Diagnostics

IntroductionBladder cancer (BC) remains the most common malignancy of the urinary tract, with non-muscle invasive BC (NMIBC) representing the vast majority of bladder cancer patients. The current standard of care (SOC) follow-up in NMIBC patients demands an intensive schedule and requires costly and burdensome methods, driving the development of alternative, non-invasive, cost-effective methods that may complement or serve as substitutes to cystoscopy and cytology. Uromonitor(R) is a urine biomarker test that detects hotspot mutations in three genes (TERT, FGFR3, and KRAS) for the evaluation of disease recurrence. The aim of the current study was to assess its performance comparing it to the current SOC methods. Materials and MethodsA total of 528 NMIBC surveillances from 439 individual patients were enrolled from December 2021 to June 2023. All subjects underwent SOC methods and provided an urine sample before undergoing cystoscopy for Uromonitor(R) analysis. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated for recurrence and compared to the gold-standard cystoscopy plus trans-urethral resection (TURBT) pathology. ResultsUromonitor(R) displayed a sensitivity of 87.2%, with only 6 recurrences failing to be detected by the urinary biomarker test, a specificity of 99.2%, a positive predictive value (PPV) of 93.2% and a negative predictive value (NPV) of 98.8%. Cystoscopy showed a total of 22 (31,88%) false positives not confirmed by TURBT, while Uromonitor(R) presented only 3 positive tests where no suspected lesions were found in cystoscopy. Sensitivity, specificity and NPV values for Uromonitor(R) also remained high across all NMIBC grades and stages. ConclusionIn the present study, we confirmed that the Uromonitor(R) biomarker test represents a reliable tool in the detection of NMIBC recurrence in patients undergoing routine surveillance, regardless of stage and grade. It offers either an alternative or a complement to the current SOC methods, providing rapid results and a non-invasive option, potentially improving patients quality of life and helping reduce the economic burden of NMIBC follow-up. To our knowledge, this is the largest single-center study assessing Uromonitor(R)s performance and thus validating its usefulness in clinical practice.

IntroductionDiagnosing and staging breast cancer with an easy and widely useable method that can be employed worldwide in the poorest and wealthiest settings is important. Mammography is a technique that might not be available in faraway clinics and it is technically challenging whereas USG can be available in most remote areas and small hospitals far from tertiary care hospitals. Even a trainee Radiology resident can use USG BIRADS and can be used diagnostically for that, it is important to define its, diagnostic accuracy with Sensitivity, Specificity, and other diagnostic parameters. AimsTo determine the Diagnostic accuracy of USG BIRADS compared to the gold standard Histopathology report MethodologyA Retrospective cohort study was conducted at a tertiary care hospital. a total of 84 female patients presenting to Surgical OPD with complaints of a breast lump or pain were enrolled from their records. Their Breast USG results were analyzed to identify their BIRADS stage correctly and then their corresponding Histopathology report was considered the gold standard to compare the USG results against. Excel, SPSS, and Revman were used to conduct analysis and create results. Results36 of these 84 patients belonged to BIRADS 1, 2, and 5 where Sensitivity, Specificity, and PPV were calculated at 100%. No one was diagnosed with BIRADS III from USG reports. For USG BIRADS 4, in total 48 patients Sensitivity was 0.667, specificity was 0.883, and PPV was 0.364. ConclusionPatients whose USG shows Benign growth or can be diagnosed in BIRADS 1, 2, 3, and 5 can be counted as accurate and precise. When the USG diagnosis describes the patient to be in BIRADS 4, the sensitivity and PPV show poor results showing a very low probability of the patient being truly positive when the diagnosis gives a positive result.

PURPOSEAcute pancreatitis (AP) is associated with high mortality and morbidity rates in case of necrotic forms. Risk assessment should be early performed to stratify patients into higher- and lower-risk of severe form to assist triage. In severe pancreatitis, capillary permeability increases, thereby contributing to capillary leakage which explains organ failures and or tissue necrosis. The aim of this study was therefore to evaluate pancreatic permeability by perfusion CT (pCT). METHODSFrom March 2018 to November 2018, patients with suspected AP and who underwent CT at admission were prospectively included. AP cases were classified according to the revised Atlanta classification. A permeability parameter, called k-trans, was measured from pCT in 3 pancreatic areas (normal parenchyma zone, defined by an area of normal CT pattern, pathological zone, defined as an area of parenchymal enlargement and or lack of enhancement and an intermediary zone defined by an area between normal and pathological areas) by to two observers. K-trans values in necrotic and interstitial forms for each zone were compared. To estimate reproducibility of the measure, inter-observer and intra-observer agreement was evaluated by a Bland and Altman test. RESULTS15 patients were enrolled (mean age 45.50 years old, +/-17.70). Four acute pancreatitis were necrotic, and 11 interstitial. Mean k-trans in pathologic zone of necrotic forms was significantly lower (mean=0.08) than in interstitial (mean=0.53), p= 0.0003. In both forms, k-trans values were significantly lower in pathologic zones than in intermediary and normal zones and the higher k-trans values were obtained in intermediary zones. Intra-observer reproducibility was good. Inter-observer reproducibility was acceptable, one bias was reported, close to zero (-0.052) with limited statistic difference. CONCLUSIONK-trans parameter, a well-known marker of tissue permeability, can be estimated by pCT. This parameter seems to be linked to local necrosis and could be used as a discriminant mean to diagnose necrotic from interstitial types of AP in the early phase of disease.

ObjectiveTo evaluate the performance of Thermalytix, an artificial intelligence-enhanced breast thermal imaging analysis software, against unaided manual interpretation of thermographic images. MethodsIn this multi-reader study, thermal imaging data of 258 symptomatic participants from a previous clinical trial were used. These images were independently manually interpreted by 3 senior trained breast radiologists. The same images were independently evaluated by Thermalytix, which uses sophisticated machine learning analysis of thermal/ vascular radiomic parameters to generate a risk score predictive of cancer . The results of manual interpretation and Thermalytix were compared with reference standard based on standard of care (combination of mammography, ultrasound and histopathology), to determine sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) and area under receiver operating characteristic curves (AUROC). ResultsThermalytix obtained showed a sensitivity and specificity of 95.2% (90% confidence interval (CI), 90.0- 100.0) and 66.7% (CI 60.1-73.3); the NPV and PPV were 97.7% (CI 95.2%-100.3%) and 58.3% (CI 48.5%-68.2%). The (sensitivity, specificity, NPV, PPV) obtained by Reader 1, Reader 2 and Reader 3 were (60.3%, 81.5%, 51.4%, 86.4%), (74.6%,50.8%, 86.1%, 32.9%) and (71.4%, 63.8%, 87.2%, 38.5%), respectively. The AUROC of Thermalytix was 0.85, 13.7% greater than manual interpretation. ConclusionThermalytix demonstrated good clinical performance with 25% higher accuracy than manual interpretation of thermal images. Thermalytix may alleviate the known subjectivity in thennography thereby improving its performance.

IntroductionSimilar to bladder cancer, about one third of upper tract urothelial carcinoma (UTUC) present variant histology (VH). We aim to evaluate the incidence, clinical characteristics and the impact on outcomes of VH in UTUC. MethodsWe consecutively enrolled 77 patients from 2009-2022 treated with radical surgery for UTUC from a secondary and a tertiary referral center. A central pathology review of all specimens was performed by one independent uropathologist for each center. We compared pure UTUC and UTUC with VH and the accuracy of endoscopic biopsy. Descriptive and comparative analysis were used to assess association with clinical characteristics and the Kaplan-Meier estimator to compare outcomes. ResultsMedian follow-up after surgery was 51 months. VH was present in 21/77 (28%) patients and 4/21 (19%) patients had multiple variants. The most frequent VH was squamous 12/21 (57%), followed by glandular 6/21 (29%) and micropapillary 3/21 (14%). Small cell neuroendocrine bladder carcinoma was present in two patients. Nested variant was found in one patient. Muscle invasive tumor ([&ge;]pT2) was present in 29/56 (52%) patients with pure UTUC and in 18/21 (86%) patients with VH (p <0.05). Presence of carcinoma in situ was seen in 14/56 (25%) patients with pure UTUC and in 15/21 (71%) with VH (p <0.05). Cumulative 8/56 (14%) with pure UTUC had a non-intravesical recurrence (6 patients with local and 2 distant recurrence) compared to 8/21 (38%) (3 local, 3 nodal, 2 distant) in the subgroup with VH (p <0.05). Opposite effect was noted for bladder recurrence: 60% for pure UTUC vs. 29% for tumors with VH (p <0.05). Review of preoperative endoscopic biopsy did not show the presence of VH in any patients. Differences in outcomes did not reach significance: 3yr-OS 63% vs 42% (p 0.28) and 3yr-CSS 77% vs. 50% (p 0.7). ConclusionAlmost a third of UTUC present VH. Presence of VH is related to more aggressive tumor characteristics and associated with unfavorable outcomes. Due to a higher rate of extravesical recurrences in UTUC with VH, Follow-up controls should include cross sectional imaging and cystoscopy.

BackgroundUrothelial carcinoma of the bladder (UBC) comprises several molecular subtypes, which are associated with different targetable therapeutic options. However, if and how these associations extend to the rare upper tract urothelial carcinoma (UTUC) remains unclear. ObjectiveIdentifying UTUC protein-based subtypes and developing a deep-learning (DL) workflow to predict these subtypes directly from histopathological H&E slides. Design, Setting, and ParticipantsSubtypes in a retrospective cohort of 163 invasive samples were assigned on the basis of the immunohistochemical expression of three luminal (FOXA1, GATA3, CK20) and three basal (CD44, CK5, CK14) markers. DL model building relied on a transfer-learning approach. Outcome Measurements and Statistical AnalysisClassification performance was measured via repeated cross-validation, including assessment of the area under the receiver operating characteristic (AUROC). The association of the predicted subtypes with histological features, PD-L1 status, and FGFR3 mutation was investigated. Results and LimitationsDistinctive luminal and basal subtypes were identified and could be successfully predicted by the DL (AUROC 95th CI: 0.62-0.99). Predictions showed morphology as well as presence of FGFR3-mutations and PD-L1 positivity that were consistent with the predicted subtype. Testing of the DL model on an independent cohort highlighted the importance to accommodate histological subtypes. ConclusionsOur DL workflow is able to predict protein-based UTUC subtypes directly from H&E slides. Furthermore, the predicted subtypes associate with the presence of targetable genetic alterations. Patient SummaryUTUC is an aggressive, yet understudied, disease. Here, we present an artificial intelligence algorithm that can predict UTUC subtypes directly from routine histopathological slides and support the identification of patients that may benefit from targeted therapy.

BackgroundRenal cancer is a serious threat to human health and causes heavy economic burden. Enhanced CT is widely used in the diagnosis of renal tumors. However, false-positive results of enhanced CT will bring unnecessary mental pain, expensive examination costs, physical injuries, and even adverse consequences such as organ removal and loss of function; while false-negative results of enhanced CT bring delayed treatment, and patients will thus have to bear the adverse consequences of poor prognosis, high treatment costs, poor quality of life, and short survival period. There is an urgent need to find convenient, cost-effective and non-invasive diagnostic methods to reduce the false-positive and false-negative rates of enhanced CT in renal tumors. The aim of this study was to evaluate the diagnostic value of YiDiXie -SS, YiDiXie-HS and YiDiXie-D in renal cancer. Patients and methods309 subjects (malignant group, n=244; benign group, n=65) were finally included in this study. Remaining serum samples from the subjects were collected and tested by applying the YiDiXie all-cancer detection kit to evaluate the sensitivity and specificity of YiDiXie-SS, YiDiXie-HS and YiDiXie-D, respectively. ResultsYiDiXie-SS had a sensitivity of 98.6% (95% CI: 95.8% - 99.6%; 204/207) and a specificity of 71.4% (95% CI: 45.4% - 88.3%; 10/14) in renal enhanced CT-positive patients. This means that the application of YiDiXie -SS reduces the false-positive rate of renal enhanced CT by 71.4% (95% CI: 45.4% - 88.3%; 10/14) with essentially no increase in malignant tumor leakage. The sensitivity of YiDiXie-HS in renal enhanced CT-negative patients was 86.5% ( 95% CI: 72.0% - 94.1%; 32/37) and its specificity was 84.3% (95% CI: 72.0% - 91.8%; 43/51). This means that the application of YiDiXie-HS reduces the false-negative rate of enhanced CT by 86.5% (95% CI: 72.0% - 94.1%; 32/37). The sensitivity of YiDiXie -D in renal enhanced CT-positive patients was 31.9% (95% CI: 25.9% - 38.5%; 66/207) and its specificity was 92.9% (95% CI: 68.5% - 99.6%; 13/14). This means that YiDiXie-SS reduces the false positive rate of enhanced CT by 92.9% (95% CI: 68.5% - 99.6%; 13/14). ConclusionYiDiXie-SS dramatically reduces the false-positive rate of renal enhanced CT with essentially no increase in delayed treatment of malignant tumors. YiDiXie-HS dramatically reduces the false-negative rate of renal enhanced CT. YiDiXie -D dramatically reduces the false-positive rate of renal enhanced CT. The YiDiXie test has significant diagnostic value in renal tumors, and is expected to solve the problems of "high false-positive rate of renal enhanced CT" and "high false-negative rate of renal enhanced CT". Clinical trial numberChiCTR2200066840.

BackgroundMammographic (or breast) density is an established risk factor for breast cancer. There are a variety of approaches to measurement including quantitative, semi-automated and automated approaches. We present a new automated measure, AutoCumulus, learnt from applying deep learning to semi-automated measures. MethodsWe used mammograms of 9,057 population-screened women in the BRAIx study for which semi-automated measurements of mammographic density had been made by experienced readers using the CUMULUS software. The dataset was split into training, testing, and validation sets (80%, 10%, 10%, respectively). We applied a deep learning regression model (fine-tuned ConvNeXtSmall) to estimate percentage density and assessed performance by the correlation between estimated and measured percent density and a Bland-Altman plot. The automated measure was tested on an independent CSAW-CC dataset in which density had been measured using the LIBRA software, comparing measures for left and right breasts, sensitivity for high sensitivity, and areas under the receiver operating characteristic curve (AUCs). ResultsBased on the testing dataset, the correlation in percent density between the automated and human measures was 0.95, and the differences were only slightly larger for women with higher density. Based on the CSAW-CC dataset, AltoCumulus outperformed LIBRA in correlation between left and right breast (0.95 versus 0.79; P<0.001), specificity for 95% sensitivity (13% versus 10% (P<0.001)), and AUC (0.638 cf. 0.597; P<0.001). ConclusionWe have created an automated measure of mammographic density that is accurate and gives superior performance on repeatability within a woman, and for prediction of interval cancers, than another well-established automated measure.

BackgroundPrecise prediction of residual tumour size following neoadjuvant systemic therapy for breast cancer is crucial in assessing response and surgical decision making. Our study is aimed at assessing the performance of conventional imaging modalities like ultrasound and mammography in predicting the residual tumour size after neoadjuvant systemic therapy and in evaluating the impact of imaging on the surgical outcomes. MethodsWe retrospectively compared the tumour size measured by ultrasonography and mammography and the residual tumour size on final histopathology in 109 patients. Concordance was defined as a size difference within 25% of the histopathology size. We also looked at the distribution of concordance between different T status and molecular subtypes, accuracy of USG in predicting pathological complete response and axillary lymph nodal metastasis and also surgical outcomes in the discordant cases. ResultsThe concordance rates of mammography and ultrasonography were 68.2% and 52.3% respectively without statistically significant difference between the two modalities (p = 0.081). Combination of both the modalities had a concordance rate of 57.8%. Ultrasonography had accuracy of 81.7% for predicting pathological complete response and 79.8% for predicting axillary nodal metastasis. We did not identify any influence of histologic subtype on the associations between preoperative measurements and pathology size or the accuracy for detecting pathological complete response (p values 0.43 and 0.46 respectively). In 12 cases, the radiology-pathology discordance could have led to large excision volume surgeries. In the overall cohort, recurrence free survival and overall survival rates at median follow up of 19.1 month were 87.2% and 95.4% respectively. ConclusionsUltrasound and mammography showed moderate concordance with pathology for estimation of the residual tumour size without any significant difference in the performance between the two. Despite the moderate concordance, surgical outcomes were fairly well managed in the discordant cases with the oncoplastic surgical techniques. Our study highlights the usefulness of the cheaper and widely available conventional imaging modalities in the developing countries where the cost of treatment is to be contained.

PurposeTo evaluate the diagnostic value of chest CT in 2019 novel coronavirus disease (COVID-19), using the reverse transcription polymerase chain reaction (RT-PCR) as a reference standard. At the same time, the imaging features of CT in confirmed COVID-19 patients would be summarized. MethodsA comprehensive literature search of 5 electronic databases was performed. The pooled sensitivity, specificity, positive predictive value, and negative predictive value were calculated using the random-effects model and the summary receiver operating characteristic (SROC) curve. We also conducted a meta-analysis to estimate the pooled incidence of the chest CT imaging findings and the 95% confidence interval (95%CI). Meta-regression analysis was used to explore the source of heterogeneity. ResultsOverall, 25 articles comprising 4,857 patients were included. The pooled sensitivity of CT was 93% (95% CI, 89-96%) and specificity was 44% (95% CI, 27-62%). The area under the SROC curve was 0.94 (95% CI, 0.91-0.96). For the RT-PCR assay, the pooled sensitivity of the initial test and the missed diagnosis rate after the second-round test were 76% (95% CI: 59-89%; I2=96%) and 26% (95% CI: 14-39%; I2=45%), respectively. According to the subgroup analysis, the diagnostic sensitivity of CT in Hubei was higher than that in other regions. Besides, the most common patterns on CT imaging finding was ground glass opacities (GGO) 58% (95% CI: 49-70%), followed by air bronchogram 51% (95% CI: 31-70%). Lesions were inclined to distribute in peripheral 64% (95% CI: 49-78%), and the incidence of bilateral lung involvement was 69% (95% CI: 58-79%). ConclusionsThere were still several cases of missed diagnosis after multiple RT-PCR examinations. In high-prevalence areas, CT could be recommended as an auxiliary screening method for RT-PCR. Key pointsO_LITaking RT-PCR as the reference standard, the pooled sensitivity of CT was 93% (95% CI, 89-96%) and the specificity was 44% (95% CI, 27-62%). The area under the SROC curve was 0.94 (95% CI, 0.91-0.96). C_LIO_LIFor the RT-PCR assay, the pooled sensitivity of the initial test and the missed diagnosis rate after the second-round test were 76% (95% CI: 59-89%) and 26% (95% CI: 14-39%), respectively. C_LIO_LIGGO was the key sign of the CT imaging, with an incidence of 58% (95% CI: 49-70%) in patients with SARS-CoV-2 infection. Pneumonia lesions were inclined to distribute in peripheral 64% (95% CI: 49-78%) and bilateral 69% (95% CI: 58-79%) lung lobes. C_LI

BackgroundNeoadjuvant chemotherapy (NACT) is used to downstage breast cancer prior to surgery. Image monitoring is essential to guide treatment and to assess in vivo chemosensitivity. Breast MRI is considered the gold-standard imaging technique; however, it is contraindicated or poorly tolerated in some patients and may be hard to access. Evidence suggests contrast enhanced spectral mammography (CESM) may approach the accuracy of MRI. This novel pilot study investigates whether the addition of digital breast tomosynthesis (DBT) to CESM increases the accuracy of response prediction. ResultsSixteen cancers in fourteen patients were imaged with CESM+DBT and MRI following completion of NACT. Ten cancers demonstrated pathological complete response (pCR) defined as absence of residual invasive disease. Greatest accuracy for predicting pCR was with CESM contrast-enhancement only (accuracy 81.3%, sensitivity 100%, specificity 57.1%), followed by MRI (accuracy 62.5%, sensitivity 44.4%, specificity 85.7%). Concordance with invasive tumour size was greater for CESM than MRI, concordance-coefficients 0.70 vs 0.66 respectively. MRI demonstrated greatest concordance with whole tumour size followed by CESM contrast-enhancement plus microcalcification, concordance-coefficients 0.86 vs 0.69. The addition of DBT did not improve accuracy for prediction of pCR or residual disease size. Whereas CESM+DBT tended to underestimate size of residual disease, MRI tended to overestimate but no significant differences were seen (p>0.05). ConclusionsCESM contrast-enhancement plus microcalcification is similar to MRI for predicting residual disease post-NACT. Size of enhancement alone demonstrates best concordance with invasive disease. Inclusion of residual microcalcification improves concordance with DCIS. The addition of DBT to CESM does not improve accuracy. HighlightsO_LINo benefit of adding DBT to CESM for NACT response prediction C_LIO_LICESM appears similar to MRI for predicting response to NACT C_LIO_LICESM has greatest accuracy for residual invasive tumour size. C_LIO_LICESM+calcification has greater accuracy for predicting residual in situ disease. C_LI

BackgroundWhile mammography is commonly used for breast cancer detection, its widespread implementation in resource-constrained nations is challenging. Artificial intelligence-based Thermalytix is a low-cost, portable, radiation-free, automated test for breast cancer detection in women of all ages. Although used in India, the efficacy of Thermalytix has not been tested in an African population. ObjectivesTo assess the agreement and correlation coefficient of Thermalytix output with radiologist-reported mammography, in a Zambian tertiary care population. MethodologyIn October 2023, 169 women were evaluated with both Thermalytix and standard mammography at Maina Soko Military Hospital at Lusaka. Thermalytix uses advanced machine learning algorithms to interpret breast thermal scans and generates a quantitative score indicating the likelihood of malignancy. All women underwent both tests, with results blinded both ways. Subsequently the Spearman correlation coefficient and level of agreement between Thermalytix output and BIRADs scoring from radiologist-interpreted mammography was calculated. Results144 women with complete data were analysed in this report, with median age of 50 years (53.5% postmenopausal, 65.3% asymptomatic). Six women were assessed as mammography test positive and 138 as mammography negative; in these, the correlation between Thermalytix and mammography using Spearman test of rank correlation was 0.9 [very strong], and using the US FDA recommended test of agreement, positive agreement was obtained in 83.3%. ConclusionDemonstrating a very strong correlation and level of agreement with mammography, along with its good sensitivity, specificity and negative predictive value in previous clinical trials, Thermalytix has the potential to be an additional tool in the early detection of breast cancer in Zambia.

BackgroundEsophageal cancer is a major health concern in Ethiopia and is often diagnosed at advanced stages because of limited access to diagnostic tools. Computed tomography is the primary imaging modality used for preoperative staging at Tikur Anbessa Specialized Hospital and its affiliated centers. However, its diagnostic accuracy has not been well studied locally. ObjectiveTo evaluate the diagnostic accuracy of preoperative CT TNM staging of esophageal cancer in surgical patients, postoperative histopathological findings were used as the gold standard. MethodThis retrospective cross-sectional study included 121 patients with histologically confirmed esophageal cancer who underwent preoperative CT staging and surgery between January 2021 and December 2024 at TASH and affiliated hospitals. Data were collected from medical records and CT images and analyzed via SPSS version 27. The sensitivity, specificity, PPV, NPV, and diagnostic accuracy were calculated. ResultsAmong the patients, 57.02% were female and 42.98% male, with a median age of 54 years. SCC was the predominant histologic type (84.3%). Preoperative CT staging revealed T3 in 67.8% and T4 in 32.2% of patients. Nodal staging revealed N0 in 77.7% of the patients. The diagnostic accuracies of CT for the T3 and T4 stages were 49.6% and 67.8%, respectively. For the N0 to N3 stages, the accuracy ranged from 61.9% to 95%. The combined sensitivity and specificity for T staging were 82.6% and 20%, respectively; for N staging, they were 88.1% and 25.4%, respectively. ConclusionCT imaging has moderate accuracy in staging esophageal cancer but has limitations, particularly in differentiating tumor depth and nodal involvement. These findings underscore the need for multimodal imaging approaches, including MRI, PET, and EUS, where available, to improve preoperative assessment and patient outcomes.

AimTo assess the generalisability of a deep learning (DL) system for screening mammography developed at New York University (NYU), USA (1, 2) in a South Australian (SA) dataset. Methods and MaterialsClients with pathology-proven lesions (n=3,160) and age-matched controls (n=3,240) were selected from women screened at BreastScreen SA from January 2010 to December 2016 (n clients=207,691) and split into training, validation and test subsets (70%, 15%, 15% respectively). The primary outcome was area under the curve (AUC), in the SA Test Set 1 (SATS1), differentiating invasive breast cancer or ductal carcinoma in situ (n=469) from age-matched controls (n=490) and benign lesions (n=44). The NYU system was tested statically, after training without transfer learning (TL), after retraining with TL and without (NYU1) and with (NYU2) heatmaps. ResultsThe static NYU1 model AUCs in the NYU test set (NYTS) and SATS1 were 83.0%(95%CI=82.4%-83.6%)(2) and 75.8%(95%CI=72.6%-78.8%), respectively. Static NYU2 AUCs in the NYTS and SATS1 were 88.6%(95%CI=88.3%-88.9%)(2) and 84.5%(95%CI=81.9%-86.8%), respectively. Training of NYU1 and NYU2 without TL achieved AUCs in the SATS1 of 65.8% (95%CI=62.2%-69.1%) and 85.9%(95%CI=83.5%-88.2%), respectively. Retraining of NYU1 and NYU2 with TL resulted in AUCs of 82.4%(95%CI=79.7-84.9%) and 86.3%(95%CI=84.0-88.5%) respectively. ConclusionWe did not fully reproduce the reported performance of NYU on a local dataset; local retraining with TL approximated this level of performance. Optimising models for local clinical environments may improve performance. The generalisation of DL systems to new environments may be challenging. Key ContributionsIn this study, the original performance of deep learning models for screening mammography was reduced in an independent clinical population. Deep learning (DL) systems for mammography require local testing and may benefit from local retraining. An openly available DL system approximates human performance in an independent dataset. There are multiple potential sources of reduced deep learning system performance when deployed to a new dataset and population.

BackgroundBreast microcalcification diagnostics are challenging due to their subtle presentation, overlapping with benign findings, and high inter-reader variability, often leading to unnecessary biopsies. While deep learning (DL) models - particularly deep convolutional neural networks (DCNNs) - have shown potential to improve diagnostic accuracy, their clinical application remains limited by the need for large annotated datasets and the "black box" nature of their decision-making. PurposeTo develop and validate a deep learning model (DCNN) using a double transfer learning (d-TL) strategy for classifying suspected mammographic microcalcifications, with explainable AI (XAI) techniques to support model interpretability. Material and methodsA retrospective dataset of 396 annotated regions of interest (ROIs) from full-field digital mammography (FFDM) images of 194 patients who underwent stereotactic vacuum-assisted biopsy at the Womens Hospital radiological department, Helsinki University Hospital, was collected. The dataset was randomly split into training and test sets (24% test set, balanced for benign and malignant cases). A ResNeXt-based DCNN was developed using a d-TL approach: first pretrained on ImageNet, then adapted using an intermediate mammography dataset before fine-tuning on the target microcalcification data. Saliency maps were generated using Gradient-weighted Class Activation Mapping (Grad-CAM) to evaluate the visual relevance of model predictions. Diagnostic performance was compared to a radiologists BI-RADS-based assessment, using final histopathology as the reference standard. ResultsThe ensemble DCNN achieved an area under the ROC curve (AUC) of 0.76, with 65% sensitivity, 83% specificity, 79% positive predictive value (PPV), and 70% accuracy. The radiologist achieved an AUC of 0.65 with 100% sensitivity but lower specificity (30%) and PPV (59%). Grad-CAM visualizations showed consistent activation of the correct ROIs, even in misclassified cases where confidence scores fell below the threshold. ConclusionThe DCNN model utilizing d-TL achieved performance comparable to radiologists, with higher specificity and PPV than BI-RADS. The approach addresses data limitation issues and may help reduce additional imaging and unnecessary biopsies.

ObjectivesTo elucidate if synthetic contrast enhanced computed tomography (CECT) images created from plain CT images using deep neural networks (DNN) could be used for screening, clinical diagnosis, and postoperative follow-up of small-diameter renal tumors by comparing the concordance rate between real and synthetic CECT images and the diagnoses according to 10 urologists. MethodsThis retrospective, multicenter study included 155 patients (artificial intelligence training cohort [n=99], validation cohort [n=56]) who underwent surgery for small-diameter ([&le;]40 mm) renal tumors, with the pathological diagnosis of renal cell carcinoma, during 2010-2020. Preoperatively, dynamic plain CT and CECT images were obtained. We created a learned DNN using pix2pix. We examined the quality of the synthetic CECT images created using this DNN and compared them with real CECT images using the zero-mean normalized cross-correlation parameter. We assessed concordance rates between real and synthetic images and diagnoses according to 10 urologists by creating a receiver operating characteristic curve and calculating the area under the curve (AUC). ResultsThe synthetic CECT images were highly concordant with the real CECT images, regardless of the existence or morphology of the renal tumor. Regarding the concordance rate, a greater AUC was obtained with synthetic CECT (AUC=0.892) than with only CT (AUC=0.720; p<0.001). ConclusionsThis study is the first to use DNN to create a high-quality synthetic CECT image that was highly concordant with a real CECT image. Synthetic CECT images could be used for urological diagnoses and clinical screening.

BackgroundUroepithelial carcinoma is a serious threat to human health and causes heavy economic burden. Enhanced CT is widely used in screening or preliminary diagnosis of uroepithelial tumors. However, false-positive results of enhanced CT will bring unnecessary mental pain, expensive examination costs, physical injuries, and other adverse consequences; while false-negative results of enhanced CT bring delayed treatment, and patients will thus have to bear the adverse consequences of poor prognosis, high treatment costs, poor quality of life, and short survival period. There is an urgent need to find convenient, cost-effective and non-invasive diagnostic methods to reduce the false-negative and false-positive rates of enhanced CT in uroepithelial tumors. The aim of this study was to evaluate the diagnostic value of YiDiXie-SS and YiDiXie-HS in uroepithelial carcinoma. Patients and methods319 subjects (malignant group, n=240; benign group, n=79) were finally included in this study. Remaining serum samples from the subjects were collected and tested by applying the YiDiXie all-cancer detection kit to evaluate the sensitivity and specificity of YiDiXie-SS and YiDiXie-HS. ResultsThe sensitivity of YiDiXie-SS in enhanced CT-positive patients was 96.3% (95% CI: 96.3% - 98.3%; 158/164)and its specificity was 64.3% (95% CI: 38.8% - 83.7%; 9/14). This means that the application of YiDiXie -SS reduces the false-positive rate of urological enhanced CT by 64.3% (95% CI: 38.8% - 83.7%; 9/14) with essentially no increase in malignancy leakage. The sensitivity of YiDiXie-HS in enhanced CT-negative patients was 85.5% (95% CI: 75.9% - 91.7%; 65/76)and its specificity was 84.6% (95% CI: 73.9% - 91.4%; 55/65). This means that the application of YiDiXie-HS reduces the false-negative rate of urological enhanced CT by 85.5% (95% CI: 75.9% - 91.7%; 65/76). ConclusionYiDiXie -SS substantially reduces the rate of urological enhanced CT false positives with essentially no increase in delayed treatment of malignancies. YiDiXie-HS substantially reduces the false negative rate of urological enhanced CT. YiDiXie -SS and YiDiXie -HS have an important diagnostic value in uroepithelial carcinoma, and are expected to solve the problems of "high false-positive rate of urological enhanced CT" and "high false-negative rate of urological enhanced CT" in uroepithelial carcinoma. Clinical trial numberChiCTR2200066840.

ObjectiveBladder cancer (BC) is the most common malignancy of the urinary system and among the most frequently diagnosed cancers worldwide. This systematic review and meta-analysis aimed to evaluate the diagnostic accuracy of commercially available urinary biomarkers tests (UBTs) for detecting BC recurrence, focusing on pooled sensitivity and specificity estimates across different tests. MethodsA systematic search was performed on PubMed and EMBASE up to May 2025 to identify studies assessing recurrence of BC in previously diagnosed patients using non-FDA approved UBTs, including Xpert Bladder Cancer, Bladder Epicheck, ADXbladder and Uromonitor. Eligible studies were synthesized using the bivariate Generalized Linear Mixed Model (GLMM) model. ResultsOut of 307 initially screened citations, 33 studies met the eligibility criteria, encompassing a total of 10,478 patients. Xpert Bladder Cancer was evaluated on 13 studies and Bladder Epicheck was assessed on 10 studies. ADXbladder and Uromonitor were assessed in four and six studies, respectively. Meta-analyses included 13 studies for Xpert Bladder Cancer and 10 studies for Bladder Epicheck, yielding pooled sensitivity (95% CI) and specificity (95% CI) estimates of 0.71 (0.61-0.79) and 0.78 (0.74-0.82) for Xpert Bladder Cancer, and 0.75 (0.61-0.86) and 0.90 (0.84-0.94) for Bladder Epicheck. For ADXbladder and Uromonitor, meta-analyses incorporated four and six studies, respectively, resulting in pooled sensitivity and specificity values of 0.55 (0.40-0.69) and 0.60 (0.44-0.75) for ADXbladder, and 0.77 (0.61-0.88) and 0.96 (0.91-0.98) for Uromonitor. ConclusionsThis meta-analysis reveals that commercially UBTs for BC recurrence have varying diagnostic accuracy. Among the evaluated tests, Uromonitor demonstrated the highest pooled sensitivity and specificity, while Xpert Bladder Cancer and Bladder Epicheck showed reliable diagnostic performance. Further research is needed particularly for less extensively studied assays to establish their diagnostic performance.

BackgroundHuman Papillomavirus (HPV) infection is a potential risk for cervical cancer, the latter is the fourth leading cause of cancer related death worldwide. Effective testing procedures are particularly important in its prevention and management. More accurate HPV tests replace traditional cytology-based screening, and they are in line with national and international guidelines that recommend primary HPV testing due to its better efficacy. The purpose of the study was to confirm that these Gynius HPV detection kits were equal in efficiency compared with Roche Cobas(R) HPV assay-based testing, which could better facilitate earlier detection and stratification by risk level for cervical lesions. MethodsThe prospective parallel control study recruited 1000 women from the district hospitals of Kibagabaga and Muhima in Rwanda. Participants were tested for HPV DNA tests using both Roche and Gynius kits and then the samples were sent to Rwanda National Reference Laboratory. We analyzed the concordance of two Gynius variants (liquid and lyophilized ) and Roche across for several different HPV subtypes, such as 16, 18 or other high risk types (31,33,35,39,45,51,52,56,58,59,66,68,73,53,82 &26). Statistics like sensitivity, specificity, predictive values and consistency (Kappa) were calculated to compare the concordance of assays. ResultsWe found that the Gynius HPV solution is exceptionally easy to implement in laboratory settings within low- and middle-income countries (LMICs). It offers effective sampling control through color detection, making it highly suitable for self-sampling procedures. Additionally, the Gynius HPV solution includes an automated extraction system that completes full plate extraction in under 20 minutes. The Gynius solution is compatible with various qPCR systems, eliminating the need for costly qPCR machine repurchases. Using lyophilized reagents, which can be stored and transported at room temperature, it bypasses the need for a cold chain and refrigeration. The Gynius solution also has a high sample processing capacity. For instance, using the BioRad CFX96, it can produce up to 1,200 results per day per machine; with the BioRad CFX384, it can generate up to 4,000 results per day--making it ideal for large-scale screening. In a comparison of 987 samples, high-risk HPV detection rates were consistent across all three tests: Roche (20%), Gynius liquid (23%), and Gynius lyophilized (21%). Concordance analysis showed high agreement rates, with both Gynius kits achieving over 80% compatibility with Roche for detecting HPV 16, 18, and other virus subtypes. ConclusionThis study provides compelling evidence for the efficacy of the Gynius HPV kits (both lyophilized and liquid formulations), which demonstrated high concordance rates with established assays for detecting high-risk HPV genotypes, particularly HPV 16 and 18. Engineered specifically for low- and middle-income countries (LMICs), the Gynius HPV kits integrate innovative sampling and detection technologies optimized for resource-limited settings, facilitating seamless implementation. These findings indicate that Gynius HPV kits present a viable and scalable solution for cervical cancer screening programs, with the potential to significantly enhance early detection capabilities and broaden preventive care accessibility.

Objectivesto evaluate the performance of Artificial Intelligence (AI) methods to detect covid-19 from chest images (X-Ray and CT scans). MethodsChest CT scans and X-Ray images collected from different centers and institutions were downloaded and combined together. Images were separated by patient and 66% of the patients were used to develop and train AI image-based classifiers. Then, the AI automated classifiers were evaluated on a separate set of patients (the remaining 33% patients). Results (Chest X-Ray)Five different data sources were combined for a total of N=9,841 patients (1,733 with covid-19, 810 with bacterial tuberculosis and 7,298 healthy patients). The test sample size was N=3,528 patients. The best AI method reached an Area Under the Curve (AUC) for covid-19 detection of 99%, with a detection rate of 96.4% at 1.0% false positive rate. Results (Chest CT scans)Two different data sources were combined for a total of N=363 patients (191 having covid-19 and 172 healthy patients). The test sample size was N=121 patients. The best AI method reached an AUC for covid-19 detection of 90.9%, with a detection rate of 90.6% at 24.6% false positive rate. ConclusionsComputer aided automatic covid-19 detection from chest X-ray images showed promising results to be used as screening tool during the covid-19 outbreak. The developed method may help to manage patients better in case access to PCR testing is not possible or to detect patients with symptoms missed in a first round of PCR testing. The method will be made available online (www.quantuscovid19.org). These results merit further evaluation collecting more images. We hope this study will allow us to start such collaborations.