False-negative Interpretations Research Articles

Statistical errors in scientific research: A narrative review

ABSTRACT Statistics is the backbone of scientific research. Statistical errors are an inevitable part of any study. A researcher needs to be aware of various types of statistical errors, their implications, and steps to minimize them. Essentially, statistical errors may creep into medical literature either by deriving false-positive or false-negative interpretations during analysis or through fallacies in data collection. Manipulation of statistical errors to derive one’s preferred conclusions from a study is unethical. We searched medical literature in databases such as PubMed, Embase, and Cochrane for the articles published between July 2014 and June 2023 followed by abstract screening which led to the study of 8 full-length articles to gather the pertinent material. This review highlights many of the common statistical errors along with examples and suggested corrective actions.

Identification and Localization of Indolent and Aggressive Prostate Cancers Using Multilevel Bi-LSTM.

Identifying indolent and aggressive prostate cancers is a critical problem for optimal treatment. The existing approaches of prostate cancer detection are facing challenges as the techniques rely on ground truth labels with limited accuracy, and histological similarity, and do not consider the disease pathology characteristics, and indefinite differences in appearance between the cancerous and healthy tissue lead to many false positive and false negative interpretations. Hence, this research introduces a comprehensive framework designed to achieve accurate identification and localization of prostate cancers, irrespective of their aggressiveness. This is accomplished through the utilization of a sophisticated multilevel bidirectional long short-term memory (Bi-LSTM) model. The pre-processed images are subjected to multilevel feature map-based U-Net segmentation, bolstered by ResNet-101 and a channel-based attention module that improves the performance. Subsequently, segmented images undergo feature extraction, encompassing various feature types, including statistical features, a global hybrid-based feature map, and a ResNet-101 feature map that enhances the detection accuracy. The extracted features are fed to the multilevel Bi-LSTM model, further optimized through channel and spatial attention mechanisms that offer the effective localization and recognition of complex structures of cancer. Further, the framework represents a promising approach for enhancing the diagnosis and localization of prostate cancers, encompassing both indolent and aggressive cases. Rigorous testing on a distinct dataset demonstrates the model's effectiveness, with performance evaluated through key metrics which are reported as 96.72%, 96.17%, and 96.17% for accuracy, sensitivity, and specificity respectively utilizing the dataset 1. For dataset 2, the model achieves the accuracy, sensitivity, and specificity values of 94.41%, 93.10%, and 94.96% respectively. These results surpass the efficiency of alternative methods.

Discordant and false-negative interpretations at digital breast tomosynthesis in the prospective Oslo Tomosynthesis Screening Trial (OTST) using independent double reading

ObjectivesTo analyze discordant and false-negatives of double reading digital breast tomosynthesis (DBT) versus digital mammography (DM) including reading times in the Oslo Tomosynthesis Screening Trial (OTST), and reclassify these in a retrospective reader study as missed, minimal sign, or true-negatives.MethodsThe prospective OTST comparing double reading DBT vs. DM had paired design with four parallel arms: DM, DM + computer aided detection, DBT + DM, and DBT + synthetic mammography. Eight radiologists interpreted images in batches using a 5-point scale. Reading time was automatically recorded. A retrospective reader study including four radiologists classified screen-detected cancers with at least one false-negative score and screening examinations of interval cancers as negative, non-specific minimal sign, significant minimal sign, and missed; the two latter groups are defined “actionable.” Statistics included chi-square, Fisher’s exact, McNemar’s, and Mann–Whitney U tests.ResultsDiscordant rate (cancer missed by one reader) for screen-detected cancers was overall comparable (DBT (31% [71/227]) and DM (30% [52/175]), p = .81), significantly lower at DBT for spiculated cancers (DBT, 19% [20/106] vs. DM, 36% [38/106], p = .003), but high (28/49 = 57%, p = 0.001) for DBT-only detected spiculated cancers. Reading time and sensitivity varied among readers. False-negative DBT-only detected spiculated cancers had shorter reading time than true-negatives in 46% (13/28). Retrospective evaluation classified the following DBT exams “actionable”: three missed by both readers, 95% (39/41) of discordant cancers detected by both modes, all 30 discordant DBT-only cancers, 25% (13/51) of interval cancers.ConclusionsDiscordant rate was overall comparable for DBT and DM, significantly lower at DBT for spiculated cancers, but high for DBT-only detected spiculated lesions. Most false-negative screen-detected DBT were classified as “actionable.”Clinical relevance statementRetrospective evaluation of false-negative interpretations from the Oslo Tomosynthesis Screening Trial shows that most discordant and several interval cancers could have been detected at screening. This underlines the potential for modern AI-based reading aids and triage, as high-volume screening is a demanding task.Key Points• Digital breast tomosynthesis (DBT) screening is more sensitive and has higher specificity compared to digital mammography screening, but high-volume DBT screening is a demanding task which can result in high discordance rate among readers.• Independent double reading DBT screening had overall comparable discordance rate as digital mammography, lower for spiculated masses seen on both modalities, and higher for small spiculated cancer seen only on DBT.• Almost all discordant digital breast tomosynthesis-detected cancers (72 of 74) and 25% (13 of 51) of the interval cancers in the Oslo Tomosynthesis Screening Trial were retrospectively classified as actionable and could have been detected by the readers.

'Earlier than Early' Detection of Breast Cancer in Israeli BRCA Mutation Carriers Applying AI-Based Analysis to Consecutive MRI Scans.

Female BRCA1/BRCA2 (=BRCA) pathogenic variants (PVs) carriers are at a substantially higher risk for developing breast cancer (BC) compared with the average risk population. Detection of BC at an early stage significantly improves prognosis. To facilitate early BC detection, a surveillance scheme is offered to BRCA PV carriers from age 25-30 years that includes annual MRI based breast imaging. Indeed, adherence to the recommended scheme has been shown to be associated with earlier disease stages at BC diagnosis, more in-situ pathology, smaller tumors, and less axillary involvement. While MRI is the most sensitive modality for BC detection in BRCA PV carriers, there are a significant number of overlooked or misinterpreted radiological lesions (mostly enhancing foci), leading to a delayed BC diagnosis at a more advanced stage. In this study we developed an artificial intelligence (AI)-network, aimed at a more accurate classification of enhancing foci, in MRIs of BRCA PV carriers, thus reducing false-negative interpretations. Retrospectively identified foci in prior MRIs that were either diagnosed as BC or benign/normal in a subsequent MRI were manually segmented and served as input for a convolutional network architecture. The model was successful in classification of 65% of the cancerous foci, most of them triple-negative BC. If validated, applying this scheme routinely may facilitate 'earlier than early' BC diagnosis in BRCA PV carriers.

PSUN51 Re-Calibrating Aldosterone Assay Interpretations: Immunoassay and Liquid Chromatography-Tandem Mass Spectrometry Measurements Across Multiple Controlled Physiologic Conditions

Abstract Background Clinicians frequently rely on aldosterone thresholds derived from older immunoassays to diagnose primary aldosteronism. However, liquid chromatography tandem mass spectrometry (LC-MS/MS) is in increasingly widespread clinical use and has been reported to yield lower aldosterone concentrations. Given that primary aldosteronism is highly prevalent and morbid, yet rarely diagnosed, incorrect interpretations of aldosterone levels can have significant public health impact. To address the potential for incorrect interpretations across the entire dynamic range of aldosterone physiology, we compared aldosterone measurements using LC-MS/MS and immunoassay under controlled physiologic conditions. Methods Untreated, normotensive, overweight volunteers (n=48) underwent prospective characterization of aldosterone production during four controlled conditions: oral sodium suppression and dietary sodium restriction (to capture the full range of Angiotensin II-mediated aldosterone production) and dexamethasone suppression and cosyntropin stimulation (to capture the full range of ACTH-mediated aldosterone production). Serum aldosterone from each condition (n=188 samples) was measured by both immunoassay and LC-MS/MS. Results Serum aldosterone concentrations by LC-MS/MS and immunoassay had a Pearson correlation of 0.69 (p<0.001), with good agreement (intraclass correlation coefficient 0.76 [95% CI 0.52-0.87]). Overall, aldosterone was 37.2% lower by LC-MS/MS than immunoassay (median 10.5 [IQR 3.8, 21.9] vs. 19.6 [9.5, 28.0] ng/dL; p<0.001) with an average bias of 7.2 (SD 12.0) ng/dL by Bland-Altman analysis. The most notable discrepancy was observed in the clinically discriminatory range below 20 ng/dL: 9.9 [7.1, 13.8] ng/dL by immunoassay corresponded to 5.5 [1.4, 8.9] ng/dL by LC-MS/MS (p<0.001), a median difference of 49.6% [20.6, 75.2%] higher by immunoassay. Following oral sodium suppression testing, the aldosterone-to-renin ratio was 4-fold higher by immunoassay than by LC-MS/MS (27.2 [19.7, 62.4] vs. 6.4 [3.5, 19.1] ng/dL per ng/mL/h; p<0.001). Conclusions Aldosterone measurements are systematically lower by LC-MS/MS than by immunoassay across the entire physiologic spectrum of aldosterone production; this effect was especially pronounced – up to 50-65% lower – when aldosterone levels were less than 20 ng/dL, meaning that clinicians should be careful to verify the assay used and the clinical context before excluding the diagnosis of primary aldosteronism. Our findings highlight the risk for false negative interpretations with increasingly common LC-MS/MS assays, thereby underscoring the urgent need to raise awareness among clinicians that interpretations of circulating aldosterone and aldosterone-to-renin ratio values need to be re-calibrated to include much lower values than previously recommended. Presentation: Sunday, June 12, 2022 12:30 p.m. - 2:30 p.m.

USG guided FNAC of thyroid improving diagnostic accuracy

Introduction: USG guided FNAC of thyroid particularly facilitates accurate FNAC of non-palpable small lesions. It has the major advantage of real time monitoring which help in accurate localisation of needle tip during aspiration. It is very useful in the non diagnostic or unsatisfactory cases (according to Bethesda System Bethesda category I). It has been suggested to reduce the rate of non-diagnostic material and subsequently reduce the false-negative interpretation. Aims: 1) To evaluate the Diagnostic accuracy of USG-guided FNAC of Thyroid. 2) To evaluate cellular yield of aspirated material in USG-guided FNAC. Materials and Methods: The present study included 100 patients with thyroid swelling undergoing USG guided FNAC who previously were diagnosed by FNAC as Non diagnostic or Unsatisfactory (According to Bethesda System Bethesda category I). USG guided FNAC was done & Smears were made. Then microscopy was done following the Bethesda system (TBS). Result: Total 100 patients with thyroid swelling undergoing USG guided FNAC who previously were diagnosed by FNAC as Bethesda category I. In present study, Out of 100 patients, maximum no. of patients were in 41 to 50 years age groups (54 cases) and majority were Female (84 Cases). The revised diagnosis in these cases according to TBS were as follow, Nodular goiter (29%), Colloid goiter (26%), lymphocytic thyroiditis (14%), Hashimoto thyroiditis (9%), Graves’ disease (2%) & granulomatous thyroiditis (1%). Suspicious for malignancy (4%) and papillary carcinoma (4%) and Follicular neoplasm (3%). There was increased cellular yield of aspirated material in 92% of cases with USG guidance. In present study false negative & True Positive rates were–92% & 8% for Conventional FNAC technique, as concluded after repeat FNAC under USG guidance.

Should Primary Stroke Centers Perform Advanced Imaging?

Should Primary Stroke Centers Perform Advanced Imaging?

Artificial intelligence system for identification of false-negative interpretations in chest radiographs.

To investigate the efficacy of an artificial intelligence (AI) system for the identification of false negatives in chest radiographs that were interpreted as normal by radiologists. We consecutively collected chest radiographs that were read as normal during 1 month (March 2020) in a single institution. A commercialized AI system was retrospectively applied to these radiographs. Radiographs with abnormal AI results were then re-interpreted by the radiologist who initially read the radiograph ("AI as the advisor" scenario). The reference standards for the true presence of relevant abnormalities in radiographs were defined by majority voting of three thoracic radiologists. The efficacy of the AI system was evaluated by detection yield (proportion of true-positive identification among the entire examination) and false-referral rate (FRR, proportion of false-positive identification among all examinations). Decision curve analyses were performed to evaluate the net benefits of applying the AI system. A total of 4208 radiographs from 3778 patients (M:F = 1542:2236; median age, 56 years) were included. The AI system identified initially overlooked relevant abnormalities with a detection yield and an FRR of 2.4% and 14.0%, respectively. In the "AI as the advisor" scenario, radiologists detected initially overlooked relevant abnormalities with a detection yield and FRR of 1.2% and 0.97%, respectively. In a decision curve analysis, AI as an advisor scenario exhibited a positive net benefit when the cost-to-benefit ratio was below 1:0.8. An AI system could identify relevant abnormalities overlooked by radiologists and could enable radiologists to correct their false-negative interpretations by providing feedback to radiologists. • In consecutive chest radiographs with normal interpretations, an artificial intelligence system could identify relevant abnormalities that were initially overlooked by radiologists. • The artificial intelligence system could enable radiologists to correct their initial false-negative interpretations by providing feedback to radiologists when overlooked abnormalities were present.

Ratiometric Electrochemical Biosensor Based on Internally Controlled Duplex PCR for Detection of Mycobacterium Tuberculosis

The pathogenic bacteria Mycobacterium tuberculosis (MTB) is responsible for tuberculosis which is well known for the globally leading cause of death. The likelihood of false negative interpretation as well as potential influence from intrinsic and extrinsic factors are considerably minimized by the incorporation of internal control (IC) detection into the developed assay platform. Ratiometric electrochemical biosensor (REC biosensor) for detection of MTB was developed based on the IC integration via duplex PCR (dPCR) and a dual-signal electrochemical readout. The MTB- or IC-specific PNA probe was labeled with methylene blue (MB) or ferrocene (FC), respectively at the C terminus, producing a strong square wave voltammetry (SWV) signal. Interaction of the ICdPCR product could induce changes in the dynamics of these two redox-labeled PNA probes (MTB-MB and IC-FC) that were attached to the screen-printed gold electrode (SPGE) via formation of a self-assembled monolayer. Using this MB as a reporter and FC as an IC, the REC ICdPCR biosensor achieved a broad detection range from 10 fM to 10 nM and a detection limit of 1.26 fM, corresponding to approximately 2.5 bacteria cells. The REC ICdPCR biosensor was applied to MTB measurement in practical samples, exhibiting high accuracy and more importantly high practicability.

Diagnostic Accuracy of 18F‐FDG‐PET/CT and 18F‐FDG‐PET/MRI in Detecting Locoregional Recurrence of HNSCC 12 Weeks after the End of Chemoradiotherapy: Single‐Center Experience with PET/MRI

Purpose In head and neck squamous cell carcinoma (HNSCC), the early diagnosis and efficient detection of recurrences and/or residual tumor after treatment play a very important role in patient's prognosis. Positron emission tomography (PET) using 2-deoxy-2-18F-fluoro-D-glucose (18F-FDG) has become an established method for the diagnosis of suspected recurrence in head and neck carcinomas. In particular, integrated PET/MRI imaging that provides optimal soft tissue contrast and less dental implant artifacts compared to PET/CT is an intriguing technique for the follow-up imaging of HNSCC patients. The aim of this study was to evaluate the benefit of PET/MRI compared to PET/CT in post-treatment follow-up imaging of HNSCC patients. Methods This retrospective observational cohort study consists of 104 patients from our center with histologically confirmed HNSCC. All patients received chemoradiotherapy (CRT) and underwent 18F-FDG-PET/CT (n = 52) or 18F-FDG-PET/MRI (n = 52) scan 12 weeks after the end of treatment. Image analysis was performed by two independent readers according to a five-point Likert scale analysis. Results PET/MRI was more sensitive (1.00 vs. 0.77) than PET/CT in the detection of locoregional recurrence. PET/MRI also had better negative (1.00 vs. 0.87) predictive values. AUCs for PET/MRI and PET/CT on patient-based analysis were 0.997 (95% CI 0.989–1.000) and 0.890 (95% CI 0.806–0.974), respectively. The comparison of sensitivity, AUCs, and negative predictive values revealed a statistically significant difference, p < 0.05. In PET/CT, false-negative and positive findings were observed in the more advanced disease stages, where PET/MRI performed better. Also, false-negative findings were located in the oropharyngeal, laryngeal, and nasopharyngeal regions, where PET/MRI made no false-negative interpretations. Conclusion Based on these results, PET/MRI might be considered the modality of choice in detecting locoregional recurrence in HNSCC patients, especially in the more advanced stages in the oral cavity, larynx, or nasopharynx.

A New Method for Solid Acid Catalyst Evaluation for Cellulose Hydrolysis

A systematic and structure-agnostic method for identifying heterogeneous activity of solid acids for catalyzing cellulose hydrolysis is presented. The basis of the method is preparation of a supernatant liquid by exposing the solid acid to reaction conditions and subsequent use of the supernatant liquid as a cellulose hydrolysis catalyst to determine the effects of in situ generated homogeneous acid species. The method was applied to representative solid acid catalysts, including polymer-based, carbonaceous, inorganic, and bifunctional materials. In all cases, supernatant liquids produced from these catalysts exhibited catalytic activity for cellulose hydrolysis. Direct comparison of the activity of the solid acid catalysts and their supernatants could not provide unambiguous detection of heterogeneous catalysis. A reaction pathway kinetic model was used to evaluate potential false-negative interpretation of the supernatant liquid test and to differentiate heterogeneous from homogeneous effects on cellulose hydrolysis. Lastly, differences in the supernatant liquids obtained in the presence and absence of cellulose were evaluated to understand possibility of false-positive interpretation, using structural evidence from the used catalysts to gain a fresh understanding of reactant–catalyst interactions. While many solid acid catalysts have been proposed for cellulose hydrolysis, to our knowledge, this is the first effort to attempt to differentiate the effects of heterogeneous and homogeneous activities. The resulting supernatant liquid method should be used in all future attempts to design and develop solid acids for cellulose hydrolysis.

Automated detection of aggressive and indolent prostate cancer on magnetic resonance imaging.

PurposeWhile multi‐parametric magnetic resonance imaging (MRI) shows great promise in assisting with prostate cancer diagnosis and localization, subtle differences in appearance between cancer and normal tissue lead to many false positive and false negative interpretations by radiologists. We sought to automatically detect aggressive cancer (Gleason pattern ≥ 4) and indolent cancer (Gleason pattern 3) on a per‐pixel basis on MRI to facilitate the targeting of aggressive cancer during biopsy.MethodsWe created the Stanford Prostate Cancer Network (SPCNet), a convolutional neural network model, trained to distinguish between aggressive cancer, indolent cancer, and normal tissue on MRI. Ground truth cancer labels were obtained by registering MRI with whole‐mount digital histopathology images from patients who underwent radical prostatectomy. Before registration, these histopathology images were automatically annotated to show Gleason patterns on a per‐pixel basis. The model was trained on data from 78 patients who underwent radical prostatectomy and 24 patients without prostate cancer. The model was evaluated on a pixel and lesion level in 322 patients, including six patients with normal MRI and no cancer, 23 patients who underwent radical prostatectomy, and 293 patients who underwent biopsy. Moreover, we assessed the ability of our model to detect clinically significant cancer (lesions with an aggressive component) and compared it to the performance of radiologists.ResultsOur model detected clinically significant lesions with an area under the receiver operator characteristics curve of 0.75 for radical prostatectomy patients and 0.80 for biopsy patients. Moreover, the model detected up to 18% of lesions missed by radiologists, and overall had a sensitivity and specificity that approached that of radiologists in detecting clinically significant cancer.ConclusionsOur SPCNet model accurately detected aggressive prostate cancer. Its performance approached that of radiologists, and it helped identify lesions otherwise missed by radiologists. Our model has the potential to assist physicians in specifically targeting the aggressive component of prostate cancers during biopsy or focal treatment.

An analysis of radiographer preliminary image evaluation - A focus on common false negatives.

IntroductionPreliminary image evaluation (PIE) is a mechanism whereby radiographers provide a preliminary evaluation of whether pathology is present in their radiographs, typically acquired within the emergency department (ED). PIE provides referrers with a timely communication of pathology prior to the availability of a radiology report. The purpose of this study was to determine the most common radiographer PIE false‐negative interpretations.MethodsEach month over a two‐year period, 100 PIEs of adult and paediatric patients were randomly reviewed in a metropolitan hospital ED. The radiographer’s PIE was compared with the radiologist’s report and categorised into basic quality indicators; true positive, true negative, false positive and false negative. The anatomical regions which most commonly indicated a false‐negative interpretation were further analysed.Results2402 cases were reviewed which resulted in an overall PIE accuracy of 88.7%. Wrists, hands, phalanges (upper), ankles, feet and phalanges (lower) reporting the highest false‐negative or false‐negative/true‐positive interpretations (60/116). Of the 60 false‐negative PIEs, 68 pathologies were identified. 41.1% (28/68) of the pathology not identified were in the phalanges. Within these regions, examinations with multiple injuries commonly reported false negatives (17/60).ConclusionsThis study demonstrated the most common false‐negative radiographer PIEs were within the upper and lower distal extremities. Specifically, the phalanges and examinations demonstrating multiple injuries reported high levels of misinterpretation. The misinterpretation in multi‐injury examinations could be attributed to ‘Subsequent Search Miss (SSM)’ error. These results provide valuable insights into areas of emphasis when providing image interpretation education.

Fusion of high b-value diffusion-weighted and unenhanced T1-weighted images to diagnose invasive breast cancer: factors associated with false-negative results.

We sought factors associated with false-negative results in the diagnosis of invasive breast cancer via non-contrast breast magnetic resonance imaging (MRI) using fused high b-value diffusion-weighted imaging (DWI) and unenhanced T1-weighted images (T1WI). Between 2018 and 2019, 316 consecutive women (mean age, 54.6 years) with invasive breast cancer who underwent preoperative breast MRI, including fused high b-value DWI and unenhanced T1WI, were retrospectively evaluated. Malignancy confidence ratings of the most suspicious breast lesions evident on fused DWI were derived by two radiologists using a 6-point Likert-type scale. Both clinicopathological and imaging features were analyzed. Multivariate regression analysis was performed to identify factors associated with false-negative DWI results in the diagnosis of invasive breast cancer. Of the 316 breast cancers, fused DWI yielded 289 (91.5%) true-positive and 27 (8.5%) false-negative results. Multivariate analysis showed that small tumor size (≤ 1 cm) (odds ratio [OR], 5.95; 95% confidence interval [CI], 2.11, 16.81; p = 0.001), presence of calcifications in the tumor (OR, 3.41; 95% CI, 1.27, 9.15; p = 0.015), and a moderate/marked background diffusion signal (ORs, 4.23 and 19.18; 95% CI, 1.31, 13.67 and 6.51, 56.46; p = 0.016 and p < 0.001, respectively) were significantly associated with false-negative results. In subgroup analysis of 141 screening-detected cancers, a marked background diffusion signal (OR, 7.94; 95% CI, 2.30, 27.35; p = 0.001) remained significantly associated with false-negative results in the multivariate analysis. In addition to histopathological features, a higher background diffusion signal was associated with false-negative results in the diagnosis of invasive breast cancer via non-contrast MRI using fused high b-value DWI and unenhanced T1WI. • Subcentimeter tumors and presence of calcifications in the tumor are associated with false-negative diffusion-weighted imaging results in the diagnosis of invasive breast cancer. • A higher degree of background diffusion signal may lead to false-negative interpretation of diffusion-weighted imaging in patients with invasive breast cancer.

Pearls and Pitfalls in the Magnetic Resonance Diagnosis of Dural Sinus Thrombosis: A Comprehensive Guide for the Trainee Radiologist.

Dural sinus thrombosis (DST) is a potentially fatal neurological condition that can be reversed with early diagnosis and prompt treatment. Non-enhanced CT scan is often the first imaging investigation in patients presenting with acute neurological symptoms; however, its poor sensitivity in detecting DST is a major drawback. Magnetic resonance (MR) imaging offers multiple advantages such as excellent contrast resolution and unenhanced venography possibilities, making it the mainstay in the non-invasive diagnosis of DST. However, physiological variations, evolution of thrombi, and incorrect selection/application of MR techniques can lead to false positive and false negative interpretations impacting patient management and outcome. This article discusses the MR techniques useful to diagnose DST and describes pitfalls, with troubleshooting methods, to ensure an accurate diagnosis. We have used multiple diagrammatic illustrations and MR images to highlight pertinent take-home points and to serve as an easy guide for day-to-day clinical practice.

Impact of Artifacts From Optical Coherence Tomography Retinal Nerve Fiber Layer and Macula Scans on Detection of Glaucoma Progression

To determine the prevalence of artifacts on segmented spectral-domain optical coherence tomography (SDOCT) images and assess their impact on the interpretation of glaucomatous progression in the retinal nerve fiber layer (RNFL) profile and macular thickness map. Retrospective reliability analysis. Retrospective review of glaucoma and glaucoma suspect eyes imaged with SDOCT during a 1-month period. All cases had at least 4 sets of RNFL and macular images at 6-month intervals. SDOCT raw B-scans were examined to determine true progression and whether artifacts impacted the original interpretation of progression based on auto-segmented change maps. The co-prevalence of artifacts in the RNFL and macula was assessed, as well as the association of clinical factors with the likelihood of artifacts. A total of 190 eyes with 760 sets of OCT RNFL and macular scans were included. Fifty percent (96/190) of eyes had artifacts, either in the circumpapillary RNFL (83/190; 43.68%) or the macula (57/190; 30.0%). Epiretinal membrane and vitreomacular traction were the most common artifacts. True progression was present on 39.5% (75/190) of scans overall. Among scans with artifacts, 23.9% (23/96) of artifacts masked true progression (ie, false-negative), 36.5% (35/96) led to an interpretation of false progression (ie, false-positive), and 39.6% (38/96) had no effect on the interpretation of progression. The presence of true progression on the RNFL scan was significantly associated with the presence of true progression on the macular scan (P < .001). Similarly, the presence of artifacts on the RNFL scan was significantly associated with artifacts on the macular scan (P < .001). In multivariable analysis, severe glaucoma, hypertension, and age were significantly associated with the presence of artifacts on RNFL (P< .05). Artifacts are highly prevalent on both circumpapillary RNFL and macular scans on SDOCT images acquired in a glaucoma clinic. Artifacts can lead to false-positive and false-negative interpretation of progression when using only the auto-segmentation change maps. Thus, careful examination of the raw B-scan images of both the RNFL and macula is critical to identify artifacts and true glaucoma progression.

Electronic health records to facilitate continuous detection of familial hypercholesterolemia

Background and aimsFamilial hypercholesterolemia (FH) is an inherited disorder associated with increased risk of coronary heart disease as a result of high LDL-cholesterol (LDL-C). The clinical diagnosis can be made with the Dutch Lipid Clinic Network criteria (DLCN criteria). FH is an underdiagnosed disorder, possibly due to false negative LDL-C interpretation during lipid lowering therapy (LLT). We hypothesized that automated health record-based integration of data can provide a signal to facilitate identification of FH patients. MethodsWe included patients with LDL-C ≥6.5 mmol/l after correction for LLT in all patients testing LDL-C in Northwest Clinics, The Netherlands. Patients previously diagnosed with FH were excluded. The primary endpoint was the additional number of patients with DLCN criteria ≥6 points after correction for LLT. Secondary endpoints were the additional number of patients with DLCN criteria ≥6 points after also adding data on patient- and family history, and LDL-C before and after correction for LLT. Analysis was performed in a daily automated routine (HiX ChipSoft). ResultsIn a total of 41,937 individual LDL-C measurements during 26 weeks, we found 351 patients with LDL-C ≥6.5 mmol/l after automated correction for LLT. FH had previously been diagnosed in 42 patients. In the remaining 309 patients (58.3% female; age: 66 ± 11 yrs (mean ± SD); 85.8% on LLT), the number of patients with DLCN criteria ≥6 points increased from 9 to 95 after correction for LLT, and to 127 after also adding patient and family history. The mean LDL-C before and after correction for LLT was 4.69 ± 1.42 mmol/l and 8.16 ± 1.68 mmol/l, respectively (mean ± SD; p < 0.001). ConclusionsWe conclude that automated medical record-based integration of LDL-C, LLT and patient- and family history can provide a crucial signal to facilitate identification of FH. Whether this signal results in subsequent genetic identification of FH patients and their relatives requires further study.

Free light chain kappa and the polyspecific immune response in MS and CIS – Application of the hyperbolic reference range for most reliable data interpretation

ObjectivesFree light chain kappa (FLC-k) in cerebrospinal fluid (CSF) is involved in intrathecal immune responses and is being investigated frequently for its diagnostic sensitivity. The objective of this study was the application and interpretation of FLC-k data in quotient diagrams with a hyperbolic reference range and to confirm the superior evaluation in comparison with another proposed reference method and cut-off values. Secondly, the performance of the FLC-k quotient diagram was analyzed in respect to MS and CIS patients and in relation to the polyspecific immune response. Materials and methodsFLC-k was analyzed in a control cohort (n = 302) and in patients with MS/CIS (n = 98) using a nephelometric FLC-k kit. The intrathecal fraction of FLC-k based on the hyperbolic reference range was calculated in comparison to various linear FLC-k indices and routine CSF parameters [oligoclonal bands (OCB), polyspecific antiviral immune response]. ResultsUsing the new hyperbolic reference range, intrathecal FLC-k synthesis was found in 20 / 302 OCB negative controls. The sensitivity in the definitive MS cohort was 100%, compared to 93% positive OCB. The linear FLC-k Index interpretation with similar sensitivity for MS, however, bares the risk for the control samples,depending on the reference range, of false positive interpretations (up to 7 at low QAlb) or false negative interpretations (up to 17/20 FLC-k positives at high QAlb). The quantitative mean intrathecal FLC-k synthesis in the CIS cohort (later MS) was even slightly higher than in initially definitive MS questioning a pathophysiological difference. A positive MRZ reaction found in 53% percent of CIS patients with intrathecal FLC-k synthesis could have allowed diagnosis of MS immediately, i.e. earlier than with the Mc Donald criteria. ConclusionsThe evaluation of FLC-k with hyperbolic reference range in quotient diagrams is superior to other analytical methods like the linear FLC-k index. We suggest a sequential CSF testing with FLC-k Reibergram evaluation, potentially followed by isoelectric focusing. With the MRZ reaction we obtain highest specificity for MS diagnosis.

Abstract B13: Applying machine learning for urine cytology—computational urothelial carcinoma analysis and diagnosis

Abstract Urothelial carcinoma is estimated to be the fourth most common cancer among men in the U.S. in 2019. The key to successful management of this cancer is early detection and timely monitoring after treatment. Urine cytology is an inexpensive, noninvasive, and easily available diagnostic method of urine analysis, but the usage of this tool for urothelial malignancy is still not satisfactory in its sensitivity and specificity. It is time-consuming and labor-intensive to find rare neoplastic cells from hundreds or even thousands of a variety of types of cells in the background, even by well-trained cytopathologists or cytotechnicians. Therefore, it is a crucial issue to avoid false-positive and false-negative results when applying urine cytology in clinical practice. Recent advances in artificial intelligence (AI) offer potential solutions to these issues through the development of an automated, impartial, and specialized computation model to 1) reduce false-negative rates and interpretation times and 2) assist pathologists with reporting using The Paris System for Reporting Urinary Cytopathology. In this study, we tested a deep learning-based image analysis model for cell classification and enumeration in urine cytology. Deidentified whole-slide images were digitalized after performing standard urine cytology examination. We used an “active learning” approach that requires minimal upfront training and improves with more usage. We collected 3 sub-images (total 3,335 cells) from one whole-slide image annotated by 3 domain experts to train the initial model. Cells were classified into seven categories: high-grade urothelial carcinoma (HGUC), cluster HGUC, atypical neoplastic cell, atypical reactive cell, inflammatory cell, epithelial cell, and unidentified cell. Domain experts reviewed and provided feedback to initiate the model training process. After the initial model was established, we performed a pilot study using 2 sub-images from five digitalized sample slides (total 10 sub-images) to further improve the training model. Our results showed that the AI model was able to learn the morphologies of six cell types. It was then able to automatically quantify total cell counts in each class that correlated with annotations by domain experts. We are now attempting to integrate this AI model with The Paris System to create a clinical decision support system for urologists, as there is a clear unmet clinical need to establish a computational model since The Paris System requires a significant amount of complex and accurate analysis, which many clinical decisions are based upon. In conclusion, our AI model demonstrates its value in providing an automated and unbiased method to analyze urine cytology with high throughput, accuracy, and reproducibility. Introduction and integration of our AI model with The Paris System will be a giant stride to optimize clinical practices in the field of urology in the imminent future. Citation Format: Wei-Lei Yang, Ching-Ming Lee, Mei-Ling Wu, Yu-Ching Peng, Ten-Jen Liu, Yi-Hsin Liu. Applying machine learning for urine cytology—computational urothelial carcinoma analysis and diagnosis [abstract]. In: Proceedings of the AACR Special Conference on Advances in Liquid Biopsies; Jan 13-16, 2020; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(11_Suppl):Abstract nr B13.

Frozen section diagnosis of gastrointestinal poorly cohesive and signet-ring cell adenocarcinoma: useful morphologic features to avoid misdiagnosis.

Frozen section examination of adenocarcinomas with poorly cohesive growth, including signet-ring cell carcinoma, is challenging. Due to their diffuse morphology, the tumor cells may be indistinct and difficult to distinguish from inflammatory or stromal cells. Misdiagnosis may result in significant adverse clinical outcome. We performed a detailed retrospective analysis of such cases to identify features that are helpful to avoid misdiagnosis at the time of frozen section. We reviewed the original frozen section slides from 50 patients with poorly cohesive carcinoma (PCC) including 32 with positive and 18 with negative frozen section slides. Tumor cells and inflammatory cells were evaluated for 17 distinct cytologic and nine architectural or stromal features. Features with 100% specificity and positive predictive value (PPV) for carcinoma included the presence of cells with a single distinct cytoplasmic mucin vacuole, focal gland formation, and perineural invasion. Features with high specificity, sensitivity, PPV, and negative predictive value (NPV) (all > 75%) included irregular nuclear contours, large nuclear size with many nuclei > 4× the size of a small lymphocyte, and disruption/obliteration of normal structures. Other features with high specificity and PPV (both ≥ 85%) but relatively low sensitivity and NPV-included crescent-shaped/indented nuclei, prominent nucleoli, anisonucleosis (> 4:1 difference in nuclear size), multinucleation, and the presence of mitotic figures. We characterized useful histologic features of poorly cohesive carcinoma that may serve to distinguish carcinoma cells from benign inflammatory or stroma cells. Knowledge of the relatively specific features in particular may help surgical pathologists avoid false-negative interpretation resulting in significant clinical morbidity.