Sources of performance variability in deep learning-based polyp detection.

PURPOSE: Validation metrics are a key prerequisite for the reliable tracking of scientific progress and for deciding on the potential clinical translation of methods. While recent initiatives aim to develop comprehensive theoretical frameworks for understanding metric-related pitfalls in image analysis problems, there is a lack of experimental evidence on the concrete effects of common and rare pitfalls on specific applications. We address this gap in the literature in the context of colon cancer screening. METHODS: Our contribution is twofold. Firstly, we present the winning solution of the Endoscopy Computer Vision Challenge on colon cancer detection, conducted in conjunction with the IEEE International Symposium on Biomedical Imaging 2022. Secondly, we demonstrate the sensitivity of commonly used metrics to a range of hyperparameters as well as the consequences of poor metric choices. RESULTS: Based on comprehensive validation studies performed with patient data from six clinical centers, we found all commonly applied object detection metrics to be subject to high inter-center variability. Furthermore, our results clearly demonstrate that the adaptation of standard hyperparameters used in the computer vision community does not generally lead to the clinically most plausible results. Finally, we present localization criteria that correspond well to clinical relevance. CONCLUSION: We conclude from our study that (1) performance results in polyp detection are highly sensitive to various design choices, (2) common metric configurations do not reflect the clinical need and rely on suboptimal hyperparameters and (3) comparison of performance across datasets can be largely misleading. Our work could be a first step towards reconsidering common validation strategies in deep learning-based colonoscopy and beyond.

Microenvironmental changes co-occur with mosaic somatic clonal expansions in normal skin and esophagus tissues.

The presence of somatic mutations, previously identified in cancers, are being increasingly recognized in normal tissues. While the role of microenvironment (ME) in tumor progression is well understood, the changes that occur in the microenvironment of normal tissues that harbor somatic mutations has not been systematically studied. Here, using normal RNA-Seq data accrued from 6544 samples across 27 tissue types from Genotype-Tissue Expression (GTEx) project, we studied the association of microenvironmental changes in the presence of somatic clonal expansions of previously implicated cancer genes. We focused our analysis on skin and esophagus since they have the highest number of samples and mutation burden together. We observed changes in microenvironmental cell-types previously implicated in tumor progression including endothelial cells, epithelial cells, pericytes, fibroblasts, chondrocytes, among others. The Epithelial-Mesenchymal-Transition (EMT) pathway is dysregulated in both skin and esophagus, along with increased hypoxia scores in samples with clonal expansions. These results suggest that microenvironmental changes play an important role in clonal expansions and potentially the initiating stages of cancer progression. Studying these changes may provide new avenues for early intervention of cancer, for targeted therapies, or enhance activities of conventional therapies.

External validation of models to predict the outcome of pregnancies of unknown location: a multicentre cohort study.

OBJECTIVE: To validate externally five approaches to predict ectopic pregnancy (EP) in pregnancies of unknown location (PUL): the M6P and M6NP risk models, the two-step triage strategy (2ST, which incorporates M6P), the M4 risk model, and beta human chorionic gonadotropin ratio cut-offs (BhCG-RC). DESIGN: Secondary analysis of a prospective cohort study. SETTING: Eight UK early pregnancy assessment units. POPULATION: Women presenting with a PUL and BhCG >25 IU/l. METHODS: Women were managed using the 2ST protocol: PUL were classified as low risk of EP if presenting progesterone ≤2 nmol/l; the remaining cases returned 2 days later for triage based on M6P. EP risk ≥5% was used to classify PUL as high risk. Missing values were imputed, and predictions for the five approaches were calculated post hoc. We meta-analysed centre-specific results. MAIN OUTCOME MEASURES: Discrimination, calibration and clinical utility (decision curve analysis) for predicting EP. RESULTS: Of 2899 eligible women, the primary analysis excluded 297 (10%) women who were lost to follow up. The area under the ROC curve for EP was 0.89 (95% CI 0.86-0.91) for M6P, 0.88 (0.86-0.90) for 2ST, 0.86 (0.83-0.88) for M6NP and 0.82 (0.78-0.85) for M4. Sensitivities for EP were 96% (M6P), 94% (2ST), 92% (N6NP), 80% (M4) and 58% (BhCG-RC); false-positive rates were 35%, 33%, 39%, 24% and 13%. M6P and 2ST had the best clinical utility and good overall calibration, with modest variability between centres. CONCLUSIONS: 2ST and M6P performed best for prediction and triage in PUL. TWEETABLE ABSTRACT: The M6 model, as part of a two-step triage strategy, is the best approach to characterise and triage PULs.

Triaging women with pregnancy of unknown location using two-step protocol including M6 model: clinical implementation study.

OBJECTIVE: The M6 risk-prediction model was published as part of a two-step protocol using an initial progesterone level of ≤ 2 nmol/L to identify probable failing pregnancies (Step 1) followed by the M6 model (Step 2). The M6 model has been shown to have good triage performance for stratifying women with a pregnancy of unknown location (PUL) as being at low or high risk of harboring an ectopic pregnancy (EP). This study validated the triage performance of the two-step protocol in clinical practice by evaluating the number of protocol-related adverse events and how effectively patients were triaged. METHODS: This was a prospective multicenter interventional study of 3272 women with a PUL, carried out between January 2015 and January 2017 in four district general hospitals and four university teaching hospitals in the UK. The final pregnancy outcome was defined as: a failed PUL (FPUL), an intrauterine pregnancy (IUP) or an EP (including persistent PUL (PPUL)). FPUL and IUP were grouped as low-risk and EP/PPUL as high-risk PUL. Serum progesterone and human chorionic gonadotropin (hCG) levels were measured at presentation in all patients. If the initial progesterone level was ≤ 2 nmol/L, patients were discharged and were asked to have a follow-up urine pregnancy test in 2 weeks to confirm a negative result. If the progesterone level was > 2 nmol/L or a measurement had not been taken, hCG level was measured again at 48 h and results were entered into the M6 model. Patients were managed according to the outcome predicted by the protocol. Those classified as 'low risk, probable FPUL' were advised to perform a urine pregnancy test in 2 weeks and those classified as 'low risk, probable IUP' were invited for a scan a week later. When a woman with a PUL was classified as high risk (i.e. risk of EP ≥ 5%) she was reviewed clinically within 48 h. One center used a progesterone cut-off of ≤ 10 nmol/L and its data were analyzed separately. If the recommended management protocol was not adhered to, this was recorded as a protocol deviation and classified as: unscheduled visit for clinician reason, unscheduled visit for patient reason or incorrect timing of blood test or ultrasound scan. The classifications outlined in the UK Good Clinical Practice (GCP) guidelines were used to evaluate the incidence of adverse events. Data were analyzed using descriptive statistics. RESULTS: Of the 3272 women with a PUL, 2625 were included in the final analysis (317 met the exclusion criteria or were lost to follow-up, while 330 were evaluated using a progesterone cut-off of ≤ 10 nmol/L). Initial progesterone results were available for 2392 (91.1%) patients. In Step 1, 407 (15.5%) patients were classified as low risk (progesterone ≤ 2 nmol/L), of whom seven (1.7%) were ultimately diagnosed with an EP. In 279 of the remaining 2218 women with a PUL, the M6 model was not applied owing to protocol deviation or because the outcome was already known (usually on the basis of an ultrasound scan) before a second hCG reading was taken; of these patients, 30 were diagnosed with an EP. In Step 2, 1038 women with a PUL were classified as low risk, of whom eight (0.8%) had a final outcome of EP. Of 901 women classified as high risk at Step 2, 275 (30.5%) had an EP. Therefore, 275/320 (85.9%) EPs were correctly classified as high risk. Overall, 1445/2625 PUL (55.0%) were classified as low risk, of which 15 (1.0%) were EP. None of these cases resulted in a ruptured EP or significant clinical harm. Sixty-two women participating in the study had an adverse event, but no woman had a serious adverse event as defined in the UK GCP guidelines. CONCLUSIONS: This study has shown that the two-step protocol incorporating the M6 model effectively triaged the majority of women with a PUL as being at low risk of an EP, minimizing the follow-up required for these patients after just two visits. There were few misclassified EPs and none of these women came to significant clinical harm or suffered a serious adverse clinical event. The two-step protocol incorporating the M6 model is an effective and clinically safe way of rationalizing the management of women with a PUL. Copyright © 2019 ISUOG. Published by John Wiley & Sons Ltd.

Assessing a single SNP located at TERT/CLPTM1L multi-cancer risk region as a genetic modifier for risk of pancreatic cancer and melanoma in Dutch CDKN2A mutation carriers.

Carriers of pathogenic variants in CDKN2A have a 70% life-time risk of developing melanoma and 15-20% risk of developing pancreatic cancer (PC). In the Netherlands, a 19-bp deletion in exon 2 of CDKN2A (p16-Leiden mutation) accounts for most hereditary melanoma cases. Clinical experience suggests variability in occurrence of melanoma and PC in p16-Leiden families. Thereby, the risk of developing cancer could be modified by both environmental and genetic contributors, suggesting that identification of genetic modifiers could improve patients' surveillance. In a recent genome-wide association study (GWAS), rs36115365-C was found to significantly modify risk of PC and melanoma in the European population. This SNP is located on chr5p15.33 and has allele-specific regulatory activities on TERT expression. Herein, we investigated the modifying capacities of rs36115365-C on PC and melanoma in a cohort of 283 p16-Leiden carriers including 29 diagnosed with PC, 171 diagnosed with melanoma, 21 diagnosed with both PC and melanoma and 62 with neither PC nor melanoma. In contrast to previously reported findings, we did not find a significant association of PC risk with risk variant presence as determined by Generalized Estimating Equations (GEE) modelling. Interestingly, carrier-ship of the risk variant had a significant protective effect for melanoma (OR - 0.703 [95% CI - 1.201 to - 0.205], p = 0.006); however, the observed association was no longer significant after exclusion of probands to assess possible influence of ascertainment. Collectively, genetic modifiers for the prediction of PC and melanoma risk in p16-Leiden carriers remain to be determined.

Saffron (Crocus sativus) intake provides nutritional preconditioning against myocardial ischemia-reperfusion injury in Wild Type and ApoE(-/-) mice: Involvement of Nrf2 activation.

BACKGROUND AND AIMS: Saffron is an antioxidant herbal derivative; however, its efficacy as a nutritional cardioprotective agent has not been fully elucidated. We investigated the cardioprotective properties of a standardized saffron aqueous extract (SFE) against ischemia/reperfusion (I/R) injury in Wild-Type (WT) and ApoE(-/-) mice and the underlying molecular mechanisms. METHODS AND RESULTS: WT and ApoE(-/-) mice were subjected to 30 min I and 2 h R, with the following per os interventions for 4 weeks: 1) WT Control Group, receiving Water for Injection (WFI); 2) WT Crocus Group, receiving SFE at a dose of 60 mg/kg/day; 3) WT Crocus + Wort group, receiving SFE as described above and wortmannin at a dose of 60 µg/kg bolus 15 min before R; 4) ApoE(-/-) Control Group, receiving WFI; 5) ApoE(-/-) Crocus Group, receiving SFE at a dose of 60 mg/kg/day and 6) ApoE(-/-) Crocus + Wort: receiving SFE as described above and wortmannin at a dose of 60 µg/kg bolus, 15 min before R. Ischemic area/area at risk (I/R%) ratio was measured. Blood samples and ischemic myocardial tissue were collected at the 10th min of reperfusion for assessment of troponin I, malondialdehyde (MDA), nitrotyrosine (NT), p-eNOS, eNOS, p-Akt, Akt, p-p42/p-p44, p-GSK3ß, GSK3ß, IL-6, Nrf2, HO-1 and MnSOD expression. The effect of SFE on Nrf2 expression was also evaluated in vitro. SFE reduced infarct size in WT (16.15 ± 3.7% vs 41.57 ± 2.48%, ***p < 0.001) and in ApoE(-/-) mice (16.14 ± 1.47% vs 45.57 ± 1.73%, ***p < 0.001). The administration of wortmannin resulted in partial inhibition of the infarct size limitation efficacy of SFE (in both WT and Apo-E(-/-) mice). Mice receiving SFE showed increased levels of eNOS, p-Akt, p-ERK1/2, p-44/p-42 and p-GSK3ß-Ser9 and reduced expression of IL-6 and iNOS; furthermore, SFE reduced the levels of MDA and NT. SFE induced Nrf2 expression and its downstream targets, HO-1 and MnSOD in the myocardium of the treated animals, and induced Nrf2 expression in vitro in a dose-dependent manner. CONCLUSIONS: SFE limits myocardial infarction in Wild-Type and ApoE(-/-) mice in a multifaceted manner including activation of Akt/eNOS/ERK1/2/GSK3-ß and through Nrf2 pathway, bestowing antioxidant protection against I/R.

Raise the bar and lower the dose: current and future strategies for radiation dose reduction in head and neck imaging.

SUMMARY: Technologic advances in CT have generated a dramatic increase in the number of CT studies, with a resultant increase in the radiation dose related to CT scanning. Such increase in radiation dose is becoming a concern for the radiology community, especially with increasing public awareness of the dose burden related to examinations. To cope with the increase in CT-related radiation exposure, it is becoming necessary to optimize CT imaging protocols and apply radiation dose reduction techniques to ensure the best imaging with the lowest radiation dose.

Quaternary structure of the human Cdt1-Geminin complex regulates DNA replication licensing.

All organisms need to ensure that no DNA segments are rereplicated in a single cell cycle. Eukaryotes achieve this through a process called origin licensing, which involves tight spatiotemporal control of the assembly of prereplicative complexes (pre-RCs) onto chromatin. Cdt1 is a key component and crucial regulator of pre-RC assembly. In higher eukaryotes, timely inhibition of Cdt1 by Geminin is essential to prevent DNA rereplication. Here, we address the mechanism of DNA licensing inhibition by Geminin, by combining X-ray crystallography, small-angle X-ray scattering, and functional studies in Xenopus and mammalian cells. Our findings show that the Cdt1:Geminin complex can exist in two distinct forms, a "permissive" heterotrimer and an "inhibitory" heterohexamer. Specific Cdt1 residues, buried in the heterohexamer, are important for licensing. We postulate that the transition between the heterotrimer and the heterohexamer represents a molecular switch between licensing-competent and licensing-defective states.

Implementation of semi-automated cloning and prokaryotic expression screening: the impact of SPINE.

The implementation of high-throughput (HTP) cloning and expression screening in Escherichia coli by 14 laboratories in the Structural Proteomics In Europe (SPINE) consortium is described. Cloning efficiencies of greater than 80% have been achieved for the three non-ligation-based cloning techniques used, namely Gateway, ligation-indendent cloning of PCR products (LIC-PCR) and In-Fusion, with LIC-PCR emerging as the most cost-effective. On average, two constructs have been made for each of the approximately 1700 protein targets selected by SPINE for protein production. Overall, HTP expression screening in E. coli has yielded 32% soluble constructs, with at least one for 70% of the targets. In addition to the implementation of HTP cloning and expression screening, the development of two novel technologies is described, namely library-based screening for soluble constructs and parallel small-scale high-density fermentation.

Virtual bronchoscopy versus thin section computed tomography in evaluation of moderate and low grade stenoses: receiver operating characteristic curve analysis.

PURPOSE: To evaluate the impact of virtual bronchoscopy, under proper threshold settings, on observer level of confidence in the assessment of bronchial abnormalities producing stenoses < or = 75% compared to interpretation of thin section computed tomography (CT) images. MATERIAL AND METHODS: Sixty-five patients with fiberoptic bronchoscopy positive for tracheobronchial abnormalities were evaluated in a blinded observer study using a commercially available virtual endoscopy software package. The findings of virtual endoscopy were compared with those of fiberoptic bronchoscopy using receiver operating characteristic curves (ROCs) and other statistical tools. RESULTS: A total of 102 lesions were identified by fiberoptic bronchoscopy, with 44 of these producing bronchial stenoses < or = 75%. Concerning the latter lesions, for virtual bronchoscopy the areas under the ROCs were 0.93 and 0.96 for the two observers, respectively, while for thin section CT the corresponding values were 0.86 and 0.88; the differences observed were statistically significant. Contrary to thin section CT, virtual bronchoscopy did not show statistically significant differences from fiberoptic bronchoscopy regarding estimation of degree of stenosis. CONCLUSION: Virtual bronchoscopy under proper threshold settings has a statistically significant impact on observer performance where moderate and low-grade bronchial stenoses are concerned and gives an estimate of the degree of stenosis more precisely than thin section CT.

Assessment of calibration methods for estimating bone mineral densities in trauma patients with quantitative CT: an anthropomorphic phantom study.

RATIONALE AND OBJECTIVES: Osteoporosis may contribute to the increased morbidity and mortality of elderly persons involved in motor vehicle accidents. Such patients commonly undergo whole-body computed tomographic (CT) studies that may be analyzed with quantitative CT. Various quantitative CT calibration techniques were investigated for use with patients who have suffered trauma, who are typically scanned on a backboard. MATERIALS AND METHODS: Lumbar simulator phantoms were used to simulate small and large patients. Vertebral spongiosa inserts with a wide range of bone and fat compositions were placed in the phantoms, and their bone mineral densities (BMDs) were measured by using calibration lines derived from the CT numbers of a calibration standard. Four calibration techniques were tested. In three the lumbar simulator and the calibration standard were scanned simultaneously, with the standard placed beneath the backboard (method 1), on top of the backboard adjacent to the lumbar simulator (method 2), or on top of the abdomen region of the lumbar simulator (method 3). The fourth technique employed a single calibration line derived from a separate scan of the calibration standard beneath the small lumbar simulator without the backboard, with correction for patient body size. RESULTS: The best overall results were obtained with the single calibration line method. The root mean square errors of the BMD values were 2.9-18.4, 2.5-7.5, 2.5-14.9, and 0.3-2.8 mg/cm3 for methods 1, 2, 3, and 4, respectively (ranges represent variations in the errors of the measured BMDs of the inserts due to changes in scanner table height and lumbar simulator phantom size). CONCLUSION: The single calibration line method is an accurate means of measuring BMD in trauma patients.

Overexpression, purification, and characterization of a thermostable chitinase (Chi40) from Streptomyces thermoviolaceus OPC-520.

A new procedure for the large-scale purification of the recombinant thermostable chitinase (Chi40) cloned from Streptomyces thermoviolaceus in various expression vectors in Escherichia coli is described. Chi40 was overproduced in the cytosolic and secreted forms. The cytosolic form (Chi40c) was highly overproduced and purified by metal-affinity and ion-exchange chromatography in large amounts. The protein was highly active and thermostable but not homogeneous, since a considerable proportion of the Chi40c protein was not correctly folded as determined by native polyacrylamide gel electrophoresis. The Chi40 protein secreted into the culture medium (Chi40s) was purified by hydrophobic interaction and ion-exchange chromatography and high amounts of correctly folded and active Chi40 protein could be recovered in a short time. The enzymatic activity of Chi40s on a synthetic and on its natural substrate, chitin, was studied. Thermostability measurements showed that Chi40 has a T(m) of 60.7 degrees C at neutral pH. (13)C-(15)N double-labeled recombinant Chi40s was also produced and purified from the pECHChi40-9 construct introduced into BL21trxB(DE3) cells grown in minimal medium in the presence of the paramagnetic elements [(13)C]glucose and (15)NH(4)Cl. The presented data open the possibility of an extensive structural study on Chi40s by X-ray crystallography and on enzyme-substrate interaction by NMR spectroscopy.

Phototimer setup for CR imaging.

A study was performed to investigate the feasibility of using the standard deviation (sigma) of the pixel values in a computed radiography (CR) image and a measure of the median incident exposure on the imaging plate (IP) as parameters for setting up phototimers in a CR system. Slabs of Lucite 4-, 6-, and 8-in.-thick were imaged with a CR system at 70, 90, and 125 kVp at various mA s values both with grid and without grid. Incident IP exposures were measured with an ionization chamber. Images were analyzed on a workstation. The sigma's in the "flat field" images were found to be approximately related to the mean incident exposure E by the relationship: sigma is proportional to E(-1/2), indicating the quantum-noise-limited operation of the system. Derived relationships between the reading sensitivity of the (IP) reader (S number) and sigma can be used to obtain images with a specific noise level. At our institution, where a 400 speed screen-film system is used for general radiography and 200 speed for chest radiography, radiologists generally find CR image quality acceptable when sigma < or = 11 (S< or =400) for general radiography (50-90 kVp), and sigma < or =8 (S< or =200) for chest radiography (125 kVp). However, factors other than the amount of x-ray quanta that form the useful image, such as the image processing mode and the amount of collimation, may affect both the sensitivity value and the image quality.

Design and evaluation of an external filter technique for exposure equalization in mammography.

We are developing an external filter method for equalizing x-ray exposure in the peripheral region of the breast. This method requires the use of only a limited number of custom-built filters for different breast shapes in a given view. This paper describes the design methodology for these external filters. The filter effectiveness was evaluated through a simulation study on 171 mediolateral and 196 craniocaudal view digitized mammograms and through imaging of a breast phantom. The degree of match between the simulated filter and the individual 3-D exposure profiles at the breast periphery was quantified. An analysis was performed to investigate the effect of filter misalignment. The simulation study indicates that the filter is effective in equalizing exposures for more than 80% of the breast images in our database. The tolerance in filter misalignment was estimated to be about +/- 2 mm for the CC view and +/- 1 mm for the MLO view at the image plane. Some misalignment artifacts were demonstrated with simulated filtered mammograms.

Metareasoning and meta-level learning in a hybrid knowledge-based architecture.

Ahybrid knowledge-based architecture integrates different problem solvers for the same (sub)task through a control unit operating at a meta-level, the metareasoner, which coordinates the use of, and the communication between, the different problem solvers. A problem solver is defined to be an association between a knowledge intensive (sub)task, an inference mechanism and a knowledge domain view operated by the inference mechanism in order to perform the (sub)task. Important issues in a hybrid system are the metareasoning and learning aspects. Metareasoning encompasses the functions performed by the metareasoner, while learning reflects the ability of the system to evolve on the basis of its experiences in problem solving. Learning occurs at different levels, learning at the meta-level and learning at the level of the specific problem solvers. Meta-level learning reflects the ability of the metareasoner to improve the overall performance of the hybrid system by improving the efficiency of meta-level tasks. Meta-level tasks include the initial planning of problem solving strategies and the dynamic adaptation of chosen strategies depending on new events occurring dynamically during problem solving. In this paper we concentrate on metareasoning and meta-level learning in the context of a hybrid architecture. The theoretical arguments presented in the paper are demonstrated in practice through a hybrid knowledge-based prototype system for the domain of breast cancer histopathology.

Cloning, overproduction, purification and crystallization of the DNA binding protein HU from the hyperthermophilic eubacterium Thermotoga maritima.

The humar gene encoding for the histone-like DNA-binding protein HU from the hyperthermophilic eubacterium Thermotoga maritima was efficiently overexpressed in Escherichia coli under the T7 promoter. The HU protein was purified using SP-Sepharose ion-exchange and heparin-affinity chromatography and was successfully crystallized in ammonium sulfate. The crystals were grown in the tetragonal form in space group P43 or P41 and have unit-cell dimensions a = b = 46.12, c = 77.56 A, alpha = beta = gamma = 90 degrees. The crystals diffract X-rays to 1.6 A resolution using synchrotron radiation and are suitable for determination of the HU structure at high resolution.

Broad beam transmission properties of some common shielding materials for use in diagnostic radiology.

Broad beam geometry was used to measure the x-ray transmission properties of gypsum wallboard, steel, plate glass, and concrete for x-ray tube potentials of 50-125 kVp using an x-ray generator having a three-phase, twelve-pulse waveform and total initial beam filtration sufficient to provide half-value layers representative of those found in common practice and required by regulatory agencies. Measurement results are presented graphically and as numerical fits to a mathematical model of broad beam transmission to permit their use in the design of protective barriers for medical diagnostic x-ray facilities.