Radiologists' vs. surgeons' reading performance in non traumatic acute abdomen CTs : P32

Purpose: Emergency room reading performances have been a point of interest in recent studies comparing radiologists to other physician groups. Our objective was to evaluate and compare the reading performances of radiologists and surgeons in an emergency room setting of non-traumatic abdominal CTs. Methods and materials: A total of ten readers representing four groups participated in this study: three senior radiologists and visceral surgeons, respectively, and two junior radiologists and surgeons, respectively. Each observer blindedly evaluated a total of 150 multi-slice acute abdominal CTs. CTs were chosen representing established proportions of acute abdomen pathologies in a Level I trauma centre from 2003 to 2005. Each answer was interpretated as right or wrong regarding pathology location, diagnosis and need for operation. Gold standard was the intraoperative result, and the clinical patient follow-up for non-operated patients. Significance was assumed at a p <.05 level. Results: Senior radiologists had a mean score of 2.38 ± 1.14, junior radiologists a score of 2.34 ± 1.14, whereas senior surgeons scored 2.07 ± 1.30 and junior surgeons 1.62 ± 1.42. No significant difference was found between the two radiologist groups, but results were significantly better for senior surgeons as compared to junior surgeons and better for the two radiologist groups as compared to each of the surgeon groups (all p <.05). Conclusion: Abdominal CT reading in an acute abdomen setting should continue to rely on an evaluation by a radiologist, whether senior or junior. Satisfying reading results can be achieved by senior visceral surgeons, but junior surgeons need more experience for a good reading performance.

A Survey of Active Learning Algorithms for Supervised Remote Sensing Image Classification

Defining an efficient training set is one of the most delicate phases for the success of remote sensing image classification routines. The complexity of the problem, the limited temporal and financial resources, as well as the high intraclass variance can make an algorithm fail if it is trained with a suboptimal dataset. Active learning aims at building efficient training sets by iteratively improving the model performance through sampling. A user-defined heuristic ranks the unlabeled pixels according to a function of the uncertainty of their class membership and then the user is asked to provide labels for the most uncertain pixels. This paper reviews and tests the main families of active learning algorithms: committee, large margin, and posterior probability-based. For each of them, the most recent advances in the remote sensing community are discussed and some heuristics are detailed and tested. Several challenging remote sensing scenarios are considered, including very high spatial resolution and hyperspectral image classification. Finally, guidelines for choosing the good architecture are provided for new and/or unexperienced user.

Structured output SVM for remote sensing image classification

In the recent years, kernel methods have revealed very powerful tools in many application domains in general and in remote sensing image classification in particular. The special characteristics of remote sensing images (high dimension, few labeled samples and different noise sources) are efficiently dealt with kernel machines. In this paper, we propose the use of structured output learning to improve remote sensing image classification based on kernels. Structured output learning is concerned with the design of machine learning algorithms that not only implement input-output mapping, but also take into account the relations between output labels, thus generalizing unstructured kernel methods. We analyze the framework and introduce it to the remote sensing community. Output similarity is here encoded into SVM classifiers by modifying the model loss function and the kernel function either independently or jointly. Experiments on a very high resolution (VHR) image classification problem shows promising results and opens a wide field of research with structured output kernel methods.

Active mitral ring for post-surgical remote correction of residual mitral regurgitation on the beating heart.

OBJECTIVES: Residual mitral regurgitation after valve repair worsens patients' clinical outcome. Postimplant adjustable mitral rings potentially address this issue, allowing the reshaping of the annulus on the beating heart under echocardiography control. We developed an original mitral ring allowing valve geometry remodelling after the implantation and designed an animal study to assess device effectiveness in correcting residual mitral regurgitation. METHODS: The device consists of two concentric rings: one internal and flexible, sutured to the mitral annulus and a second external and rigid. A third conic element slides between the two rings, modifying the shape of the flexible ring. This sliding element is remotely activated with a rotating tool. Animal model: in adult swine, under cardio pulmonary bypass and cardiac arrest, we shortened the primary chordae of P2 segment to reproduce Type III regurgitation and implanted the active ring. We used intracardiac ultrasound to assess mitral regurgitation and the efficacy of the active ring to correct it. RESULTS: Severe mitral regurgitation (3+ and 4+) was induced in eight animals, 54 ± 6 kg in weight. Vena contracta width decreased from 0.8 ± 0.2 to 0.1 cm; proximal isovelocity surface area radius decreased from 0.8 ± 0.2 to 0.1 cm and effective regurgitant orifice area decreased from 0.50 ± 0.1 to 0.1 ± 0.1 cm(2). Six animals had a reversal of systolic pulmonary flow that normalized following the activation of the device. All corrections were reversible. CONCLUSIONS: Postimplant adjustable mitral ring corrects severe mitral regurgitation through the reversible modification of the annulus geometry on the beating heart. It addresses the frequent and morbid issue of recurrent mitral valve regurgitation.

Inhibition of mouse mammary tumor virus-induced T cell responses in vivo by antibodies to an open reading frame protein.

Minor lymphocyte stimulating (Mls) antigens specifically stimulate T cell responses that are restricted to particular T cell receptor (TCR) beta chain variable domains. The Mls phenotype is genetically controlled by an open reading frame (orf) located in the 3' long terminal repeat of mouse mammary tumor virus (MMTV); however, the mechanism of action of the orf gene product is unknown. Whereas predicted orf amino acid sequences show strong overall homology, the 20-30 COOH-terminal residues are strikingly polymorphic. This polymorphic region correlates with TCR V beta specificity. We have generated monoclonal antibodies to a synthetic peptide encompassing the 19 COOH-terminal amino acid residues of Mtv-7 orf, which encodes the Mls-1a determinant. We show here that these antibodies block Mls responses in vitro and can interfere specifically with thymic clonal deletion of Mls-1a reactive V beta 6+ T cells in neonatal mice. Furthermore, the antibodies can inhibit V beta 6+ T cell responses in vivo to an infectious MMTV that shares orf sequence homology and TCR specificity with Mtv-7. These results confirm the predicted extracellular localization of the orf COOH terminus and imply that the orf proteins of both endogenous and exogenous MMTV interact directly with TCR V beta.

Influence on screening performance of second reading strategies in two lowvolume programs

BACKGROUND: The European Guidelines specify a minimum of 5,000 screening cases to be read yearly by radiologists carrying out second reading in non-centralized programs. This professional requirement is difficult to reach and/or to implement in regional programs covering a sparse population with a high number of participating radiology units, so that alternative blind double reading strategies must be devised. OBJECTIVE: To evaluate the effect on breast cancer screening performances of two second reading strategies used in non-centralized, low-volume programs. METHODS: Reading performances in two Swiss regional breast cancer screening programs (cantons of Wallis and Vaud), covering female populations, aged 50-69, of about 31'000 and 72'000 inhabitants were computed and compared. Both programs had similar screening regimens and organizations, but differed with respect to second reading. One setting applied a selective strategy whereby only experienced radiologists performed second reading; the other elicited not to restrict second readers on the basis of their individual screening activity. Analysis included some 140,000 mammograms performed between 1999 and 2005. RESULTS: Overall, screening performances improved with increasing total volume of reading, albeit not in a linear fashion. Regardless of setting, radiologists attained a higher level of screening accuracy when performing second rather than first readings, and incident rather than prevalent screening cases. The effect of a selective, small group of second readers appeared to impact favorably on the false-positive rate and other indicators of screening quality. As the learning curve depends on the number of mammograms read, these distinct strategies may bear different outcome in the long run. Implications and practical issues for low-volume programs are discussed.

Self-expanding mini-cannula for remote perfusion with pediatric scenarios.

The aim of this report is to address the benefits of the minimal invasive venous drainage in a pediatric cardio surgical scenario. Juvenile bovine experiments (67.4+/-11 kg) were performed. The right atrium was cannulated in a trans-jugular way by using the self-expandable (Smart Stat, 12/20F, 430 mm) venous cannula (Smartcannula LLC, Lausanne, Switzerland) vs. a 14F 250 mm (Polystan Lighthouse) standard pediatric venous cannula. Establishing the cardiopulmonary bypass (CPB), the blood flows were assessed for 20 mmHg, 30 mmHg and 40 mmHg of driving pressure. Venous drainage (flow in l/min) at 20 mmHg, 30 mmHg, and 40 mmHg drainage load was 0.26+/-0.1, 0.35+/-0.2 and 0.28+/-0.08 for the 14F standard vs. 1.31+/-0.22, 1.35+/-0.24 and 1.9+/-0.2 for the Smart Stat 12/20F cannula. The 43 cm self-expanding 12/20F Smartcannula outperforms the 14F standard cannula. The results described herein allow us to conclude that usage of the self-expanding Smartcannula also in the pediatric patients improves the flow and the drainage capacity, avoiding the insufficient and excessive drainage. We believe that similar results may be expected in the clinical settings.