Initial evaluation of a continuous speech recognition program for radiology

The aims of this work were to measure the accuracy of one continuous speech recognition product and dependence on the speaker's gender and status as a native or nonnative English speaker, and evaluate the product's potential for routine use in transcribing radiology reports. IBM MedSpeak/Radiology software, version 1.1 was evaluated by 6 speakers. Two were nonnative English speakers, and 3 were men. Each speaker dictated a set of 12 reports. The reports included neurologic and body imaging examinations performed with 6 different modalities. The dictated and original report texts were compared, and error rates for overall, significant, and subtle significant errors were computed. Error rate dependence on modality, native English speaker status, and gender were evaluated by performing ttests. The overall error rate was 10.3 +/- 3.3%. No difference in accuracy between men and women was found; however, significant differences were seen for overall and significant errors when comparing native and nonnative English speakers (P = .009 and P = .008, respectively). The speech recognition software is approximately 90% accurate, and while practical implementation issues (rather than accuracy) currently limit routine use of this product throughout a radiology practice, application in niche areas such as the emergency room currently is being pursued. This methodology provides a convenient way to compare the initial accuracy of different speech recognition products, and changes in accuracy over time, in a detailed and sensitive manner.     

Neurodevelopmental outcome in high-risk preterm infants treated with inhaled nitric oxide

Inhaled nitric oxide (iNO) is used to treat preterm infants with hypoxaemic respiratory failure. In this study we describe the long-term survival and neurodevelopmental status of high-risk preterm infants enrolled into a randomized controlled trial of iNO therapy. Information regarding long-term outcome was available for all 25 children enrolled in the original trial who survived until discharge from hospital. Formal, blinded, developmental assessment and neurological examinations were performed in 21 out of 22 children still alive at 30 mo of age, corrected for prematurity. No significant differences were found in long-term mortality (12/20 vs 8/22, RR 1.65, 95% CI 0.87-3.3), neurodevelopmental delay (4/7 vs 9/14, RR 0.89, 95% CI 0.37-1.75), severe neurodisability (0/7 vs 5/14, p = 0.12) or cerebral palsy (0/7 vs 2/14, p = 0.53) between iNO-treated and control infants. CONCLUSION: In this study there was no evidence of a significant effect on either survival or long-term neurodevelopmental status in infants treated with iNO.     

Use of the Complete Rockall Score and the Forrest Classification to Assess Outcome in Patients with Non-variceal Upper Gastrointestinal Bleeding Subject to After-hours Endoscopy: A Retrospective Cohort Study

Objectives:

To evaluate the usefulness of the Forrest classification and the complete Rockall score with customary cut-off values for assessing the risk of adverse events in patients with upper gastrointestinal bleeding (UGI-B) subject to after-hours emergency oesophago-gastro-duodenoscopy (E-EGD) within six hours after admission.

Methods:

The medical records of patients with non-variceal UGI-B proven by after-hours endoscopy were analysed. For ''high risk'' situations (Forrest stage Ia–IIb/complete Rockall score > 2), univariate analysis was conducted to evaluate odds ratio for reaching the study endpoints (30-day and one-year mortality, re-bleeding, hospital stay ≥ 3 days).

Results:

During the study period (75 months), 86 cases (85 patients) met the inclusion criteria. Patients ''age was 66.36 ± 14.38 years; 60.5% were male. Mean duration of hospital stay was 15.21 ± 19.24 days. Mortality rate was 16.7% (30 days) and 32.9% (one year); 14% of patients re-bled. Univariate analysis of post-endoscopic Rockall score ≥ 2 showed an odds ratio of 6.09 for death within 30 days (p = 0.04). No other significant correlations were found.

Conclusion:

In patients with UGI-B subject to after-hours endoscopy, a ''high-risk'' Rockall score permits an estimation of the risk of death within 30 days but not of re-bleeding. A ''high-risk '' Forrest score is not significantly associated with the study endpoints.     

Association of granulocyte-macrophage colony-stimulating factor with the crystalloid granules of human eosinophils

We have previously shown that normal-density human peripheral blood eosinophils transcribe and translate mRNA for granulocyte-macrophage colony-stimulating factor (GM-CSF) and that the intracellular distribution was granular as assessed by light microscopy immunocytochemistry. The present study was conducted to confirm this apparent association between GM-CSF and the crystalloid granule using a subcellular fractionation method for human eosinophils and immunogold electron microscopy (EM). Highly purified (> 99%, by negative selection using anti-CD16 immunomagnetic microbeads) human peripheral blood eosinophils were obtained from four asthmatic subjects (not taking systemic medication), homogenized and density fractionated (5 x 10(7) cells/subject) on linear Nycodenz gradients. Twenty-four fractions were collected from each cell preparation and analyzed for marker enzyme activities as well as total protein. Dot blot analysis with specific monoclonal antibodies (MoAbs) was used to detect the eosinophil granule proteins major basic protein (MBP) and eosinophil cationic protein (ECP). An anti-CD9 MoAb was used as an eosinophil plasma membrane marker. Lactate dehydrogenase (LDH) was used as a cytosolic marker. Immunoreactivity for GM-CSF was detected by a specific enzyme-linked immunosorbent assay using a polyclonal antihuman GM-CSF antibody and confirmed by dot blot. GM-CSF coeluted with the cellular fractions containing granule markers (MBP, ECP, eosinophil peroxidase, hexosaminidase, and arylsulphatase), but not those containing cytoplasm (LDH+) or membrane (CD9+) markers. EM examination of pooled fractions associated with the peak of GM-CSF immunoreactivity confirmed that they contained crystalloid and small granules, but not plasma membrane. In addition, quantification, using immunogold labeling with an anti/GM-CSF MoAb, indicated preferential localization of gold particles over the eosinophil granule cores of intact cells. Thus, our results indicate that GM-CSF resides as a granule-associated, stored mediator in unstimulated human eosinophils.     

Surviving the 'heartsink' experience

The authors describe a pilot workshop designed to help doctorsachieve a greater understanding of and ability to cope withtheir ‘heartsink’ patients. Participants were askedto list their personal objectives in attending and a numberof cases were discussed in the group. A ‘heartsink survivalkit’ was provided which consisted of skills and strategieswhich are useful in difficult consultations and an approachto reassessing the goals of the relationship which might promotea more realistic understanding. The workshop was videotapedand two scenarios are presented. The implications for furthertraining are discussed.     

Primitive neuroectodermal tumours of the cerebrum

Summary Eighteen cases of cerebral tumour composed partly or totally of primitive embryonal cells are reported. These lesions comprise 2.8% of all primary cerebral hemisphere tumours in the histopathology files of The Royal Marsden Hospital between 1971 and 1980 inclusive. Most exhibited some degree of differentiation towards neuronal or glial elements and, as more than one type of differentiation was often present in the same lesion, we agree with others that the term primitive neuroectodermal tumour (PNET) is more appropriate to describe these lesions than terms based on histogenesis. The extent of the primitive component varied, but usually accounted for more than 80% of the tumour. Although the tumours bear some similarities to posterior fossa medulloblastomas, they exhibit important differences in histology, immunohistology, natural history and response to treatment. Nearly all PNETs examined expressed some glial fibrillary acidic (GFAP) both in primitive areas and zones of astrocytic differentiation. GFAP staining may thus be of value in distinguishing PNETs from undifferentiated non-neurogenic tumours. Of 14 patients referred for radiotherapy, the survival rate at 3 years was 29% (\414) and 5 years 25% (\312). Patients with tumours in which at least 90% of the tissue was undifferentiated exhibited an extremely poor prognosis with none of 9 patients still alive at 3 years in contrast to 3 of 5 patients (60%) with tumours showing less than 90% undifferentiation. Radical tumour removal, where feasible, followed by irradiation of the whole cerebrospinal axis is recommended. Adjuvant chemotherapy with such agents as CCNU and Vincristine may be of value: the 3 long term survivors in the present series (7–11 years), including one who presented disseminated intracranial disease, received such adjuvant treatment.