Search PubMed⌕ Search

Biomedical subjects

U Hahn

Publications and source records attributed to U Hahn.

At least 91 records · Page 5Linked to original sources
PubMed1999

How knowledge drives understanding--matching medical ontologies with the needs of medical language processing.

In this article, we introduce a knowledge-based approach to medical text understanding. From an in-depth consideration of deep sentence and text understanding we distill basic requirements for an adequate knowledge representation framework. These requirements are then matched with currently available medical ontologies (thesauri, terminologies, etc.). A fundamental trade-off is recognized between large-scale conceptual coverage on the one hand, and formal mechanisms for integrity preservation and conceptual expressiveness on the other hand. We discuss various shortcomings of the most wide-spread ontologies to capture medical knowledge in-the-large. As a result, we argue for the need of a formally sound and expressive model along the lines of KL-ONE-style terminological representation systems in the format of description logics. These provide an adequate methodology for designing more sophisticated, flexible medical ontologies serving the needs of 'deep' knowledge applications which are by no means restricted to medical language processing.

Artificial Intelligence↗
PubMed1999

Discourse structures in medical reports--watch out! The generation of referentially coherent and valid text knowledge bases in the MEDSYNDIKATE system.

The automatic analysis of medical narratives currently suffers from neglecting text structure phenomena such as referential relations between discourse units. This has unwarranted effects on the descriptional adequacy of medical knowledge bases automatically generated from texts. The resulting representation bias can be characterized in terms of incomplete, artificially fragmented and referentially invalid knowledge structures. We focus here on four basic types of textual reference relations, viz. pronominal and nominal anaphora, textual ellipsis and metonymy and show how to deal with them in an adequate text parsing device. Since the types of reference relations we discuss show an increasing dependence on conceptual background knowledge, we stress the need for formally grounded, expressive conceptual representation systems for medical knowledge. Our suggestions are based on experience with MEDSYNDIKATE, a medical text knowledge acquisition system designed to properly deal with various sorts of discourse structure phenomena.

Artificial Intelligence↗
PubMed1999

Stage III and oestrogen receptor negativity are associated with poor prognosis after adjuvant high-dose therapy in high-risk breast cancer.

We report on the efficacy and toxicity of a sequential high-dose therapy with peripheral blood stem cell (PBSC) support in 85 patients with high-risk stage II/III breast cancer. There were 71 patients with more than nine tumour-positive axillary lymph nodes. An induction therapy of two cycles of ifosfamide (total dose, 7.5 g m(-2)) and epirubicin (120 mg m(-2)) was given, and PBSC were harvested during G-CSF-supported leucocyte recovery following the second cycle. The PBSC-supported high-dose chemotherapy consisted of two cycles of ifosfamide (total dose, 12,000 mg m(-2)), carboplatin (900 mg m(-2)) and epirubicin (180 mg m(-2)). Patients were autografted with a median number of 3.7 x 10(6) CD34+ cells kg(-1) (range, 1.9-26.5 x 10(6)) resulting in haematological reconstitution within approximately 2 weeks following high-dose therapy. The toxicity was moderate in general, and there was no treatment-related toxic death. Twenty-one patients relapsed between 3 and 30 months following the last cycle of high-dose therapy (median, 11 months). The probability of disease-free and overall survival at 4 years were 60% and 83%, respectively. According to a multivariate analysis, patients with stage II disease had a significantly better probability of disease-free survival (74%) in comparison to patients with stage III disease (36%). The probability of disease-free survival was also significantly better for patients with oestrogen receptor-positive tumours (70%) compared to patients with receptor-negative ones (40%). Bone marrow samples collected from 52 patients after high-dose therapy were examined to evaluate the prognostic relevance of isolated tumour cells. The proportion of patients presenting with tumour cell-positive samples did not change in comparison to that observed before high-dose therapy (65% vs 71%), but a decrease in the incidence and concentration of tumour cells was observed over time after high-dose therapy. This finding was true for patients with relapse and for those in remission, which argues against a prognostic significance of isolated tumour cells in bone marrow. In conclusion, sequential high-dose chemotherapy with PBSC support can be safely administered to patients with high-risk stage II/III breast cancer. Further intensification of the therapy, including the addition of non-cross resistant drugs or immunological approaches such as the use of antibodies against HER-2/NEU, may be envisaged for patients with stage III disease and hormone receptor-negative tumours.

Adult↗
PubMed1999

Stereological analysis and modelling of gradient structures

Gradient structures are inhomogeneous along a particular gradient direction but homogeneous perpendicular to that direction. Consequently, structural parameters such as volume fraction or surface area density are local characteristics which depend on the 'vertical' coordinate with respect to the 'vertical' gradient axis. Analogously, models for gradient structures have model parameters depending on the vertical coordinate z. For example, a Voronoi tessellation with a gradient is generated by a gradient point process with a local intensity which is a function of z. Similarly, a gradient germ grain model is obtained from a gradient point process where the grain size distribution may also depend on z. For a gradient Boolean model, local volume fraction VV(z) and local surface area density SV(z) can be calculated from the model parameters. Stereological methods for gradient structures are based on vertical sections parallel to the gradient direction. Estimation of VV(z), SV(z) and local length density LV(z) is done by lineal analysis using horizontal test lines with vertical coordinate z. Similarly, lineal analysis is used to estimate local mean cell volume of gradient tessellations. For the estimation of local particle number density and size in the spirit of the Wicksell problem the use of kernel methods and distributional assumptions is required.

Journal Article↗
PubMed1999

[Cardiac MRI for determining functional left ventricular parameters].

PURPOSE: To prove the accuracy of MR methods in the determination of left ventricular (LV) functional parameters and anatomy. MATERIALS AND METHODS: At 1.5 T, 20 healthy volunteers and 22 patients with aortic valvular disease (stenosis n = 15, regurgitation n = 7) were examined. Functional parameters like cardiac output, ejection fraction, end-diastolic volume, aortic flow maximum, and time interval from the R-wave to maximum flow were obtained using a velocity encoding 2D FLASH sequence (TR 24 ms, TE 5 ms, venc 250 cm/sec) and segmented breath-hold cine FLASH 2D technique (TR 100 ms, TE 4.8 ms, flip angle 25 degrees, temporal resolution 50 ms). Invasive measurements (Fick principle) served as gold standard, intra- and interobserver variability were determined. RESULTS: Differences of functional parameters between normal volunteers and patients were detectable at a high level of significance (p < 0.0001). For cardiac output a superior correlation with the gold standard was found using flow measurements (r = 0.66, p < 0.0007) compared to volumetric calculations from cine studies (r = 0.47, p < 0.02). Interobserver variability was 2.5 +/- 2.7%/4.5 +/- 6.9% (flow quantification/calculations from cine studies), intraobserver variability was 1.7 +/- 1.6%/3.3 +/- 2.2%. CONCLUSIONS: MRI is an appropriate tool for determining LV functional parameters and anatomy. Differences between normal volunteers and patients with aortic valvular disease can be detected reliably. Flow measurements turned out to be more accurate than calculations from cine images. Therefore, flow quantification techniques should be preferred for clinical use.

Adult↗
PubMed1999

[Optimization of numerical measurement parameters for ECG-triggered MRI snapshot-FLASH myocardial perfusion studies].

BACKGROUND AND AIMS: In MR examinations of myocardial perfusion by the use of Snapshot-FLASH sequences it is of major importance that the achievable signal difference between pre- and normal postcontrast myocardium be maximized. METHODS: In ECG-triggered Snapshot-FLASH sequences the signal intensity of the myocardium depends on the flip angle alpha, the inversion time TI and the trigger delay TD (both depending on the cardiac frequency f) for unchanged slice thickness (SL), matrix size (MA), repetition time (TR) and echo time (TE). Therefore a simulation of the signal behavior of pre- and postcontrast myocardium based on Bloch's equations was performed by varying the flip angle alpha, TI and TD for different cardiac frequencies in order to determine an optimized combination of the measurement parameters. RESULTS: In normal heart rates (50-70 beats/min) maximal signal differences between pre- and normal postcontrast myocardium can be reached for inversion times TI = 170-200 ms and a flip angle alpha = 11 degrees. For higher heart rates again alpha = 11 degrees and TI = 200-220 ms with shortened TD (TD = 0 for f > 90 beats/min) were found to be optimal. The calculated values were semiquantitatively confirmed in phantom and volunteer measurements. CONCLUSIONS: The described method allows cardiac frequency dependent optimization of the Snapshot-FLASH measurement parameters alpha, TI and TD in order to reach a maximum in signal contrast between normal and malperfused myocardium.

Computer Graphics↗
PubMed1999

[Value of CT in diagnosis of respirator-associated pneumonia].

PURPOSE: To analyse the diagnostic accuracy of computed tomography (CT) in ventilator-associated pneumonia (VAP). MATERIALS AND METHODS: 23 patients on mechanical ventilation with a new pulmonary abnormality on chest X-ray were examined with both spiral-CT and high-resolution CT. The diagnosis VAP was made according to prospectively defined criteria. Bronchoscopic specimen asservation with protected specimen brushing (PSB) served as gold standard. RESULTS: With PSB, 11 of 23 patients were found to have VAP. CT showed a sensitivity and specificity of 53% and 63%, respectively. Ground glass infiltrates appeared to have a 100% specificity but were found in only 5/11 patients. CONCLUSIONS: CT is not the method of choice for diagnosing VAP. Ground glass infiltrates seeming to be highly specific are only inconstantly found.

Adult↗
PubMed1999

[Diagnosis of renal artery stenosis in 1.0 T using 3D phase contrast magnetic resonance angiography and dynamic contrast medium perfusion].

PURPOSE: To assess renal artery stenosis (RAS) by 3D phase contrast (PC) MR angiography and dynamic perfusion imaging of the kidneys. METHODS: On a standard 1.0 T MR imaging system (Magnetom Expert, Siemens), 32 patients with angiographically proven unilateral RAS were examined using a 3D PC sequence (TR 40 ms/TE 9 ms/venc 30 cm/s). An ECG-gated Turbo-FLASH 2D sequence (TR 4.5 ms/TE 2.2 ms/TIeff. 400 ms) was applied to study the first pass of paramagnetic contrast agent (0.1 mmol Gd-DTPA/kg) through the kidneys. Signal intensity (SI) over time curves of the renal cortex were obtained and evaluated considering temporal relation and percentage of maximum SI compared to the aorta and normal kidneys. Analysis of the MRA was performed by two independent blinded readers. The gold-standard DSA was interpreted by consensus reading of two experienced radiologists. RESULTS: RAS was detected by 3D PC MRA with a sensitivity of 93% and specificity of 81% (ppv 82%, npv 93%, accuracy 87%, kappa = 0.61). Maximum SI in RAS was significantly decreased (p < 0.001-0.0001). A temporally delayed enhancement of 1.5 +/- 1.3 s was found for RAS > 75% (p < 0.002) but not for RAS < 75% (p > 0.1). CONCLUSIONS: 3D PC MRA is capable of detecting RAS in a high percentage of patients. Dynamic perfusion imaging of the kidneys, applied additionally, can confirm the diagnosis and give valuable information about the hemodynamic relevance of RAS in suspected unilateral disease.

Contrast Media↗
PubMed1999

MR angiography and flow quantification of the internal mammary artery graft after minimally invasive direct coronary artery bypass.

OBJECTIVE: Six patients who had undergone minimally invasive direct coronary artery bypass surgery were examined to evaluate an MR imaging protocol that provided information about cardiac function, bypass graft patency, and flow characteristics with a single examination. CONCLUSION: Preliminary results suggest that our imaging protocol allows accurate follow-up of patients after minimally invasive direct coronary artery bypass surgery. Bypass graft patency was correctly determined in all patients. In four patients, anastomoses were visualized by MR angiography, and flow measurements revealed a volume range of 28-84 ml/min (native and grafted internal mammary arteries) and a trend for the flow values of bypass grafts to be lower than those of native vessels. Interobserver reproducibility was good (r = .99; slope, .98).

Contrast Media↗
PubMed1999

Renal MR angiography at 1.0 T: three-dimensional (3D) phase-contrast techniques versus gadolinium-enhanced 3D fast low-angle shot breath-hold imaging.

OBJECTIVE: The purpose of this study was to evaluate the diagnostic usefulness of three different MR angiographic techniques at 1.0 T. SUBJECTS AND METHODS: In 22 patients with renal artery stenosis confirmed at intraarterial catheter angiography, we also performed unenhanced and gadolinium-enhanced three-dimensional phase-contrast MR angiography and gadolinium-enhanced single breath-hold three-dimensional fast low-angle shot MR angiography. We determined circulation time to optimize signal acquisition in gadolinium-enhanced breath-hold MR angiography after bolus injection of contrast material. RESULTS: Sensitivity, defined as the detection of a hemodynamically significant stenosis (>50% luminal narrowing), was 85% for enhanced phase-contrast MR angiography, 91% for gadolinium-enhanced MR angiography, and 95% for unenhanced phase-contrast MR angiography. The combination of unenhanced phase-contrast MR angiography and gadolinium-enhanced MR angiography yielded 100% sensitivity for hilar artery stenoses. There were 13 false-positive findings with unenhanced phase-contrast MR angiography, 10 with enhanced phase-contrast MR angiography, and four with gadolinium-enhanced MR angiography (specificity: 38%, 52%, and 79%, respectively). Accessory renal arteries were not seen on unenhanced or enhanced phase-contrast MR angiography (0/8 patients) but were detected with gadolinium-enhanced MR angiography in five of the eight patients. Interobserver agreement (kappa = .62) was best with gadolinium-enhanced MR angiography. The quality of the images was unsatisfactory for adequate evaluation of segmental renal arteries with all three MR angiographic techniques. CONCLUSION: A combination of unenhanced phase-contrast MR angiography and gadolinium-enhanced MR angiography at 1.0 T proved useful as a screening protocol for renal artery stenosis.

Adult↗
PubMed1999

Streamlining semantic interpretation for medical narratives.

We introduce two abstraction mechanisms by which the process of semantic interpretation of medical narratives can be simplified and further optimized. One relates to generalized triggering conditions, the other to inheritance-based specifications of semantic rules. The proposed methodology leads to a parsimonious inventory of abstract, simple and domain-independent semantic interpretation schemata whose effectiveness has been evaluated on a medical text corpus.

Evaluation Studies as Topic↗
PubMed1999

Towards a multilingual morpheme thesaurus for medical free-text retrieval.

We introduce a methodology for the segmentation of complex compounds into medically plausible morphemes. A tool for thesaurus compilation and management is presented, and design principles for a multilingual morpheme thesaurus are outlined. Our goal is to enhance the quality of medical free-text retrieval by replacing lexically based through morpheme-based search procedures.

Germany↗
PubMed1998

Beam-position monitors in the X-ray undulator beamline at PETRA.

At the 12 GeV storage ring PETRA, the first synchrotron radiation beamline uses a 4 m-long undulator. The beamline, with a length of 130 m between source and sample, delivers hard X-ray photons usable up to 300 keV. The photon beam has a total power of 7 kW. Combined with the high brilliance, the powerful beam is very critical for all beamline components. Copper, located at a distance of 26 m, hit by the full undulator beam, melts within 20 ms. Different monitors are described for stable, safe and reliable operation of beam and experiments.

Journal Article↗
PubMed1998

Peptide design aided by neural networks: biological activity of artificial signal peptidase I cleavage sites.

De novo designed signal peptidase I cleavage sites were tested for their biological activity in vivo in an Escherichia coli expression and secretion system. The artificial cleavage site sequences were generated by two different computer-based design techniques, a simple statistical method, and a neural network approach. In previous experiments, a neural network was used for feature extraction from a set of known signal peptidase I cleavage sites and served as the fitness function in an evolutionary design cycle leading to idealized cleavage site sequences. The cleavage sites proposed by the two algorithms were active in vivo as predicted. There seems to be an interdependence between several cleavage site features for the constitution of sequences recognized by signal peptidase. It is concluded that neural networks are useful tools for sequence-oriented peptide design.

Amino Acid Sequence↗
PubMed1998

Similarity and rules: distinct? Exhaustive? Empirically distinguishable?

The distinction between rule-based and similarity-based processes in cognition is of fundamental importance for cognitive science, and has been the focus of a large body of empirical research. However, intuitive uses of the distinction are subject to theoretical difficulties and their relation to empirical evidence is not clear. We propose a 'core' distinction between rule- and similarity-based processes, in terms of the way representations of stored information are 'matched' with the representation of a novel item. This explication captures the intuitively clear-cut cases of processes of each type, and resolves apparent problems with the rule/similarity distinction. Moreover, it provides a clear target for assessing the psychological and AI literatures. We show that many lines of psychological evidence are less conclusive than sometimes assumed, but suggest that converging lines of evidence may be persuasive. We then argue that the AI literature suggests that approaches which combine rules and similarity are an important new focus for empirical work.

Cognition↗