Radiology reports: a quantifiable and objective textual approach.

AIM: To examine the feasibility of using automated lexical analysis in conjunction with machine learning to create a means of objectively characterising radiology reports for quality improvement. MATERIALS AND METHODS: Twelve lexical parameters were quantified from the collected reports of four radiologists. These included the number of different words used, number of sentences, reading grade, readability, usage of the passive voice, and lexical metrics of concreteness, ambivalence, complexity, passivity, embellishment, communication and cognition. Each radiologist was statistically compared to the mean of the group for each parameter to determine outlying report characteristics. The reproducibility of these parameters in a given radiologist's reporting style was tested by using only these 12 parameters as input to a neural network designed to establish the authorship of 60 unknown reports. RESULTS: Significant differences in report characteristics were observed between radiologists, quantifying and characterising deviations of individuals from the group reporting style. The 12 metrics employed in a neural network correctly identified the author in each of 60 unknown reports tested, indicating a robust parametric signature. CONCLUSION: Automated and quantifiable methods can be used to analyse reporting style and provide impartial and objective feedback as well as to detect and characterise significant differences from the group. The parameters examined are sufficiently specific to identify the authors of reports and can potentially be useful in quality improvement and residency training.

A pilot study to assess the possible methods of determining the burden of obstructive sleep apnoea syndrome in primary care.

INTRODUCTION: A significant minority of otherwise healthy adults may suffer from disordered breathing during sleep. The commonest problem, known as Obstructive Sleep Apnoea Syndrome (OSAS), results in poor quality sleep, daytime hypersomnolence and excess risk of road traffic crashes. It is also associated with occupational injuries. OSAS can be successfully treated, reducing costs of hospitalisation. There is a gap in the literature regarding the burden of patients with OSAS in primary care, particularly because there is no agreed method for screening. OBJECTIVES: This pilot study was designed to determine the feasibility of screening for OSAS in a UK inner-city multi-ethnic primary care population, to investigate ways in which it might be detected, and to gain an awareness of the potential size of the burden of OSAS METHODS: We tested two methods of screening: firstly, postal questionnaires to a random sample of 240 patients from a subset of the population at highest risk of OSAS (men aged 35-65 years); and secondly, the same questionnaire plus two screening tools for detecting hypersomnolence, given to patients attending the practice for pre-registration health checks. RESULTS: Despite reminders, only 40% of postal questionnaires were returned and there were only 67 (28%) usable responses. The prevalence of snoring was 55%. Almost half of those patients who responded (46%) had a Body Mass Index (BMI) associated with a high risk of OSAS. This was not a positive responder bias because there were no significant differences in BMI between responders and non-responders. 12% had a collar size of greater than 17.5, whilst 34% reported daytime sleepiness, and 24% reported witnessed apnoea. Screening during pre-registration health checks proved both feasible and productive with 38 patients of differing linguistic abilities completing the assessment. CONCLUSION: Screening for OSAS in primary care by means of a postal questionnaire produced a low response. Whilst there was no evidence of a responder bias in terms of BMI, the numbers reporting large collar size, daytime sleepiness and witnessed apnoea suggest that a sizeable proportion of the population is at risk of OSAS. Questionnaires used in this study provide a means of identifying patients with symptoms suggestive of OSAS, although the sensitivity, specificity and positive predictive value of a self-administered instrument needs to be confirmed by a larger study incorporating ventilatory monitoring during sleep. The addition of screening to pre-registration health checks is feasible, but it would take a long time to screen all those at risk for OSAS by this route.

How well can radiologists using neural network software diagnose pulmonary embolism?

OBJECTIVE: This study evaluated and optimized the performance of an automated artificial neural network image interpreter in the diagnosis of pulmonary embolism on ventilation-perfusion lung scans. The computer interpretations were compared with the interpretations of three experienced observers. MATERIALS AND METHODS: Digital data were obtained from 100 patients with normal findings on chest radiographs who were undergoing both radionuclide ventilation-perfusion scanning and pulmonary angiography. Interpretations of differently trained neural networks were compared with those of three experienced nuclear medicine practitioners unaware of the clinical diagnosis. RESULTS: Machines running neural networks performed similarly to experienced scan interpreters in the detection of pulmonary embolism. Both the human observers and the networks performed best in cases with large emboli. Neural network performance was best in the right lung, when the networks were trained using only cases with large emboli and when networks were trained independently in the right and left lungs. The best predictions resulted from a collaborative interpretation incorporating both the human and computer predictions. CONCLUSION: Computers running artificial neural networks using scan data obtained directly from the anterior and posterior ventilation and perfusion images, without human involvement, perform comparably with experienced observers in patients with normal findings on chest radiographs. Human observers can improve their interpretations by incorporating computer output to formulate diagnostic prediction. The method of training the networks is critical to optimizing performance.

Polyclonal, nonspecific 111In-IgG scintigraphy in the evaluation of complicated osteomyelitis and septic arthritis.

BACKGROUND: In this investigation we tested the hypothesis that 111In-IgG scintigraphy can differentiate infectious from sterile inflammatory processes in patients with complicated osteomyelitis or septic arthritis. METHODS: A prospective university hospital based study was performed over 18 months. We studied 31 sites of suspected infection, in 25 adult patients, (age 18 to 74 years, 12 females and 13 males) referred with clinical presentations compatible with complicated osteomyelitis or septic arthritis and in whom proof of the infection was likely to be obtained. The clinical setting in these patients was previous trauma, recent surgery, peripheral vascular disease or adjacent soft tissue infection. Whole body scintigraphy was performed at 1-6, 18-24 and 42-48 hours after administration of 55 MBq of 111In-IgG and results were compared to radiographs, 99mTc-MDP skeletal scintigraphy, biopsy specimens (9 sites) or synovial fluid aspirates (4 sites) and clinical follow-up. RESULTS: Of the 31 sites evaluated, 68% (21/31) were interpreted as negative for abnormal tracer accumulation and 32% (10/31) were considered positive. In patients who underwent biopsy and/or synovial fluid aspiration, 6 of 7 sites were correctly interpreted as positive; sensitivity 86%. Five of 6 sites were correctly interpreted as negative; specificity 83%. When all patients were considered using clinical follow-up in addition to culture results, 9 of 10 sites were correctly interpreted as positive (sensitivity 90%) and 20 of 21 patients were correctly interpreted as negative (specificity 95%). CONCLUSIONS: 111In-IgG scintigraphy is useful for detection of musculoskeletal infection in patients in whom sterile inflammatory events simulate infectious processes.

A relative of the broad-host-range plasmid RSF1010 detected in Erwinia amylovora.

Streptomycin- and sulfonamide-resistant Erwinia amylovora CA3R from California contained an 8.7-kb plasmid, pEa8.7, with a sulII-strA-strB resistance region; furthermore, PCR, sequencing, hybridization, and restriction analyses showed that pEa8.7 was closely related or identical to broad-host-range plasmid RSF1010. Although RSF1010 has been found in a variety of bacteria, this is the first report of its presence in plant pathogenic bacteria.

Interaction of Escherichia coli RecA protein with LexA repressor. I. LexA repressor cleavage is competitive with binding of a secondary DNA molecule.

Essential to the two distinct cellular events of genetic recombination and SOS induction in Escherichia coli, RecA protein promotes the homologous pairing and exchange of DNA strands and the proteolytic cleavage of the LexA repressor, respectively. Since both of these activities require single-stranded DNA (ssDNA) and ATP, the inter-relationship between these reactions was investigated and found to display many parallels. The extent of active complex formed between RecA protein and M13 ssDNA, as measured by both ATP hydrolysis and LexA proteolysis, is stimulated in a similar manner by either a reduction in magnesium ion concentration or the presence of single-stranded DNA binding (SSB) protein. However, unexpectedly, SSB protein inhibits both LexA proteolysis and ATP hydrolysis (in assays containing repressor) at concentrations of RecA protein that are substoichiometric to the ssDNA, arguing that LexA repressor affects the competition between RecA and SSB proteins for limited ssDNA binding sites. Additionally, attenuation of LexA repressor cleavage in the presence of double-stranded DNA or by an excess of ssDNA suggests that interaction of the RecA nucleoprotein filament with either LexA repressor or a secondary DNA molecule is mutually exclusive. The significance of these results is discussed in the context of both the regulation of inducible responses to DNA damage, and the competitive relationship between the processes of SOS induction and genetic recombination.

Impact of fluorodeoxyglucose positron emission tomography on the clinical management of patients with glioma.

The past decade has seen the identification of many clinical settings in the treatment of primary brain tumors in which information from fluorodeoxyglucose positron emission tomography (FDG-PET) might be useful, if not essential, to therapeutic formulation. FDG-PET is currently used at referral centers in the management of primary brain tumors. The clinical pattern of FDG-PET use was assessed and its value compared to other information sources in clinical decision making. The clinical records of 75 glioma patients who were evaluated by FDG-PET were reviewed. The range of circumstances in which FDG-PET was employed included: pretherapeutic baseline studies for monitoring the effect of a therapy (1% of all cases), mapping of hypermetabolic regions before surgery or biopsy (2%), mapping of hypermetabolic regions before radiotherapy (2%), postsurgical evaluation for residual tumor (2%), assessment of the malignancy of a mass as a substitute for biopsy (11%), and distinguishing between radiation necrosis and recurrent tumor (87%). Other sources of information that contributed to the therapeutic management of patients included: gadolinium-enhanced MRI, contrast-CT, and clinical findings.

Pleural effusion and ventilation/perfusion scan interpretation for acute pulmonary embolus.

UNLABELLED: This study was conducted to determine if pleural effusion size affects ventilation/perfusion (V/Q) scan interpretation algorithms for acute pulmonary embolus (PE). METHODS: Retrospective analysis identified 163 consecutive patients undergoing angiography for PE with radiographic evidence for pleural effusion. V/Q scanning was performed in 94 (58%) of cases and reported using original Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) criteria. Effusions were classified as small, large and/or bilateral. Radiographic and scintigraphic results were compared with regard to size and location of abnormalities. RESULTS: Of the 163 patients, 57 (35%) had angiographically-proven PE, 77 (47%) had at least one large pleural effusion and 86 (53%) had a small effusion; 33 (43%) with large effusions and 24 (28%) with small effusions had emboli at angiography. Thirty-six of 119 patients (30%) with clear chest radiographs (a control group) had PE. Thus, large effusions were associated with a higher incidence of PE than those with small effusions or clear lungs (p < 0.05). Of those with V/Q scanning, 26 of 94 (28%) had a solitary large effusion, with 12 (46%) positive for emboli. V/Q-matched abnormalities limited to effusion size were found in 16 with a solitary large effusion and 10 with a solitary small effusion. In both groups, 50% were angiographically positive for emboli. Twenty-three (66%) of 35 with bilateral effusions had corresponding V/Q-matched defects at one (n = 11) or both (n = 12) lung bases, and 9 (39%) were positive for emboli. In total, 45% with a V/Q-matched defect of equivalent size to the effusion were angiographically positive for PE. CONCLUSION: Pulmonary emboli are associated with pleural effusions of all sizes. Matched V/Q defects corresponding to radiographically-evident pleural effusions are of intermediate probability for PE. Thus, revision of the traditional lung scan interpretive criteria based upon pleural effusion size is not warranted.

Neural networks in ventilation-perfusion imaging.

PURPOSE: To optimize the performance of artificial neural networks in the prediction of pulmonary embolism from ventilation-perfusion (V-P) scans. MATERIALS AND METHODS: Neural networks were constructed with a set of V-P scan criteria that included sharpness and completeness of perfusion defects and involved quantification of abnormalities by using a continuous numeric scale. Several network parameters were systematically varied. Networks were trained with 150 cases and tested with 30 different cases. Findings were compared with those of pulmonary angiography. RESULTS: Networks capable of performing as well as experienced nuclear medicine physicians could be constructed with few V-P scan features. A brief training period was optimal (50-100 iterations). Further training diminished network performance. CONCLUSION: Effective neural networks can be constructed by using a limited number of unconventional V-P scan features. Several parameters can be adjusted to optimize performance.

Neural networks in ventilation-perfusion imaging. Part II. Effects of interpretive variability.

PURPOSE: To evaluate the usefulness of a neural network developed by one physician and used by another. MATERIALS AND METHODS: Intra- and interobserver variability were analyzed in image categorization of ventilation-perfusion (V-P) scans. This information was used to estimate network performance when it was used by a physician who did not train the network. RESULTS: Network training was optimized by using input parameters that demonstrated both individually high correlations with pulmonary embolism and good reproducibility in multiple interpretations. CONCLUSION: Potential variability exists in the performance of a network when it is supplied with input data by different physicians. The clinical usefulness of a network depends heavily on the similarity of interpretive styles between the network trainer and the user.

X-ray diffraction analysis of crystals containing twofold symmetric nucleosome core particles.

Nucleosome core particles containing a DNA palindrome and purified chicken erythrocyte histone octamer have been reconstituted and crystallized. The dyad symmetry of the palindrome extends the dyad symmetry of the histone octamer to result in a twofold symmetric particle. This ensures that the structure determined by X-ray diffraction will yield a true representation of the DNA strand rather than the twofold averaged structure which would result from using a non-palindromic DNA sequence. The crystals provide isotropic diffraction to 3.2 A with observed reflections extending to d spacings of about 2.8 A using a rotating-anode Cu Kalpha X-ray source. Although the DNA palindrome is a factor contributing to the quality of the diffraction data, another significant factor is an improved preparative technique which enriches for correctly phased nucleosome core particles.

Large-scale production of palindrome DNA fragments.

Our structural studies of nucleosomes necessitated the production of over 100 mg of a 146-bp perfect palindrome DNA for use in the reconstitution of perfectly symmetrical nucleosome core particles for detailed X-ray crystallographic analysis. The propagation of palindromic DNA DNA sequences by bacterial culture is hindered by the instability of these sequences during bacterial replication and recombination. While the loss of some palindrome sequences can be eliminated by the use of sbcB or sbcC mutants of Escherichia coli, not all palindrome-containing plasmids are faithfully maintained by these strains. The production of large quantities of palindrome DNA that involves production of plasmid containing multiple copies of the repeating unit of the palindrome which are isolated by restriction digestion and ligated in vitro to form the palindrome DNA. The procedure has resulted in the production of over 20 mg of a 146-bp DNA fragment in 2 weeks.

Preparative separation of nucleosome core particles containing defined-sequence DNA in multiple translational phases.

The nucleosome core particle is composed of an octamer of core histone proteins and about 146 bp of DNA. When reconstituted from purified histone octamer and defined-sequence, nucleosome positioning DNA fragments, the DNA will bind to the histone core in a number of translational phases with respect to the dyad symmetry axis of the histone octamer. Only one of these phases contains symmetrically bound DNA, and it is this species which is required for crystallization and X-ray diffraction studies. We have developed a technique for separating nucleosome core particles, containing defined-sequence 146 bp DNA, which differ only in translational phasing of the DNA with respect to the histone octamer core.

Pulmonary embolism: prediction of the usefulness of initial ventilation-perfusion scanning with chest radiographic findings.

PURPOSE: To evaluate the usefulness of ventilation-perfusion scanning when chest radiographic findings are abnormal to decide which patients should initially undergo angiography because of suspected pulmonary embolism. MATERIALS AND METHODS: The records of 951 patients who underwent examination between April 1, 1992, and August 24, 1993, were reviewed, and the results of ventilation-perfusion scanning and chest radiography were correlated. RESULTS: Approximately 8% of patients with radiographic findings of no acute disease, pleural effusion, and linear atelectasis had high-probability ventilation-perfusion scans. Pulmonary edema and parenchymal consolidation categories had fewer high-probability studies (3% and 1%, respectively; P < .01). The chance of obtaining a nondiagnostic result was similar for the first three categories (12%) but was more likely for pleural effusion and parenchymal consolidation (36% and 82%, respectively; P < .01). CONCLUSION: The results of ventilation-perfusion scanning are not useful with focal radiographic consolidation. It may be appropriate to proceed directly to pulmonary angiography in many of these patients.

The integrin alpha 9 beta 1 mediates cell attachment to a non-RGD site in the third fibronectin type III repeat of tenascin.

We have previously reported the sequence of the integrin alpha 9 subunit, a partner of the beta 1 subunit that is expressed in basal keratinocytes, hepatocytes, airway epithelial cells, and smooth and skeletal muscle. In the present study, we have stably expressed alpha 9 beta 1 on the surface of the human embryonic kidney cell line 293 and the human colon carcinoma cell line SW480 and used these transfected cells lines to identify ligand(s) for this integrin. Transfected cells did not appear to utilize alpha 9 beta 1 for attachment to the extracellular matrix proteins fibronectin, laminin, vitronectin, fibrinogen, thrombospondin, or type I or IV collagen. However, in contrast to mock transfectants, both 293 cells and SW480 cells expressing alpha 9 beta 1 adhered to intact chicken tenascin. By utilizing a variety of recombinant fragments of tenascin, we were able to localize the binding site for alpha 9 beta 1 to the third type III repeat. This repeat contains the arginine-glycine-aspartic acid (RGD) tripeptide that has been shown to serve as a binding site in tenascin for alpha v-integrins. However, the RGD site does not appear to be the binding site for alpha 9 beta 1, as the attachment of alpha 9 transfectants to this fragment was not inhibited by RGD peptide, nor by changing the RGD site to RAD or RAA.

Sequence and tissue distribution of the integrin alpha 9 subunit, a novel partner of beta 1 that is widely distributed in epithelia and muscle.

The integrin family of adhesion receptors consists of several heterodimeric glycoproteins, each composed of one alpha and one beta subunit. A novel integrin alpha subunit partial cDNA isolated from TGF-beta stimulated guinea pig airway epithelial cells has previously been reported (Erle, D.J., D. Sheppard, J. Bruess, C. Rüegg, and R. Pytela. 1991. Am. J. Respir. Cell Mol. Biol. 5:170-177). We have now determined cDNA and amino acid sequence for the human homolog of this subunit, named alpha 9, from a human lung cDNA library, a human small intestine cDNA library, and cDNA from the cell lines U937, HL-60 and Tera-2. This sequence is predicted to encode a 1006-amino acid mature protein that shares 39% identity with the previously identified integrin subunit alpha 4. By Northern blot analysis, alpha 9 mRNA was detected in the human carcinoma cell lines Tera-2 and Caco-2. Anti-peptide antibodies against the predicted COOH-terminal sequence of alpha 9 immunoprecipitated a heterodimer (140 kD/115 kD nonreduced; 150 kD/130 kD reduced) from Tera-2 lysates. Immunodepletion of beta 1-containing integrins with Tera-2 lysates removed alpha 9 immunoreactivity, suggesting that beta 1 is the principal beta subunit partner for alpha 9 in these cells. alpha 9 was detected by immunohistochemistry in airway epithelium, in the basal layer of squamous epithelium, and in smooth muscle, skeletal muscle, and hepatocytes.