Automated Scoring Method Research Articles

Artificial intelligence-assisted quantitative CT analysis of airway changes following SABR for central lung tumors

IntroductionUse of stereotactic ablative radiotherapy (SABR) for central lung tumors can result in up to a 35% incidence of late pulmonary toxicity. We evaluated an automated scoring method to quantify post-SABR bronchial changes by using artificial intelligence (AI)-based airway segmentation. Materials and methodsCentral lung SABR patients treated at Amsterdam UMC (AUMC, internal reference dataset) and Peter MacCallum Cancer Centre (PMCC, external validation dataset) were identified. Patients were eligible if they had pre- and post-SABR CT scans with ≤ 1 mm resolution. The first step of the automated scoring method involved AI-based airway auto-segmentation using MEDPSeg, an end-to-end deep learning-based model. The Vascular Modeling Toolkit in 3D Slicer was then used to extract a centerline curve through the auto-segmented airway lumen, and cross-sectional measurements were computed along each bronchus for all CT scans. For AUMC patients, airway stenosis/occlusion was evaluated by both visual assessment and automated scoring. Only the automated method was applied to the PMCC dataset. ResultsStudy patients comprised 26 from AUMC, and 33 from PMCC. Visual scoring identified stenosis/occlusion in 8 AUMC patients (31 %), most frequently in the segmental bronchi. After airway auto-segmentation, minor manual edits were needed in 9 % of patients. Segmentation for a single scan averaged 83sec (range 73–136). Automated scoring nearly doubled detected airway stenosis/occlusion (n = 15, 58 %), and allowed for earlier detection in 5/8 patients who had also visually scored changes. Estimated rates were 48 % and 66 % at 1- and 2-years, respectively, for the internal dataset. The automated detection rate was 52 % in the external dataset, with 1- and 2-year risks of 56 % and 61 %, respectively. ConclusionAn AI-based automated scoring method allows for detection of more bronchial stenosis/occlusion after lung SABR, and at an earlier time-point. This tool can facilitate studies to determine early airway changes and establish more reliable airway tolerance doses.

DReD–A Descriptive Relation Dataset for Expanding Relation Extraction

Relation extraction is a fundamental topic in document information extraction. Traditionally, datasets for relation extraction have been annotated with named entities and classified with a subset of relation categories. Models then predict either the entities and relations (end-to-end) or assume the entities are given and only classify the relations. However, current approaches are limited by datasets with a narrow definition of entities and relations. We seek to remedy this by introducing our Descriptive Relation Dataset (DReD), which contains 3286 annotations for descriptions of relations between more general noun phrases inspired by linguistic theory. We benchmark our dataset using several seq2seq models and find that T5 achieves the best results with a ROUGE-1 score of 75.5. We verify the usefulness of DreD by collecting feedback on 100 predictions and comparing human judgment to automated scoring methods. Finally, we verify that relations can be described accurately by transforming the CoNLL04 and Re-TACRED datasets and mapping sentence templates to relation categories. T5 achieves competitive accuracy on CoNLL-04 and Re-TACRED with an F1 score of 78.6 and 90.4, respectively. With this paper, we prove that relations can be described, therefore overcoming the limitations set by previous datasets and approaches. We publicly provide our dataset and training code at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/logan-markewich/DReD</uri> .

Combining Human and Automated Scoring Methods in Experimental Assessments of Writing: A Case Study Tutorial

In a randomized trial that collects text as an outcome, traditional approaches for assessing treatment impact require that each document first be manually coded for constructs of interest by human raters. An impact analysis can then be conducted to compare treatment and control groups, using the hand-coded scores as a measured outcome. This process is both time and labor-intensive, which creates a persistent barrier for large-scale assessments of text. Furthermore, enriching one’s understanding of a found impact on text outcomes via secondary analyses can be difficult without additional scoring efforts. The purpose of this article is to provide a pipeline for using machine-based text analytic and data mining tools to augment traditional text-based impact analysis by analyzing impacts across an array of automatically generated text features. In this way, we can explore what an overall impact signifies in terms of how the text has evolved due to treatment. Through a case study based on a recent field trial in education, we show that machine learning can indeed enrich experimental evaluations of text by providing a more comprehensive and fine-grained picture of the mechanisms that lead to stronger argumentative writing in a first- and second-grade content literacy intervention. Relying exclusively on human scoring, by contrast, is a lost opportunity. Overall, the workflow and analytical strategy we describe can serve as a template for researchers interested in performing their own experimental evaluations of text.

Explainable automated evaluation of the clock drawing task for memory impairment screening.

The clock drawing task (CDT) is frequently used to aid in detecting cognitive impairment, but current scoring techniques are time-consuming and miss relevant features, justifying the creation of an automated quantitative scoring approach. We used computer vision methods to analyze the stored scanned images (N=7,109), and an intelligent system was created to examine these files in a study of aging World Trade Center responders. Outcomes were CDT, Montreal Cognitive Assessment (MoCA) score, and incidence of mild cognitive impairment (MCI). The system accurately distinguished between previously scored CDTs in three CDT scoring categories: contour (accuracy=92.2%), digits (accuracy=89.1%), and clock hands (accuracy=69.1%). The system reliably predicted MoCA score with CDT scores removed. Predictive analyses of the incidence of MCI at follow-up outperformed human-assigned CDT scores. We created an automated scoring method using scanned and stored CDTs that provided additional information that might not be considered in human scoring.

Application of a semi-automated dicentric scoring system in triage and monitoring occupational radiation exposure.

The dicentric chromosome assay (DCA) is considered the gold standard for radiation biodosimetry, but it is limited by its long dicentric scoring time and need for skilled scorers. The automation of scoring dicentrics has been considered a strategy to overcome the constraints of DCA. However, the studies on automated scoring methods are limited compared to those on conventional manual DCA. Our study aims to assess the performance of a semi-automated scoring method for DCA using ex vivo and in vivo irradiated samples. Dose estimations of 39 blind samples irradiated ex vivo and 35 industrial radiographers occupationally exposed in vivo were estimated using the manual and semi-automated scoring methods and subsequently compared. The semi-automated scoring method, which removed the false positives of automated scoring using the dicentric chromosome (DC) scoring algorithm, had an accuracy of 94.9% in the ex vivo irradiated samples. It also had more than 90% accuracy, sensitivity, and specificity to distinguish binary dose categories reflecting clinical, diagnostic, and epidemiological significance. These data were comparable to those of manual DCA. Moreover, Cohen's kappa statistic and McNemar's test showed a substantial agreement between the two methods for categorizing in vivo samples into never and ever radiation exposure. There was also a significant correlation between the two methods. Despite of comparable results with two methods, lower sensitivity of semi-automated scoring method could be limited to assess various radiation exposures. Taken together, our findings show the semi-automated scoring method can provide accurate dose estimation rapidly, and can be useful as an alternative to manual DCA for biodosimetry in large-scale accidents or cases to monitor radiation exposure of radiation workers.

Investigating Relationship between EFL Teachers’ Personality Traits and Fluency Ratings

The objective of this study was to investigate the connection between EFL teachers" personality traits and fluency perceptions, assuming that personality could play a role in deciding listeners" fluency perceptions in the EFL context. Twenty-four listeners (13 NSs and 11 NNSs of English teachers) were asked to judge the fluency of 28 Korean high school students" narrative speech samples and complete the Big Five Aspects Scales personality test (DeYong et al., 2007). Listeners" mean fluency ratings were compared to their personality scores. Statistical analysis results indicated that Extraversion, one of the Big Five personality traits, was moderately correlated with fluency scores in the expected direction, suggesting that more extraverted people rated L2 speech as fluent. On the other hand, the other four traits were not substantially associated with fluency scores. This study not only emphasizes the significance of personality in influencing fluency perception, but also shows that fluency scores assigned by EFL teachers should be interpreted with caution. It opens up the possibility of using a more objective and accurate automated scoring method to assess oral fluency.

Automated Essay Scoring Effect on Test Equating Errors in Mixed-format Test

Scoring constructed-response items can be highly difficult, time-consuming, and costly in practice. Improvements in computer technology have enabled automated scoring of constructed-response items. However, the application of automated scoring without an investigation of test equating can lead to serious problems. The goal of this study was to score the constructed-response items in mixed-format tests automatically with different test/training data rates and to investigate the indirect effect of these scores on test equating compared with human raters. Bidirectional long-short term memory (BLSTM) was selected as the automated scoring method for the best performance. During the test equating process, methods based on classical test theory and item response theory were utilized. In most of the equating methods, errors of the equating resulting from automated scoring were close to the errors occurring in equating processes conducted by human raters. It was concluded that automated scoring can be applied because it is convenient in terms of equating.

Abstract P1-03-01: An international multicenter study to evaluate reproducibility of automated scoring methods for assessment of Ki67 in breast cancer

Abstract Background: The nuclear proliferation biomarker Ki67 has multiple potential roles in breast cancer, including prognosis-based decisions, but unacceptable between-laboratory variability has limited its clinical value. The International Ki67 Working Group (IKWG) has undertaken a systematic program to determine whether Ki67 immunohistochemistry can be analytically validated and standardized across laboratories. Technological advances and broader availability of devices for automated assessment of stained slides raise the possibility that these machines may improve on reproducibility of traditional pathologist-based visual Ki67 assessment. Aims: To characterize reproducibility of automated machine-measured Ki67 expression using slides previously analyzed in the IKWG phase 3 study that evaluated reproducibility of visual Ki67 assessment. Methods: Two sets of 30 previously stained slides containing core-cut biopsy sections of breast tumors were circulated to 14 laboratories for scanning and automated assessment of Ki67 expression. Sites were instructed to return average and maximum percentage of tumor cells positive for Ki67 for each slide, where maximum is designed to reflect “hot spot” analysis. Two laboratories returned scores from 2 operators; not all laboratories reported values for maximum Ki67 scores. Different operators were treated as distinct laboratories in analyses. Sixteen and 10 score sets were available for average and maximum Ki67 analyses, respectively, encompassing 7 unique scanner and 10 software platforms. Pre-specified analyses included evaluation of reproducibility across all laboratories as well as within a subgroup limited to those using Aperio scanners. The primary reproducibility metric was intraclass correlation coefficient between laboratories (ICC), regardless of device platform or software. Results: Geometric means across 30 cases for 16 operators ranged from 11.06% to 38.11% with overall mean 16.75% (95% CI:14.45-19.42) for average scores. Geometric means for 10 operators ranged from 16.44% to 68.73% with overall mean 25.16% (95% CI: 18.71-33.84) for maximum scores. ICC for automated average scores across 16 operators was 0.83 (95% CI: 0.73-0.91) and ICC for maximum scores across 10 operators was 0.63 (95% CI: 0.44-0.80) although one outlier lab dramatically affected results. For the laboratories using the Aperio platform (8 score sets), ICC for automated average scores was 0.89 (95% CI; 0.81-0.96). These results are similar to ICC of 0.87 (95%CI; 0.81-0.93) reported using these same slides in the Phase 3 visual assessment reproducibility study in which observers counted 500 cells per slide (Leung et al, NPJBrCancer, in press). Conclusions: Between-laboratory reproducibility for automated machine assessment of average Ki67 is similar to that for pathologist-based visual assessment of Ki67. However, the observed ICC was markedly numerically lower for the maximum score method compared to the average method, suggesting that the maximum score may not be useful as a reproducible measure of proliferation. Automated average scoring methods show promise for standardization of Ki67 scoring, supporting future studies to clinically validate Ki67. Citation Format: Rimm DL, McShane LM, Leung SCY, Bai Y, Bane AL, Bartlett JMS, Bayani J, Chang MC, Dean M, Denkert C, Enwere E, Galderisi C, Gholap A, Hugh JC, Jadhav A, Kornaga E, Laurinavicius A, Levenson R, Lima J, Miller K, Pantanowitz L, Piper T, Ruan J, Srinivasan M, Virk S, Wu Y, Yang H, Hayes DF, Nielsen TO, Dowsett M. An international multicenter study to evaluate reproducibility of automated scoring methods for assessment of Ki67 in breast cancer [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-03-01.

Human sperm sex chromosome disomy and sperm DNA damage assessed by the neutral comet assay.

Is there an association between human sperm sex chromosome disomy and sperm DNA damage? An increase in human sperm XY disomy was associated with higher comet extent; however, there was no other consistent association of sex chromosome disomies with DNA damage. There is limited published research on the association between sex chromosome disomy and sperm DNA damage and the findings are not consistent across studies. We conducted a cross-sectional study of 190 men (25% ever smoker, 75% never smoker) from subfertile couples presenting at the Massachusetts General Hospital Fertility Clinic from January 2000 to May 2003. Multiprobe fluorescence in situ hybridization for chromosomes X, Y and 18 was used to determine XX, YY, XY and total sex chromosome disomy in sperm nuclei using an automated scoring method. The neutral comet assay was used to measure sperm DNA damage, as reflected by comet extent, percentage DNA in the comet tail, and tail distributed moment. Univariate and multiple linear regression models were constructed with sex chromosome disomy (separate models for each of the four disomic conditions) as the independent variable, and DNA damage parameters (separate models for each measure of DNA damage) as the dependent variable. Men with current or past smoking history had significantly greater comet extent (µm: regression coefficients with 95% CI) [XX18: 15.17 (1.98, 28.36); YY18: 14.68 (1.50, 27.86); XY18: 15.41 (2.37, 28.45); Total Sex Chromosome Disomy: 15.23 (2.09, 28.38)], and tail distributed moment [XX18: 3.01 (0.30, 5.72); YY18: 2.95 (0.24, 5.67); XY18: 3.04 (0.36, 5.72); Total Sex Chromosome Disomy: 3.10 (0.31, 5.71)] than men who had never smoked. In regression models adjusted for age and smoking, there was a positive association between XY disomy and comet extent. For an increase in XY disomy from 0.56 to 1.47% (representing the 25th to 75th percentile), there was a mean increase of 5.08 µm in comet extent. No other statistically significant findings were observed. A potential limitation of this study is that it is cross-sectional. Cross-sectional analyses by nature do not lend themselves to inference about directionality for any observed associations; therefore we cannot determine which variable is the cause and which one is the effect. A small sample size may be a further limitation. Comparison of these findings to other studies is limited due to methodological differences. Although consistent associations across sex chromosome disomies or DNA damage measures were not observed, this study highlights the need to explore etiologies of sperm DNA damage and sex chromosome disomy to better understand the potential mechanistic overlaps between the two. This work was supported by NIOSH Grant T42 OH008416, and NIH/NIEHS Grants ES 009718, ES 000002, and R01 ES017457. During the study M.E.M. was affiliated with the Department of Environmental Health at the Harvard School of Public Health. N/A.

Comparison of visual and automated assessment of tumour inflammatory infiltrates in patients with colorectal cancer

BackgroundCancer-associated inflammation is increasingly recognised to be an important determinant of oncological outcome. In colorectal cancer, the presence of peri-tumoural inflammatory/lymphocytic infiltrates predicts improved survival. To date, these infiltrates, assessed visually on haematoxylin and eosin (H&E) stained sections, have failed to enter routine clinical practice, partly due to their subjective assessment and considerable inter-observer variation. The present study aims to develop an automated scoring method to enable consistent and reproducible assessment of tumour inflammatory infiltrates in colorectal cancer. Methods154 colorectal cancer patients who underwent curative resection were included in the study. The local inflammatory infiltrate was assessed using the method described by Klintrup–Makinen. H&E tumour sections were uploaded to an image analysis programme (Slidepath, Leica Biosystems). An image analysis algorithm was developed to count the inflammatory cells at the invasive margin. The manual and automated assessments of the tumour inflammatory infiltrates were then compared. ResultsThe automated inflammatory cell counts assessed using the freehand annotation method (p<0.001) and the rectangular box method (p<0.001) were significantly associated with both K–M score (p<0.001) and K–M grade (p<0.001). The inflammatory cell counts were divided using quartiles to group tumours with similar inflammatory cell densities. There was good agreement between the manual and automated scoring methods (intraclass correlation coefficient (ICC)=0.82). Similar to the visual K–M scoring system, the automated K–M classification of the inflammatory cell counts, using quartiles, was significantly associated with venous invasion (p<0.05) and modified Glasgow Prognostic Score (mGPS) (p⩽0.05). On univariate survival analysis, both automated K–M category (p<0.05) and automated K–M grade (p<0.005) were associated with cancer-specific survival. ConclusionThe results of the present study demonstrate that automated assessment effectively recapitulates the clinical value of visual assessment of the local inflammatory cell infiltrate at the invasive margin of colorectal tumours. In addition, it is possible to obtain an objective assessment of tumour inflammatory infiltrates using routinely stained H&E sections. An automated, computer-based scoring method is therefore a workable and cost-effective approach to clinical assessment of local immune cell infiltrates in colorectal cancer.

Comparison of Automated Scoring Methods for a Computerized Performance Assessment of Clinical Judgment

Growing reliance on complex constructed response items has generated considerable interest in automated scoring solutions. Many of these solutions are described in the literature; however, relatively few studies have been published that compare automated scoring strategies. Here, comparisons are made among five strategies for machine-scoring examinee performances of computer-based case simulations, a complex item format used to assess physicians’ patient-management skills as part of the Step 3 United States Medical Licensing Examination. These strategies utilize expert judgments to obtain various (a) case-specific or (b) generic scoring algorithms. The various compromises between efficiency, validity, and reliability that characterize each scoring approach are described and compared.

Immunohistochemical analysis of breast tissue microarray images using contextual classifiers

Background: Tissue microarrays (TMAs) are an important tool in translational research for examining multiple cancers for molecular and protein markers. Automatic immunohistochemical (IHC) scoring of breast TMA images remains a challenging problem. Methods: A two-stage approach that involves localization of regions of invasive and in-situ carcinoma followed by ordinal IHC scoring of nuclei in these regions is proposed. The localization stage classifies locations on a grid as tumor or non-tumor based on local image features. These classifications are then refined using an auto-context algorithm called spin-context. Spin-context uses a series of classifiers to integrate image feature information with spatial context information in the form of estimated class probabilities. This is achieved in a rotationally-invariant manner. The second stage estimates ordinal IHC scores in terms of the strength of staining and the proportion of nuclei stained. These estimates take the form of posterior probabilities, enabling images with uncertain scores to be referred for pathologist review. Results: The method was validated against manual pathologist scoring on two nuclear markers, progesterone receptor (PR) and estrogen receptor (ER). Errors for PR data were consistently lower than those achieved with ER data. Scoring was in terms of estimated proportion of cells that were positively stained (scored on an ordinal scale of 0-6) and perceived strength of staining (scored on an ordinal scale of 0-3). Average absolute differences between predicted scores and pathologist-assigned scores were 0.74 for proportion of cells and 0.35 for strength of staining (PR). Conclusions: The use of context information via spin-context improved the precision and recall of tumor localization. The combination of the spin-context localization method with the automated scoring method resulted in reduced IHC scoring errors.

TU‐E‐BRC‐09: An Automated Scoring Method for Motion Artifacts in Cine Four‐Dimensional Computed Tomography Images

Purpose: The aim of this work is to develop an automated method to objectively score the severity of image artifacts in four‐dimensional (4D) computed tomography (CT). Methods: We developed a metric based on the difference of the two‐dimensional normalized correlation coefficients (NCC) between the edge slices at couch transitions and the NCC between edge slices and their neighbors within their respective couch positions. We applied our method to the 4D‐CT images of ten patients that were initially retrospectively sorted using the breathing trace with miscalculated phase values, and then re‐sorted using the same breathing trace with recalculated phase values after manually correcting the peaks of inhalation. The metric was calculated for both sets of 4D‐CT images and a score was determined to assess the severity of the image artifacts in both sets. The accuracy of our method to identify the 4D‐CT image set with fewer or lesser artifacts was validated by two human observers who evaluated three coronal and two sagittal cross‐sections from the same locations in both image sets. The two sets were displayed side by side and each observer independently scrolled through ten respiratory phases of both sets simultaneously (blinded to the selection method of each set), and marked the image set that appeared to have fewer or lesser artifacts. Results: Our method determined that the 4D‐CT sets using the breathing trace with recalculated phase values after manually correcting the peaks of inhalation to result in images with fewer or lesser artifacts. This is in good agreement with the findings of both observers. The observers agreed forty times and disagreed one time among total fifty selections. Conclusions: In this study we have demonstrated proof of principle that our method is able to replicate the findings of two human observers in determining 4D‐CT images with fewer or lesser artifacts.The authors greatly acknowledge the support from Mr. and Mrs. T. Archer Morgan Fellowship in Radiation Oncology at Stanford University.

SU-GG-I-122: An Automated Scoring Method to Quantitatively Evaluate Imaging Artifacts in 4D-CT Images

Purpose: To develop an automated scoring method based on normalized cross correlation to evaluate imaging artifacts due to respiratory motion in four‐dimensional‐computed tomography (4D‐CT) images.Method and Materials:Imaging artifacts in 4D CT have been extensively documented in the literature. Manual analysis of the magnitude of the artifacts is subjective and time consuming and there has no automated methods been reported for the objective quantification of the artifacts. We have developed an automated method to score the 4D‐CT images based on the calculation of the normalized cross‐correlation coefficients (NCC) between adjacent image slices at couch transitions (edge slices). We used three scoring criteria to evaluate the imaging artifacts: the minimum of the NCC (score #1), the average NCC (score #2), and the mean of the difference between the NCC of edge slices and the mean NCC of the edge slices with their respective neighboring slices within the same couch position (score #3). Our method was tested using a commercially available 4D motion phantom and subsequently applied to the 4D‐CT scans of 30 lungcancer patients. Results: Our method is able to detect imaging artifacts in the reconstructed respiratory phases of the 4D‐CT scans that are verified visually. In particular, score #1 correlates well with the maximum magnitudes of two types of artifacts that were measured manually by an unbiased observer. In contrast, scores #2 and #3 showed no correlations. Conclusions: Our automated scoring method based on normalized cross correlation is suitable to quantitatively evaluate imaging artifacts. Score #1 appears to be the most reliable measure to quantitate artifacts and has been verified visually. Used in conjunction with score #1, score #2 can be an indicator of the overall goodness of the image quality while score #3 indicates the ‘badness’ of edge slices as compared with their respective neighboring slices.

Abstract 786: Alterations of TGFβ/Smad signaling in human head and neck squamous cell carcinomas

Abstract Head-and-neck squamous cell carcinoma (HNSCC) ranks as the sixth most frequent cancer worldwide. Based on earlier in vitro studies of HNSCC cell lines and genetic analysis of human HNSCC specimens, we postulated that disruptions of Transforming Growth Factor- β (TGF-β) signaling contribute to the development and progression of HNSCC. Smad proteins are the intracellular components of the TGF-β signaling pathway. In a previous immunohistochemical study of the expression and activation of Smads in 170 HNSCC specimens assembled in a tissue microarray, phospho-Smad2 (pSmad2), an active form of Smad2 and Smad4 were absent in 14% and 22% of cases respectively. Moreover, there was a trend towards improved overall survival (OS) among patients with pSmad2-negative tumors. In the present study, we analyzed the pattern of expression of (p)Smad2, (p)Smad3 and Smad4 in a much larger 4-fold redundant tissue array of 798 HNSCC tumor samples obtained from 346 patients. In addition, we compared a new automated scoring method to standard visual scoring of immunostained slides. Each tissue core was visually scored as positive or negative, and the four scores averaged to derive a visual score (VS). Automated scores (AS) were obtained by scanning all slides using a Zeiss MedMicro whole slide scanner at 40×equivalent resolution. Images of individual cores were extracted from the scanned archive and staining maps of chromogen (brown) were isolated from the counter stain using our previously reported color decomposition algorithm. AS was computed based on the staining characteristic of each core and a final score for each specimen was obtained by averaging the scores of four matched cores. Receiver operating characteristic (ROC) curves demonstrated that the AS scoring method performed extremely well compared the VS “gold standard”, with area under the curve (AUC) values of &amp;gt;80% for all 5 Smad immunostains. We found that 19% (95% CI: 16-21%), 40% (95% CI: 36-43%) and 12% (95% CI: 10-15%) of HNSCC specimens failed to express pSmad2, pSmad3 or Smad4, respectively. Loss of TGF-β receptor signaling, as indicated by both pSmad2 and −3 negativity, was observed in 11% (95% CI: 9-13%) of specimens. In addition, 10% (95% CI: 8-12%) of specimens failed to express Smad4. These results are consistent with our previous findings in a smaller sample set. Moreover, among 198 patients with survival information, patients with pSmad2/3 negative tumors had a significantly higher OS when compared with patients with pSmad2/3 positive tumors (p=0.0090). In contrast, loss of Smad4 expression was not associated with poor survival. These results indicate that inactivation of TGF-β receptor signaling or of Smad4 in HNSCC have different clinical implications and may reflect underlying differences in biological significance. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 786.

Ein Vergleich zwischen Aufzeichnung und Auswertung von Schlaf-EEG durch QUISI, Version 1.0 und visueller Auswertung durch PSG

QUISI has been developed as a one-channel, ambulatory EEG recording device with 3 electrodes placed on the forehead close to Fp1, Fz and Fp2. While Fz serves as ground electrode, the channel Fp1−Fp2 was recorded, filtered and analyzed by Fourier-Transform with the subsequent determination of power spectrum estimates. These variables served as input variables for neural network technique classification, taking 12 input variables, 2 hidden layers and 7 outcome variables for 0=Movement Time, 1=Wake, 2=REM, 3=S1, 4=S2, 5=S3 and 6=S4. Out of 118 primary values, 12 were selected using evolutionary and genetic algorithms. 8 neural networks were established using 8 different patients. Every 30 s sleep EEG epoch (segment) was subsequently classified 8 times. The final decision for each epoch of 30 s was made taking the median of the 8 classifications. Subsequently smoothing rules were applied in analogy to the Rechtschaffen and Kales smoothing rules. The analysis is fully automated. There is no artefact rejection and no removal of any sleep epoch. Once the rule was established, it was validated by forward classification of 38 sleep disturbed patients, none of whom had been used for the establishment of the classification rule. Every patient had been recorded with polysomnography (PSG) as well according to the standards of the German Sleep Research Society and with the QUISI equipment. The results show an acceptable agreement between PSG with visual scoring and QUISI with automatic scoring. While SPT, TST and SOL showed high correlations, REM and SWS were generally overestimated. Subsequently we evaluated the pathologies in QUISI and PSG. Out of the 38 patients, only 1 patient had a normal profile both in QUISI and PSG. Out of the 37 patients with pathologies, 17 showed exactly the same pathologies in the PSG and QUISI profile. In 1 patient PSG and QUISI showed completely different pathologies. In 10 patients QUISI did not find at least one pathology seen in the PSG, in 4 patients QUISI found more pathologies than were seen in the PSG, in 5 patients both applied. In its current version QUISI often over-estimates SWS and REM in certain patients, as known from other automated scoring methods described earlier in the literature. QUISI is considered as the beginning of a development with the aim of obtaining an affordable and easily self-applicable sleep EEG recording device, which can be applied before a therapy control and after adminssion to a sleep laboratory. QUISI however can in no way replace sleep laboratory diagnostics.