Natural Language Processing Techniques for Extracting and Categorizing Finding Measurements in Narrative Radiology Reports.

BACKGROUND: Accumulating quantitative outcome parameters may contribute to constructing a healthcare organization in which outcomes of clinical procedures are reproducible and predictable. In imaging studies, measurements are the principal category of quantitative para meters. OBJECTIVES: The purpose of this work is to develop and evaluate two natural language processing engines that extract finding and organ measurements from narrative radiology reports and to categorize extracted measurements by their "temporality". METHODS: The measurement extraction engine is developed as a set of regular expressions. The engine was evaluated against a manually created ground truth. Automated categorization of measurement temporality is defined as a machine learning problem. A ground truth was manually developed based on a corpus of radiology reports. A maximum entropy model was created using features that characterize the measurement itself and its narrative context. The model was evaluated in a ten-fold cross validation protocol. RESULTS: The measurement extraction engine has precision 0.994 and recall 0.991. Accuracy of the measurement classification engine is 0.960. CONCLUSIONS: The work contributes to machine understanding of radiology reports and may find application in software applications that process medical data.

The use of a reference tissue arterial input function with low-temporal-resolution DCE-MRI data.

Pharmacokinetic modeling is a promising quantitative analysis technique for cancer diagnosis. However, diagnostic dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) of the breast is commonly performed with low temporal resolution. This limits its clinical utility. We investigated for a range of temporal resolutions whether pharmacokinetic parameter estimation is impacted by the use of data-derived arterial input functions (AIFs), obtained via analysis of dynamic data from a reference tissue, as opposed to the use of a standard AIF, often obtained from the literature. We hypothesized that the first method allows the use of data at lower temporal resolutions than the second method. Test data were obtained by downsampling high-temporal-resolution rodent data via a k-space-based strategy. To fit the basic Tofts model, either the data-derived or the standard AIF was used. The resulting estimates of K(trans) and v(e) were compared with the standard estimates obtained by using the original data. The deviations in K(trans) and v(e), introduced when lowering temporal resolution, were more modest using data-derived AIFs compared with using a standard AIF. Specifically, lowering the resolution from 5 to 60 s, the respective changes in K(trans) were 2% (non-significant) and 18% (significant). Extracting the AIF from a reference tissue enables accurate pharmacokinetic parameter estimation for low-temporal-resolution data.

Interactive image-guided neuroendoscopy: development and early clinical experience.

Technical advances and pioneering surgeons have established neuroendoscopy as an accepted diagnostic and therapeutic tool. The clinical indications for endoscopy, variety of operative techniques and number of endoscopic surgeons continue to increase steadily. However, there are fundamental limits to the scope of freehand endoscopy principally governed by the need for direct vision of anatomical and pathological structures. In addition, whilst the expert neuroendoscopist is only occasionally disorientated by complex distorted anatomy, the rising number of novices are likely to be mislead relatively often. We report the integration of neuroendoscopy with an optical neuronavigation system to provide interactive image-guided neuroendoscopy. This combination both removes the constraining requirement for direct vision and provides accurate localisation to guide the surgeon during surgery. We describe the clinical application of this method to two cases where image-guided endoscopy was essential to the safe completion of the procedure.

The EASI project--improving the effectiveness and quality of image-guided surgery.

In recent years, advances in computer technology and a significant increase in the accuracy of medical imaging have made it possible to develop systems that can assist the clinician in diagnosis, planning, and treatment. This paper deals with an area that is generally referred to as computer-assisted surgery, image-directed surgery, or image-guided surgery. We report the research, development, and clinical validation performed since January 1996 in the European Applications in Surgical Interventions (EASI) project, which is funded by the European Commission in their "4th Framework Telematics Applications for Health" program. The goal of this project is the improvement of the effectiveness and quality of image-guided neurosurgery of the brain and image-guided vascular surgery of abdominal aortic aneurysms, while at the same time reducing patient risks and overall cost. We have developed advanced prototype systems for preoperative surgical planning and intraoperative surgical navigation, and we have extensively clinically validated these systems. The prototype systems and the clinical validation results are described in this paper.

Clinical introduction of an adjustable rigid instrument holder for frameless stereotactic interventions.

Interactive image guidance is now in routine use for open neurosurgical procedures and has demonstrated patient benefits. However, freehand interactive guidance is not an appropriate replacement for the traditional frame-based stereotactic procedures of biopsy, electrode placement, and functional lesioning. These point-based procedures require precise target localization and direct instrument guidance to avoid collateral brain injury. To perform true frameless stereotactic procedures requires a guide that is also adjustable for positioning, lockable, and adaptable to multiple instruments. We describe such a device, which is employed for the guidance of biopsy needles, shunts, electrodes, and endoscopes during neuronavigation. The method of frameless stereotactic biopsy retrieval with an infrared-based neuronavigation system is described, clinical results are given, and further areas of application discussed.

Evaluation of optimal age at first conception in gilts from data collected in commercial swine herds.

The purpose of this study was to find the economically optimal period of first conception in gilts, addressing the issues of lifetime reproductive performance and expected herd life. A profit equation was used to combine the effects into one economic parameter. The data were from 14,910 gilts on 54 farms throughout The Netherlands. The average number of pigs born alive in the first litter increased with older age at conception. In the second litter a similar, but much smaller, effect was observed. Age at first conception had no effect on number of pigs born alive in the third or greater litter. Gilts bred at an older age had a shorter expected herd life than gilts bred at a younger age. Evaluating the reasons for culling revealed that infertility became a more important reason with increasing age at first conception. The proportion culled for infertility increased linearly from 18% at conception on d 200 to 24.5% at conception on d 300. Combining the effect of litter size and herd life led to the conclusion that the profit per gilt (sow) was not significantly affected by her age at first conception. It is concluded that the optimal economic age at first conception was considered to be approximately 200 to 220 d of age when the cost of housing and feed of the gilt from entry to first conception were taken into account.

Population dynamics of bulk milk somatic cell counts.

The population dynamics for bulk milk SCC in 27,000 Dutch dairy farms was studied over 6 yr; SCC declined markedly in 1984 and 1985. This coincided with the introduction of the quota system in the European Community. The CV of SCC was .36, ranging from .15 to .89. Month of year had a significant effect on SCC with the highest SCC in October and the lowest in April. The dynamics of the annual SCC were modeled with a state transition model. Farmers managing herds with a fairly low SCC (less than 450,000) had a reasonable chance to remain in these states. Herds with a higher annual SCC were much more variable. The model simulated, the potential reactions of the farmers to the decrease of the SCC limit from 750,000 to 500,000 starting January 1, 1989. The simulations showed a further decrease in mean annual SCC, although the number of farms in the category with a very low SCC (less than 150,000) is not likely to increase. In all simulations an increased number of farmers will receive a penalty due to the more strict regulations of the European Community.

VAMPP: a veterinary automated management and production control program for swine breeding farms. I. Main features.

To improve the herd health and production management on swine breeding farms the Department of Herd Health and Ambulatory Clinic (University of Utrecht) developed a computer package. The basic package includes action lists, physical performance reports, analysis and economic programs. During the development of the software package it was emphasised that the programs should be user friendly (easy to handle, with no special codes), and should have an enormous flexibility to guarantee easy updating and a long life for the whole package. The Standard MUMPS system, which offers a unique data base structure and access method, was thus employed. The software runs on a central mini-computer with telephone lines for farmers and veterinary surgeons. In addition, the 'post-in post-out' method can be used; a micro-computer version of the package also is available.

VAMPP: a veterinary automated management and production control program for swine breeding farms. II. Output.

The Department of Herd Health and Ambulatory Clinic of the Veterinary Faculty (State University of Utrecht, The Netherlands) has developed the VAMPP package for swine breeding farms. The present article is the second of two articles describing the VAMPP-system. The objective of this computer package is to improve productivity on swine breeding farms through automated data processing and interpretation and action based on the computer output. The output produced by the VAMPP-programs is described in the present article and consists of: action lists, physical performance reports, analysis graphs and tables and economic reports.

The cycle index, an alternative to the farrowing index converted to a yearly basis.

The farrowing index is an important parameter for the evaluation of sow productivity. The value of the farrowing index calculated over a relatively short period of time and converted to a yearly basis, depends largely on the number of litters produced during this period. Random differences in the numbers of litters produced in subsequent periods will strongly influence the farrowing indices of these periods. The cycle index is designed to obtain a more valuable and consistent parameter. The cycle index is based on reproductive data of sows that farrowed during a certain period and the number of production days lost because of culling of sows during that period.

Index Lists provided by the VAMPP computer programme for dairy herd health and production control. III. Interpretation and analysis of data on Index List 3.

In a series of four papers computerized herd reports for dairy herd health and production control purposes, the Index Lists, are presented. This third paper deals with Index List 3 on youngstock rearing and cow culling. The information from this list is suitable for monitoring rearing efficiency and culling policy in a veterinary herd health and production control programme, especially when used in combination with information from the other index lists. Methods for interpretation and evaluation of the data are provided. It is demonstrated how prompt computerized analysis of current information is used to detect deviations in index figures and to advise the farmer.

Index Lists provided by the VAMPP computer programme for dairy herd health and production control. IV. Interpretation and analysis of data on Index List 4.

In a series of 4 papers computerized herd reports on various aspects of dairy farming, the Index Lists, are presented for herd health and production control purposes. This fourth paper regards Index List 4 on nutritional disorders, body condition and ration composition. The information from this list is suitable to monitor nutritional efficiency and feeding management, especially when used in combination with information from the other index lists. Methods for interpretation and evaluation of the data listed are provided. Evaluations should take into account the aspects related to nutrition. It is demonstrated how deviations of index figures are detected and how advice to farm management is built up by means of actual information and prompt computerized analysis.

Index Lists provided by the VAMPP computer programme for dairy herd health and production control. I. Interpretation and analysis of data on Index List 1.

In a series of 4 papers the Index Lists as provided by the VAMPP computer programme are presented. The information on Index List 1, covering milk production, milk quality and udder health, is suitable to monitor actual herd performance, especially when used in connection with the other index lists. With computer programmes becoming available to veterinary control programmes, proper interpretation and prompt analysis of data is more critical. This paper outlines how deviations of index figures from reference values are detected, how analysis may be performed and how advice is built-up.

Index Lists provided by the VAMPP computer programme for dairy herd health and production control. II. Interpretation and analysis of data on Index List 2.

In a series of four papers computerized herd reports on various aspects of dairy farming, the Index Lists, are presented for herd health and production control purposes. This second paper presents Index List 2, on calving, breeding efficiency and lameness. The information from this list is suitable to monitor herd performance, especially when used in combination with information from other index lists. Methods for interpretation and evaluation of the data listed are provided. Evaluations should take those aspects related to reproduction and lameness into account. The use of current information and prompt computerized analysis in detecting deviations in index figures and providing advice to farm management is discussed.

VAMPP: a veterinary automated management and production control programme for dairy farms (the application of MUMPS for data processing).

The application of MUMPS in a computerised recording system for herd health and production control on dairy farms is reviewed. MUMPS is an interactive multi-user database management system, which is both an operating system and a high level computer language. In this system, coding of veterinary and management events prior to data entry is not needed. Programmes and data structure can easily be adapted and extended due to the features of MUMPS. The system for dairy farms allows epidemiological analyses, due to the flexibility of the programme. The programme is used by farmers and veterinary surgeons by means of terminals linked to a central computer. The system provides action lists for farmers and veterinary surgeons; the information on these lists is presented in a multidisciplinary way. Several herd reports and analyses, including frequency distributions and graphs, are given. These reports enable the investigation of cross-relations between farm aspects, and aid in the detection of problem areas.