), specificity (SP) and area under the curve (AUC). An overall ACC of 92.0% was achieved for caries detection, with an ACC of 85.5-95.6%, SE of 42.9-93.3%, SP of 82.1-99.4% and AUC of 0.702-0.909 for the classification of caries. Furthermore, 97.0% of the cases were accurately localised. Fully and partially correct segmentation was achieved in 52.9% and 44.1% of the cases, respectively. The validated AI-based model showed promising diagnostic performance in detecting and classifying caries using an independent image dataset. Future studies are needed to investigate the validity, reliability and practicability of AI-based models using dental photographs from different image sources and/or patient groups.

Head-to-head comparison of contrast-enhanced CT, dual-layer spectral-detector CT, and Gd-EOB-DTPA-enhanced MR in detecting neuroendocrine tumor liver metastases

PurposeTo explore the optimal of kiloelectron voltage (keV) of virtual monoenergetic imaging (VMI) of dual-layer spectral-detector CT (DLCT) in detecting neuroendocrine tumor liver metastases (NETLM) and to investigate diagnostic performance of polyenergetic images (PEI), DLCT, and Gd-EOB-DTPA-enhanced MR. MethodsSeventy-two patients with suspected NETLM who underwent DLCT and Gd-EOB-DTPA-enhanced MR were retrospectively enrolled. Tumor signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were compared between PEI and VMI at 40–140 keV. Two radiologists read the CT examinations with and without VMI separately in consensus. Two other radiologists read the Gd-EOB-DTPA-enhanced MR in consensus. The diagnostic performance was evaluated. Reference standard was histopathology, follow-up, and interpretation of all available imaging. ResultsThe highest SNR and CNR were observed at VMI40keV, significantly higher than PEI in the arterial and venous phases (all P<0.01). A total of 477 lesions were identified (396 metastases, 81 benign lesions). Per-lesion AUC was 0.86, 0.91, and 0.97 (PEI, DLCT, and Gd-EOB-DTPA-enhanced MR, respectively). Sensitivity of PEI, DLCT, and Gd-EOB-DTPA-enhanced MRI were 0.76, 0.86, and 0.95, respectively. DLCT significantly improved sensitivity compared to PEI. MR had significantly higher sensitivity than DLCT and PEI. Subgroup analysis demonstrated that the difference in diagnostic performance was concentrated on lesions < 10 mm. ConclusionThe image quality of VMI40keV is higher than that of PEI. DLCT with VMI40keV provides better diagnostic sensitivity for NETLM detection than PEI. Gd-EOB-DTPA-enhanced MR yielded the best diagnostic performance for NETLM detection.

Diagnostic Performance of 18F-FDG PET/CT According to Delay After Treatment to Detect Subclinical Recurrence of Head and Neck Squamous Cell Carcinoma.

Head and neck squamous cell carcinoma (HNSCC) remains a malignancy with high rates of locoregional recurrence and poor prognosis for recurrent cases. Early detection of subclinical lesions is challenging but critical for effective patient management. Imaging surveillance after treatment, particularly 18F-FDG PET/CT, has shown promise in the diagnosis of HNSCC recurrence. The aim was to evaluate the diagnostic performance of 18F-FDG PET/CT according to delay after treatment in detecting subclinical recurrence (SCR) in HNSCC patients. Methods: In this retrospective study, all 18F-FDG PET/CT scans were performed at a single center. All adults with histologically proven HNSCC who were treated with curative intent between January 1, 2006, and December 31, 2021, were included. They had a normal clinical examination before each scan. Patients who underwent an intensive follow-up strategy after treatment had 18F-FDG PET/CT with an intravenous contrast agent at 3-6 mo and annually thereafter for 5 y. The primary endpoint was diagnostic performance (positive and negative predictive values, sensitivity, specificity, and accuracy). Results: In total, 2,566 18F-FDG PET/CT scans were performed among 852 patients, with an average of 3 scans per patient. The overall diagnostic performance measures were as follows: positive predictive value (88%), negative predictive value (98%), sensitivity (98%), specificity (89%), and accuracy (93%). There were no significant differences in diagnostic performance over time. The scans detected 126 cases of SCR (14.8%) and 118 cases of metachronous cancer (13.8%). The incidence of SCR decreased over time, with the highest detection rate in the first 2 y after treatment. Positive predictive value improved over time, reaching 90% for the digital Vision 600 system (third period) compared with 76% for the analog Gemini GXLi system (first period, P < 0.001). Multivariate analysis identified advanced stage, high body mass index, and initial PET/CT upstaging as predictive factors for detection of SCR. Conclusion: Our study demonstrates that 18F-FDG PET/CT has high diagnostic performance in detecting SCR during follow-up after treatment of HNSCC, especially in the first 2 y. Advanced tumor stage, initial PET/CT upstaging, and high body mass index were associated with a higher likelihood of SCR detection. The routine use of 18F-FDG PET/CT during follow-up seems justified for patients with HNSCC.

Knowledge-Augmented Deep Learning for Segmenting and Detecting Cerebral Aneurysms With CT Angiography: A Multicenter Study.

Background Deep learning (DL) could improve the labor-intensive, challenging processes of diagnosing cerebral aneurysms but requires large multicenter data sets. Purpose To construct a DL model using a multicenter data set for accurate cerebral aneurysm segmentation and detection on CT angiography (CTA) images and to compare its performance with radiology reports. Materials and Methods Consecutive head or head and neck CTA images of suspected unruptured cerebral aneurysms were gathered retrospectively from eight hospitals between February 2018 and October 2021 for model development. An external test set with reference standard digital subtraction angiography (DSA) scans was obtained retrospectively from one of the eight hospitals between February 2022 and February 2023. Radiologists (reference standard) assessed aneurysm segmentation, while model performance was evaluated using the Dice similarity coefficient (DSC). The model's aneurysm detection performance was assessed by sensitivity and comparing areas under the receiver operating characteristic curves (AUCs) between the model and radiology reports in the DSA data set with use of the DeLong test. Results Images from 6060 patients (mean age, 56 years ± 12 [SD]; 3375 [55.7%] female) were included for model development (training: 4342; validation: 1086; and internal test set: 632). Another 118 patients (mean age, 59 years ± 14; 79 [66.9%] female) were included in an external test set to evaluate performance based on DSA. The model achieved a DSC of 0.87 for aneurysm segmentation performance in the internal test set. Using DSA, the model achieved 85.7% (108 of 126 aneurysms [95% CI: 78.1, 90.1]) sensitivity in detecting aneurysms on per-vessel analysis, with no evidence of a difference versus radiology reports (AUC, 0.93 [95% CI: 0.90, 0.95] vs 0.91 [95% CI: 0.87, 0.94]; P = .67). Model processing time from reconstruction to detection was 1.76 minutes ± 0.32 per scan. Conclusion The proposed DL model could accurately segment and detect cerebral aneurysms at CTA with no evidence of a significant difference in diagnostic performance compared with radiology reports. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Payabvash in this issue.

Time-dependent diffusion MRI-based microstructural mapping for differentiating high-grade serous ovarian cancer from serous borderline ovarian tumor

PurposeTo investigate the value of microstructural characteristics derived from time-dependent diffusion MRI in distinguishing high-grade serous ovarian cancer (HGSOC) from serous borderline ovarian tumor (SBOT) and the associations of immunohistochemical markers with microstructural features. MethodsTotally 34 HGSOC and 12 SBOT cases who received preoperative pelvic MRI were retrospectively included in this study. Two radiologists delineated the tumors to obtain the regions of interest (ROIs). Time-dependent diffusion MRI signals were fitted by the IMPULSED (imaging microstructural parameters using limited spectrally edited diffusion) model, to extract microstructural parameters, including fraction of the intracellular component (fin), cell diameter (d), cellularity and extracellular diffusivity (Dex). Apparent diffusion coefficient (ADC) values were obtained from standard diffusion-weighted imaging (DWI). The parameters of HGSOCs and SBOTs were compared, and the diagnostic performance was evaluated. The associations of microstructural indexes with immunopathological parameters were assessed, including Ki-67, P53, Pax-8, ER and PR. ResultsIn this study, fin, cellularity, Dex and ADC had good diagnostic performance levels in differentiating HGSOC from SBOT, with AUCs of 0.936, 0.909, 0.902 and 0.914, respectively. There were no significant differences in diagnostic performance among these parameters. Spearman analysis revealed in the HGSOC group, cellularity had a significant positive correlation with P53 expression (P = 0.028, r = 0.389) and Dex had a significant positive correlation with Pax-8 expression (P = 0.018, r = 0.415). ICC showed excellent agreement for all parameters. ConclusionTime-dependent diffusion MRI had value in evaluating the microstructures of HGSOC and SBOT and could discriminate between these tumors.

Paper 41: Does MR Arthrography Improve Evaluation of Patients with Posterior Shoulder Instability? An Analysis from the MOON Shoulder Instability Group

Objectives: Posterior shoulder instability is an uncommon injury, accounting for 2-12% of shoulder instability patients. Magnetic resonance imaging (MRI) is often utilized to non-invasively evaluate the joint for labral, cartilage, and bony injury. MR arthrography (MRA) may improve the evaluation of these intra-articular structures, but requires additional cost, time, and an invasive procedure. Moreover, its benefits for patients with posterior shoulder instability are unclear. Accurate imaging is critical in patients undergoing surgery because certain findings may influence the surgical approach. Thus, the purpose of this study is to compare the accuracy of MRA and non-enhanced MRI in detecting labral, chondral, and osseous lesions in patients with posterior shoulder instability. Methods: There were 291 patients with unilateral posterior shoulder instability who had preoperative MRA or MRI between 2006 and 2021 in the MOON database. Images were reviewed by a fellowship-trained musculoskeletal radiologist with 10+ years of experience and an orthopaedic surgeon. Evaluators were blinded to age, gender, medical history, MR report, and arthroscopic findings, but were aware of the patient's history of posterior shoulder instability. Examinations were evaluated for presence, location, morphology, and length of glenoid labral tears. Tear length and localization were noted using a standard clock face. The presence and size of humeral/glenoid osseous and cartilage lesions was also noted.Intraoperative findings from arthroscopic evaluation were documented by the surgeon using a standard response questionnaire. Diagnostic performance of standard MRA and MRI were evaluated using sensitivity and specificity with arthroscopic evaluation as the gold-standard. Chi-squared or Fisher’s exact tests were used to compare the sensitivities and specificities of MRA and MRI. Significance was defined as P &lt; 0.05. Results: 200 patients (69%) had MRA, while 91 patients had non-contrast MRIs of the shoulder. On imaging evaluation, 273 (94%) patients had labral tears with a mean tear size of 156º. 39 (13%) patients had humeral cartilage lesions, whereas 40 (14%) patients had glenoid cartilage lesions. 34 (12%) patients had glenoid bone loss greater than 5%. The mean glenoid bone loss was 12% (range, 5% - 26%). A reverse Hill-Sachs lesion was present in 35 (12%) patients. 16 (6%) patients had bipolar bone loss. The sensitivity for detection of labral tears was high for both MRA (0.95) and standard MRI (0.98) (Tables 1 and 2), though the specificity was low for both modalities (0.25 and 0.40, respectively). There was no difference in the sensitivities or specificities of the two imaging modalities in the detection of labral tears. MRA had a lower sensitivity than MRI in the detection of glenoid cartilage lesions (0.31 vs 0.88, p &lt; 0.001), but no difference was observed for specificity. There was no difference in the sensitivities and specificities of MRA and MRI for humeral cartilage lesions (p = 0.354).MRA had a higher sensitivity than MRI in assessing glenoid bone loss (0.78 vs 0.50, p &lt; 0.001), though there was no difference in specificity. There was no difference between imaging modalities in the evaluation of humeral bone loss (p = 1.000). Conclusions: In this multicenter study, MRA was the most commonly utilized preoperative advanced imaging study for patients with posterior shoulder instability. MRA had a higher sensitivity in detecting glenoid bone loss but lower sensitivity in detecting glenoid cartilage lesions compared to standard non-contrast MRI. There was no difference in diagnostic performance between MRA and MRI in the diagnosis of labral tears and humeral chondral and osseous lesions. MRA may provide an advantage over standard MRI in the evaluation of glenoid bone loss, which can influence the treatment plan for patients undergoing surgical management of posterior shoulder instability.

International Multi-Specialty Expert Physician Preoperative Identification of Extranodal Extension n Oropharyngeal Cancer Patients using Computed Tomography: Prospective Blinded Human Inter-Observer Performance Evaluation.

Extranodal extension (pENE) is a critical prognostic factor in oropharyngeal cancer (OPC) that drives therapeutic disposition. Determination of pENE from radiological imaging has been associated with high inter-observer variability. However, the impact of clinician specialty on human observer performance of imaging-detected extranodal extension (iENE) remains poorly understood. To characterize the impact of clinician specialty on the accuracy of pre-operative iENE in human papillomavirus-positive (HPV+) OPC using computed tomography (CT) images. This prospective observational human performance study analyzed pre-therapy CT images from 24 HPV+ OPC patients, with duplication of 6 scans (n=30) of which 21 were pathologically confirmed pENE. Thirty-four expert observers, including 11 radiologists, 12 surgeons, and 11 radiation oncologists, independently assessed these scans for iENE and reported human-detected radiologic criteria and observer confidence. The primary outcomes included accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and Brier score for each physician, compared to ground-truth pENE. The significance of radiographic signs for prediction of pENE were determined through logistic regression analysis. Fleiss' kappa measured interobserver agreement, and Hanley-MacNeil AUC discrimination testing. Median accuracy across all specialties was 0.57 (95%CI 0.39 to 0.73), with no specialty showing discriminate performance greater than random estimation (median AUC 0.64, 95%CI 0.44 to 0.83). Significant differences between radiologists and surgeons in Brier scores (0.33 vs. 0.26, p < 0.01), radiation oncologists and surgeons in sensitivity (0.48 vs. 0.69, p > 0.1), and radiation oncologists and radiologists/surgeons in specificity (0.89 vs. 0.56, p > 0.1). Indistinct capsular contour and nodal necrosis were significant predictors of correct pENE status among all specialties. Interobserver agreement was weak for all the radiographic criteria, regardless of specialty (κ<0.6). Multiobserver testing shows physician discrimination of HPV+OPC pENE on pre-operative CT remains non-different than blind guessing, with high interrater variability and low diagnostic accuracy, regardless of clinician specialty. While minor differences in diagnostic performance among specialties are noted, they do not significantly affect the overall poor agreement and discrimination rates observed. The findings underscore the need for further research into automated detection systems or enhanced imaging techniques to improve the accuracy and reliability of iENE assessments in clinical practice.

Comparative analysis of computer-aided imaging collaboration: MRI versus CT for detection of knee joint injuries in athletes

ObjectiveThis study aims to compare the diagnostic efficacy of collaborative computer-aided (CAD) imaging diagnosis techniques in detecting knee joint injuries among athletes. The focus is on analyzing the differences in diagnostic performance between Magnetic Resonance Imaging (MRI) and Multilayer Spiral Computed Tomography (MSCT). The objective is to assess the performance of these two imaging modalities in diagnosing knee joint injuries. MethodsA total of 242 suspected knee joint injury patients were enrolled in the study. The patients underwent MRI and MSCT examinations using a self-developed Knee Joint Injury CAD system for auxiliary diagnosis. The Knee Joint Injury CAD system, employing the Monte Carlo and Markov chain algorithms, was utilized to analyze MRI and CT images, assisting physicians in fracture diagnosis. False-positive annotations were removed through discussions and consultations among physicians to obtain independent detection results. Arthroscopy was utilized as the gold standard. The diagnostic results for MRI and CT assisted by the CAD software were recorded, and metrics such as diagnostic sensitivity, specificity, and overall accuracy were calculated. Furthermore, the accuracy of diagnosing bone and ligament injuries was analyzed. ResultsThe MRI diagnostic sensitivity of the collaborative Knee Joint Injury CAD system was 94.75%, with a specificity of 97.42% and an accuracy of 96.01%. In contrast, the diagnostic sensitivity of the collaborative Knee Joint Injury CAD system for MSCT was 89.36%, with a specificity of 85.71% and an accuracy of 89.86%. A comparison between the two detection methods reveals that MRI outperforms CT in terms of imaging quality but requires a longer examination time. The average imaging quality score for MRI was 6.89 ± 0.27, with an average examination time of 28.71 ± 7.64 minutes. On the other hand, the average imaging quality score for CT was 6.71 ± 0.42, with an average examination time of 9.95 ± 2.42 min. ConclusionIn diagnosing knee joint injuries, the collaborative Knee Joint Injury CAD system with MRI assistance in computer-aided medical diagnosis demonstrates higher diagnostic sensitivity, specificity, and accuracy compared to the collaborative Knee Joint Injury CAD system with MSCT. Therefore, the clinical application of a collaborative Knee Joint Injury CAD system with MRI exhibits superior diagnostic performance in knee joint injuries, especially in detecting ligament and soft tissue injuries. However, it is worth noting that a collaborative Knee Joint Injury CAD system with MRI has a higher imaging quality score and longer examination time. The trade-off between these factors needs to be considered in clinical practice, and the choice of imaging technology should be based on specific circumstances.

Intraoperative in vivo confocal endomicroscopy of the glioma margin: performance assessment of image interpretation by neurosurgeon users.

Confocal laser endomicroscopy (CLE) is an intraoperative real-time cellular resolution imaging technology that images brain tumor histoarchitecture. Previously, we demonstrated that CLE images may be interpreted by neuropathologists to determine the presence of tumor infiltration at glioma margins. In this study, we assessed neurosurgeons' ability to interpret CLE images from glioma margins and compared their assessments to those of neuropathologists. In vivo CLE images acquired at the glioma margins that were previously reviewed by CLE-experienced neuropathologists were interpreted by four CLE-experienced neurosurgeons. A numerical scoring system from 0 to 5 and a dichotomous scoring system based on pathological features were used. Scores from assessments of hematoxylin and eosin (H&E)-stained sections and CLE images by neuropathologists from a previous study were used for comparison. Neurosurgeons' scores were compared to the H&E findings. The inter-rater agreement and diagnostic performance based on neurosurgeons' scores were calculated. The concordance between dichotomous and numerical scores was determined. In all, 4275 images from 56 glioma margin regions of interest (ROIs) were included in the analysis. With the numerical scoring system, the inter-rater agreement for neurosurgeons interpreting CLE images was moderate for all ROIs (mean agreement, 61%), which was significantly better than the inter-rater agreement for the neuropathologists (mean agreement, 48%) (p < 0.01). The inter-rater agreement for neurosurgeons using the dichotomous scoring system was 83%. The concordance between the numerical and dichotomous scoring systems was 93%. The overall sensitivity, specificity, positive predictive value, and negative predictive value were 78%, 32%, 62%, and 50%, respectively, using the numerical scoring system and 80%, 27%, 61%, and 48%, respectively, using the dichotomous scoring system. No statistically significant differences in diagnostic performance were found between the neurosurgeons and neuropathologists. Neurosurgeons' performance in interpreting CLE images was comparable to that of neuropathologists. These results suggest that CLE could be used as an intraoperative guidance tool with neurosurgeons interpreting the images with or without assistance of the neuropathologists. The dichotomous scoring system is robust yet simple and may streamline rapid, simultaneous interpretation of CLE images during imaging.

Usar una herramienta comercial de inteligencia artificial no entrenada para COVID-19 mejora ligeramente la interpretación de las radiografías con neumonía COVID-19, especialmente entre lectores inexpertos

IntroductionOur objective is to evaluate how useful an artificial intelligence (AI) tool is to chest radiograph readers with various levels of expertise for the diagnosis of COVID-19 pneumonia when the tool has been trained on a non-COVID-19 pneumonia pathology. MethodsData was collected for patients who had previously undergone a chest radiograph and digital tomosynthesis due to suspected COVID-19 pneumonia. The gold standard consisted of the readings of two expert radiologists who assessed the presence and distribution of COVID-19 pneumonia on the images. Six medical students, two radiology trainees, and two other expert thoracic radiologists participated as additional readers. Two radiograph readings and a third supported by the AI Thoracic Care Suite tool were performed. COVID-19 pneumonia distribution and probability were assessed along with the contribution made by AI. Agreement and diagnostic performance were analysed. ResultsThe sample consisted of 113 cases, of which 56 displayed lung opacities, 52.2% were female, and the mean age was 50.70±14.9. Agreement with the gold standard differed between students, trainees, and radiologists. There was a non-significant improvement for four of the six students when AI was used. The use of AI by students did not improve the COVID-19 pneumonia diagnostic performance but it did reduce the difference in diagnostic performance with the more expert radiologists. Furthermore, it had more influence on the interpretation of mild pneumonia than severe pneumonia and normal radiograph findings. AI resolved more doubts than it generated, especially among students (31.30 vs 8.32%), followed by trainees (14.45 vs 5.7%) and radiologists (10.05% vs 6.15%). ConclusionFor expert and lesser experienced radiologists, this commercial AI tool has shown no impact on chest radiograph readings of patients with suspected COVID-19 pneumonia. However, it aided the assessment of inexperienced readers and in cases of mild pneumonia.

Screening for Primary Aldosteronism by Mass Spectrometry Versus Immunoassay Measurements of Aldosterone: A Prospective Within-Patient Study.

Measurements of aldosterone by mass spectrometry are more accurate and less prone to interferences than immunoassay measurements, and may produce a more accurate aldosterone:renin ratio (ARR) when screening for primary aldosteronism (PA). Differences in diagnostic performance of the ARR using mass spectrometry vs immunoassay measurements of aldosterone were examined in 710 patients screened for PA. PA was confirmed in 153 patients and excluded in 451 others. Disease classifications were not achieved in 106 patients. Areas under receiver-operating characteristic curves (AUROC) and other measures were used to compare diagnostic performance. Mass spectrometry-based measurements yielded lower plasma aldosterone concentrations than immunoassay measurements. For the ARR based on immunoassay measurements of aldosterone, AUROCs were slightly lower (P = 0.018) than those using mass spectrometry measurements (0.895 vs 0.906). The cutoff for the ARR to reach a sensitivity of 95% was 30 and 21.5 pmol/mU by respective immunoassay and mass spectrometry-based measurements, which corresponded to specificities of 57% for both. With data restricted to patients with unilateral PA, diagnostic sensitivities of 94% with specificities >81% could be achieved at cutoffs of 68 and 52 pmol/mU for respective immunoassay and mass spectrometry measurements. Mass spectrometry-based measurements of aldosterone for the ARR provide no clear diagnostic advantage over immunoassay-based measurements. Both approaches offer limited diagnostic accuracy for the ARR as a screening test. One solution is to employ the higher cutoffs to triage patients likely to have unilateral PA for further tests and possible adrenalectomy, while using the lower cutoffs to identify others for targeted medical therapy.German Clinical Trials Register ID: DRKS00017084.

Ultrasound radiomics-based artificial intelligence model to assist in the differential diagnosis of ovarian endometrioma and ovarian dermoid cyst.

Accurately differentiating between ovarian endometrioma and ovarian dermoid cyst is of clinical significance. However, the ultrasound appearance of these two diseases is variable, occasionally causing confusion and overlap with each other. This study aimed to develop a diagnostic classification model based on ultrasound radiomics to intelligently distinguish and diagnose the two diseases. We collected ovarian ultrasound images from participants diagnosed as patients with ovarian endometrioma or ovarian dermoid cyst. Feature extraction and selection were performed using the Mann-Whitney U-test, Spearman correlation analysis, and the least absolute shrinkage and selection operator (LASSO) regression. We then input the final features into the machine learning classifiers for model construction. A nomogram was established by combining the radiomic signature and clinical signature. A total of 407 participants with 407 lesions were included and categorized into the ovarian endometriomas group (n = 200) and the dermoid cyst group (n = 207). In the test cohort, Logistic Regression (LR) achieved the highest area under curve (AUC) value (0.981, 95% CI: 0.963-1.000), the highest accuracy (94.8%), and the highest sensitivity (95.5%), while LightGBM achieved the highest specificity (97.1%). A nomogram incorporating both clinical features and radiomic features achieved the highest level of performance (AUC: 0.987, 95% CI: 0.967-1.000, accuracy: 95.1%, sensitivity: 88.0%, specificity: 100.0%, PPV: 100.0%, NPV: 88.0%, precision: 93.6%). No statistical difference in diagnostic performance was observed between the radiomic model and the nomogram (P > 0.05). The diagnostic indexes of radiomic model were comparable to that of senior radiologists and superior to that of junior radiologist. The diagnostic performance of junior radiologists significantly improved with the assistance of the model. This ultrasound radiomics-based model demonstrated superior diagnostic performance compared to those of junior radiologists and comparable diagnostic performance to those of senior radiologists, and it has the potential to enhance the diagnostic performance of junior radiologists.

Learning imaging in axial spondyloarthritis: more than just a matter of experience

ObjectiveReliable interpretation of imaging findings is essential for the diagnosis of axial spondyloarthritis (axSpA) and requires a high level of experience. We investigated experience-dependent differences in diagnostic accuracies using X-ray...

Deep learning-accelerated T2-weighted imaging versus conventional T2-weighted imaging in the female pelvic cavity: image quality and diagnostic performance.

The deep learning (DL)-based reconstruction algorithm reduces noise in magnetic resonance imaging (MRI), thereby enabling faster MRI acquisition. To compare the image quality and diagnostic performance of conventional turbo spin-echo (TSE) T2-weighted (T2W) imaging with DL-accelerated sagittal T2W imaging in the female pelvic cavity. This study evaluated 149 consecutive female pelvic MRI examinations, including conventional T2W imaging with TSE (acquisition time = 2:59) and DL-accelerated T2W imaging with breath hold (DL-BH) (1:05 [0:14 × 3 breath-holds]) in the sagittal plane. In 294 randomly ordered sagittal T2W images, two radiologists independently assessed image quality (sharpness, subjective noise, artifacts, and overall image quality), made a diagnosis for uterine leiomyomas, and scored diagnostic confidence. For the uterus and piriformis muscle, quantitative imaging analysis was also performed. Wilcoxon signed rank tests were used to compare the two sets of T2W images. In the qualitative analysis, DL-BH showed similar or significantly higher scores for all features than conventional T2W imaging (P <0.05). In the quantitative analysis, the noise in the uterus was lower in DL-BH, but the noise in the muscle was lower in conventional T2W imaging. In the uterus and muscle, the signal-to-noise ratio was significantly lower in DL-BH than in conventional T2W imaging (P <0.001). The diagnostic performance of the two sets of T2W images was not different for uterine leiomyoma. DL-accelerated sagittal T2W imaging obtained with three breath-holds demonstrated superior or comparable image quality to conventional T2W imaging with no significant difference in diagnostic performance for uterine leiomyomas.

Sex Differences in Prehospital Identification of Large Vessel Occlusion in Patients With Suspected Stroke.

Differences in clinical presentation of acute ischemic stroke between men and women may affect prehospital identification of anterior circulation large vessel occlusion (aLVO). We assessed sex differences in diagnostic performance of 8 prehospital scales to detect aLVO. We analyzed pooled individual patient data from 2 prospective cohort studies (LPSS [Leiden Prehospital Stroke Study] and PRESTO [Prehospital Triage of Patients With Suspected Stroke Study]) conducted in the Netherlands between 2018 and 2019, including consecutive patients ≥18 years suspected of acute stroke who presented within 6 hours after symptom onset. Ambulance paramedics assessed clinical items from 8 prehospital aLVO detection scales: Los Angeles Motor Scale, Rapid Arterial Occlusion Evaluation, Cincinnati Stroke Triage Assessment Tool, Cincinnati Prehospital Stroke Scale, Prehospital Acute Stroke Severity, gaze-face-arm-speech-time, Conveniently Grasped Field Assessment Stroke Triage, and Face-Arm-Speech-Time Plus Severe Arm or Leg Motor Deficit. We assessed the diagnostic performance of these scales for identifying aLVO at prespecified cut points for men and women. Of 2358 patients with suspected stroke (median age, 73 years; 47% women), 231 (10%) had aLVO (100/1114 [9%] women and 131/1244 [11%] men). The area under the curve of the scales ranged from 0.70 (95% CI, 0.65-0.75) to 0.77 (95% CI, 0.73-0.82) in women versus 0.69 (95% CI, 0.64-0.73) to 0.75 (95% CI, 0.71-0.79) in men. Positive predictive values ranged from 0.23 (95% CI, 0.20-0.27) to 0.29 (95% CI, 0.26-0.31) in women versus 0.29 (95% CI, 0.24-0.33) to 0.37 (95% CI, 0.32-0.43) in men. Negative predictive values were similar (0.95 [95% CI, 0.94-0.96] to 0.98 [95% CI, 0.97-0.98] in women versus 0.94 [95% CI, 0.93-0.95] to 0.96 [95% CI, 0.94-0.97] in men). Sensitivity of the scales was slightly higher in women than in men (0.53 [95% CI, 0.43-0.63] to 0.76 [95% CI, 0.68-0.84] versus 0.49 [95% CI, 0.40-0.57] to 0.63 [95% CI, 0.55-0.73]), whereas specificity was lower (0.79 [95% CI, 0.76-0.81] to 0.87 [95% CI, 0.84-0.89] versus 0.82 [95% CI, 0.79-0.84] to 0.90 [95% CI, 0.88-0.91]). Rapid arterial occlusion evaluation showed the highest positive predictive values in both sexes (0.29 in women and 0.37 in men), reflecting the different event rates. aLVO scales show similar diagnostic performance in both sexes. The rapid arterial occlusion evaluation scale may help optimize prehospital transport decision-making in men as well as in women with suspected stroke.

Synovial Fluid C-reactive Protein Clinical Decision Limit and Diagnostic Accuracy for Periprosthetic Joint Infection.

Introduction C-reactive protein (CRP) has long served as a prototypical biomarker for periprosthetic joint infection (PJI). Recently, synovial fluid (SF)-CRP has garnered interest as a diagnostic tool, with several studies demonstrating its diagnostic superiority over serum CRP for the diagnosis of PJI. Although previous studies have identified diagnostic thresholds for SF-CRP, they have been limited in scope and employed various CRP assays without formal validation for PJI diagnosis. This study aimed to conduct a formal single clinical laboratory validation to determine the optimal clinical decision limit of SF-CRP for the diagnosis of PJI. Methods A retrospective analysis of prospectively collected data was performed using receiver operating characteristic (ROC) and area under the curve (AUC) analyses. Synovial fluid samples from hip and knee arthroplasties, received from over 2,600 institutions,underwent clinical testing for PJI at a single clinical laboratory (CD Laboratories, Zimmer Biomet, Towson, MD) between 2017 and 2022. Samples were assayed for SF-CRP, alpha-defensin, white blood cell count, neutrophil percentage, and microbiological culture. After applying selection criteria, the samples were classified with the 2018 ICM PJI scoring system as "infected," "not infected," or "inconclusive." Data were divided into training and validation sets. The Youden Index was employed to optimize the clinical decision limit. Results A total of 96,061 samples formed the training (n = 67,242) and validation (n = 28,819) datasets. Analysis of the biomarker median values, culture positivity, anatomic distribution, and days from aspiration to testing revealed nearly identical specimen characteristics in both the training set and validation set. SF-CRP demonstrated an AUC of 0.929 (95% confidence interval (CI): 0.926-0.932) in the training set, with an optimal SF-CRP clinical decision limit for PJI diagnosis of 4.45 mg/L. Applying this cutoff to the validation dataset yielded a sensitivity of 86.1% (95% CI: 85.0-87.1%) and specificity of 87.1% (95% CI: 86.7-87.5%). No statistically significant difference in diagnostic performance was observed between the validation and training sets. Conclusion This study represents the largest single clinical laboratory evaluation of an SF-CRP assay for PJI diagnosis. The optimal CRP cutoff (4.45 mg/L) for PJI, which yielded a sensitivity of 86.1% and a specificity of 87.1%, is specific to the assay methodology and laboratory performing the assay. We propose that an SF-CRP test with a laboratory-validated optimal clinical decision limit for PJI may be preferable, in a clinical diagnostic setting, to serum CRP tests that do not have laboratory-validated clinical decision limits for PJI.

Diagnostic performance with and without artificial intelligence assistance in real-world screening mammography

PurposeTo evaluate artificial intelligence-based computer-aided diagnosis (AI-CAD) for screening mammography, we analyzed the diagnostic performance of radiologists by providing and withholding AI-CAD results alternatively every month. MethodsThis retrospective study was approved by the institutional review board with a waiver for informed consent. Between August 2020 and May 2022, 1819 consecutive women (mean age 50.8 ± 9.4 years) with 2061 screening mammography and ultrasound performed on the same day in a single institution were included. Radiologists interpreted screening mammography in clinical practice with AI-CAD results being provided or withheld alternatively by month. The AI-CAD results were retrospectively obtained for analysis even when withheld from radiologists. The diagnostic performances of radiologists and stand-alone AI-CAD were compared and the performances of radiologists with and without AI-CAD assistance were also compared by cancer detection rate, recall rate, sensitivity, specificity, accuracy and area under the receiver-operating-characteristics curve (AUC). ResultsTwenty-nine breast cancer patients and 1790 women without cancers were included. Diagnostic performances of the radiologists did not significantly differ with and without AI-CAD assistance. Radiologists with AI-CAD assistance showed the same sensitivity (76.5%) and similar specificity (92.3% vs 93.8%), AUC (0.844 vs 0.851), and recall rates (8.8% vs. 7.4%) compared to standalone AI-CAD. Radiologists without AI-CAD assistance showed lower specificity (91.9% vs 94.6%) and accuracy (91.5% vs 94.1%) and higher recall rates (8.6% vs 5.9%, all p < 0.05) compared to stand-alone AI-CAD. ConclusionRadiologists showed no significant difference in diagnostic performance when both screening mammography and ultrasound were performed with or without AI-CAD assistance for mammography. However, without AI-CAD assistance, radiologists showed lower specificity and accuracy and higher recall rates compared to stand-alone AI-CAD.

Head-to-head comparison of cone-beam breast computed tomography and mammography in the diagnosis of primary breast cancer: A systematic review and meta-analysis

IntroductionTo compare the diagnostic performance of cone-beam breast computed tomography (CBBCT) and mammography (MG) in primary breast cancer. MethodsPubMed, Embase, Web of Science, China National Knowledge Infrastructure, WanFang DATA, and China Science and Technology Journal databases were searched comprehensively from inception to March 2023. Sensitivity and specificity were calculated using bivariate random-effects models, and a summary receiver operating characteristic (SROC) curve was constructed. Bivariate I2 statistics and meta-regression analyses were also performed. The differences in diagnostic performance between CBBCT and MG were analysed using Z-test statistics. Clinical utility was explored using Fagan’s nomogram, and quality assessment was conducted utilising the Quality Assessment of Diagnostic Accuracy Studies-2 checklist. ResultsThe summary sensitivity and specificity for CBBCT in diagnosing primary breast cancer were 0.92 (95 % CI: 0.87–0.94) and 0.79 (95 % CI: 0.71–0.85), respectively, and the area under the curve (AUC) of the SROC was 0.93 (95 % CI: 0.90–0.95). For MG, the summary sensitivity and specificity were 0.77 (95 % CI: 0.69–0.83) and 0.75 (95 % CI: 0.66–0.82), respectively, with an AUC of 0.83 (95 % CI: 0.80–0.86). The Z-test revealed that the summary sensitivity of CBBCT was significantly higher than that of MG (P < 0.001). Additionally, the summary AUC of CBBCT was significantly higher than that of MG (P < 0.001). ConclusionThe diagnostic performance of CBBCT for primary breast cancer was better than that of MG. However, the results of both the CBBCT and MG are based on studies with small sample sizes. Further studies with larger sample sizes and more comprehensive designs are required to address this issue.

An economically efficient strategy for diagnosing atypia of undetermined significance or follicular lesion of undetermined significance thyroid nodules with ultrasound-based risk stratification systems and BRAFV600E testing.

The management of thyroid nodules classified as atypia of undetermined significance or follicular lesion of undetermined significance (AUS/FLUS) has been a subject of ongoing debate. Therefore, the aim of this study was to investigate a cost-effective approach for managing these nodules by combining BRAFV600E mutation analysis with the guidelines provided by the American Thyroid Association (ATA) or the American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TIRADS). This study included 762 AUS/FLUS nodules in 551 patients with a postoperative pathology. A preoperative BRAFV600E gene test and an evaluation using the ATA guidelines and ACR-TIRADS were performed. Two combined diagnostic approaches were employed: In method 1, all nodules underwent BRAFV600E gene testing, and nodules testing positive for BRAFV600E or for risk stratification systems (RSSs) were diagnosed as malignant, while those with negative results in both tests were considered benign. In method 2 (modified combination method), nodules were reclassified into low-risk (category 2 and 3 in the ATA guidelines and ACR-TIRADS), medium-risk (category 4), and high-risk (category 5) groups based on the malignancy rate of the RSSs. BRAFV600E gene testing was applied only with the medium-risk group. Nodules with positive BRAFV600E mutation were upgraded to the high-risk group, while negative cases remained in the medium-risk group. Both malignancy rates and positive BRAFV600E mutation rates increased with the increase in RSS category (P<0.001). The combination of ACR with BRAFV600E gene testing significantly improved the area under the curve (AUC) compared to the use of ACR or BRAFV600E alone (the AUCs for ACR combined with BRAFV600E, modified ACR combined with BRAFV600E, ACR alone, and BRAFV600E alone were 0.875, 0.878, 0.832, and 0.839, respectively; P<0.05 for both combinations vs. ACR or BRAFV600E alone). Similarly, ATA combined with BRAFV600E showed significant improvements in AUC compared to ATA alone (the AUCs for ATA combined with BRAFV600E, modified ATA combined with BRAFV600E, and ATA alone were 0.851, 0.846, 0.809, respectively; P<0.001 for both combination methods vs. ATA alone), but there was no significant difference observed compared to using BRAFV600E alone (P=0.450 and P=0.680 for both combination methods vs. BRAFV600E). Notably, the AUC of ACR combined with BRAFV600E was greater than that of ATA combined with BRAFV600E (P=0.047 and P=0.007 for both combination methods, respectively). There were no significant differences in diagnostic performance between the two combination approaches (P=0.428 for ACR combined with BRAFV600E and P=0.314 for ATA combined with BRAFV600E). Performing BRAFV600E gene testing only on the medium-risk groups (modified combination method) significantly reduced the rate of BRAFV600E gene testing (P<0.001) without increasing the false-negative rate (P=0.818 and P=0.394 for ACR and ATA, respectively). Incorporating the BRAFV600E gene test exclusively for nodules in the medium-risk group significantly improved diagnostic efficacy, reduced the utilization of gene tests, and maintained a consistent false-negative rate.

Diagnostic performance of generative artificial intelligences for a series of complex case reports.

Diagnostic performance of generative artificial intelligences (AIs) using large language models (LLMs) across comprehensive medical specialties is still unknown. We aimed to evaluate the diagnostic performance of generative AIs using LLMs in complex case series across comprehensive medical fields. We analyzed published case reports from the American Journal of Case Reports from January 2022 to March 2023. We excluded pediatric cases and those primarily focused on management. We utilized three generative AIs to generate the top 10 differential-diagnosis (DDx) lists from case descriptions: the fourth-generation chat generative pre-trained transformer (ChatGPT-4), Google Gemini (previously Bard), and LLM Meta AI 2 (LLaMA2) chatbot. Two independent physicians assessed the inclusion of the final diagnosis in the lists generated by the AIs. Out of 557 consecutive case reports, 392 were included. The inclusion rates of the final diagnosis within top 10 DDx lists were 86.7% (340/392) for ChatGPT-4, 68.6% (269/392) for Google Gemini, and 54.6% (214/392) for LLaMA2 chatbot. The top diagnoses matched the final diagnoses in 54.6% (214/392) for ChatGPT-4, 31.4% (123/392) for Google Gemini, and 23.0% (90/392) for LLaMA2 chatbot. ChatGPT-4 showed higher diagnostic accuracy than Google Gemini (P < 0.001) and LLaMA2 chatbot (P < 0.001). Additionally, Google Gemini outperformed LLaMA2 chatbot within the top 10 DDx lists (P < 0.001) and as the top diagnosis (P = 0.010). This study demonstrated the diagnostic performance of generative AIs including ChatGPT-4, Google Gemini, and LLaMA2 chatbot. ChatGPT-4 exhibited higher diagnostic accuracy than the other platforms. These findings suggest the importance of understanding the differences in diagnostic performance among generative AIs, especially in complex case series across comprehensive medical fields, like general medicine.