Contrastive domain adaptation with consistency match for automated pneumonia diagnosis.

Pneumonia can be difficult to diagnose since its symptoms are too variable, and the radiographic signs are often very similar to those seen in other illnesses such as a cold or influenza. Deep neural networks have shown promising performance in automated pneumonia diagnosis using chest X-ray radiography, allowing mass screening and early intervention to reduce the severe cases and death toll. However, they usually require many well-labelled chest X-ray images for training to achieve high diagnostic accuracy. To reduce the need for training data and annotation resources, we propose a novel method called Contrastive Domain Adaptation with Consistency Match (CDACM). It transfers the knowledge from different but relevant datasets to the unlabelled small-size target dataset and improves the semantic quality of the learnt representations. Specifically, we design a conditional domain adversarial network to exploit discriminative information conveyed in the predictions to mitigate the domain gap between the source and target datasets. Furthermore, due to the small scale of the target dataset, we construct a feature cloud for each target sample and leverage contrastive learning to extract more discriminative features. Lastly, we propose adaptive feature cloud expansion to push the decision boundary to a low-density area. Unlike most existing transfer learning methods that aim only to mitigate the domain gap, our method instead simultaneously considers the domain gap and the data deficiency problem of the target dataset. The conditional domain adaptation and the feature cloud generation of our method are learning jointly to extract discriminative features in an end-to-end manner. Besides, the adaptive feature cloud expansion improves the model's generalisation ability in the target domain. Extensive experiments on pneumonia and COVID-19 diagnosis tasks demonstrate that our method outperforms several state-of-the-art unsupervised domain adaptation approaches, which verifies the effectiveness of CDACM for automated pneumonia diagnosis using chest X-ray imaging.

Subclinical Tuberculosis Disease-A Review and Analysis of Prevalence Surveys to Inform Definitions, Burden, Associations, and Screening Methodology.

While it is known that a substantial proportion of individuals with tuberculosis disease (TB) present subclinically, usually defined as bacteriologically-confirmed but negative on symptom screening, considerable knowledge gaps remain. Our aim was to review data from TB prevalence population surveys and generate a consistent definition and framework for subclinical TB, enabling us to estimate the proportion of TB that is subclinical, explore associations with overall burden and program indicators, and evaluate the performance of screening strategies. We extracted data from all publicly available prevalence surveys conducted since 1990. Between 36.1% and 79.7% (median, 50.4%) of prevalent bacteriologically confirmed TB was subclinical. No association was found between prevalence of subclinical and all bacteriologically confirmed TB, patient diagnostic rate, or country-level HIV prevalence (P values, .32, .4, and .34, respectively). Chest Xray detected 89% (range, 73%-98%) of bacteriologically confirmed TB, highlighting the potential of optimizing current TB case-finding policies.

Small lung nodules detection based on local variance analysis and probabilistic neural network.

BACKGROUND AND OBJECTIVE: In medical examinations doctors use various techniques in order to provide to the patients an accurate analysis of their actual state of health. One of the commonly used methodologies is the x-ray screening. This examination very often help to diagnose some diseases of chest organs. The most frequent cause of wrong diagnosis lie in the radiologist's difficulty in interpreting the presence of lungs carcinoma in chest X-ray. In such circumstances, an automated approach could be highly advantageous as it provides important help in medical diagnosis. METHODS: In this paper we propose a new classification method of the lung carcinomas. This method start with the localization and extraction of the lung nodules by computing, for each pixel of the original image, the local variance obtaining an output image (variance image) with the same size of the original image. In the variance image we find the local maxima and then by using the locations of these maxima in the original image we found the contours of the possible nodules in lung tissues. However after this segmentation stage we find many false nodules. Therefore to discriminate the true ones we use a probabilistic neural network as classifier. RESULTS: The performance of our approach is 92% of correct classifications, while the sensitivity is 95% and the specificity is 89.7%. The misclassification errors are due to the fact that network confuses false nodules with the true ones (6%) and true nodules with the false ones (2%). CONCLUSIONS: Several researchers have proposed automated algorithms to detect and classify pulmonary nodules but these methods fail to detect low-contrast nodules and have a high computational complexity, in contrast our method is relatively simple but at the same time provides good results and can detect low-contrast nodules. Furthermore, in this paper is presented a new algorithm for training the PNN neural networks that allows to obtain PNNs with many fewer neurons compared to the neural networks obtained by using the training algorithms present in the literature. So considerably lowering the computational burden of the trained network and at same time keeping the same performances.

Is chest X-ray screening for lung cancer in smokers cost-effective? Evidence from a population-based study in Italy.

BACKGROUND: After implementation of the PREDICA annual chest X-ray (CXR) screening program in smokers in the general practice setting of Varese-Italy a significant reduction in lung cancer-specific mortality (18 %) was observed. The objective of this study covering July 1997 through December 2006 was to estimate the cost-effectiveness of this intervention. METHODS: We examined detailed information on lung cancer (LC) cases that occurred among smokers invited to be screened in the PREDICA study (Invitation-to-screening Group, n = 5815 subjects) to estimate costs and quality-adjusted life-years (QALYs) from LC diagnosis until death. The control group consisted of 156 screening-eligible smokers from the same area, uninvited and unscreened, who developed LC and were treated by usual care. We calculated the incremental net monetary benefit (INMB) by comparing LC management in screening participants (n = 1244 subjects) and in the Invitation-to-screening group versus control group. RESULTS: The average number of QALYs since LC diagnosis was 1.7, 1.49 and 1.07, respectively, in screening participants, the invitation-to-screening group, and the control group. The average total cost (screening + management) per LC case was higher in screening participants (€17,516) and the Invitation-to-screening Group (€16,167) than in the control group (€15,503). Assuming a maximum willingness to pay of €30,000/QALY, we found that the intervention was cost-effective with high probability: 79 % for screening participation (screening participants vs. control group) and 95 % for invitation-to-screening (invitation-to-screening group vs. control group). CONCLUSIONS: Based on the PREDICA study, annual CXR screening of high-risk smokers in a general practice setting has high probability of being cost-effective with a maximum willingness to pay of €30,000/QALY.

Chest X-ray screening for lung cancer: overdiagnosis, endpoints, and randomized population trials.

Publication of the National Lung Screening Trial (NLST) generated excitement by concluding that CT screening reduces lung cancer mortality when compared to chest X-ray (CXR) screening. In contrast, CXR screening has long been considered to be ineffective. This is because randomized population trials (RPTs) have failed to demonstrate significant mortality reductions in populations randomized to CXR screening. While these studies demonstrate that CXR screening is associated with significant survival advantages, these advantages have been widely interpreted as spurious, due to the inference that CXR screening leads to substantial lung cancer overdiagnosis. Indeed, the reality of the overdiagnosis hypothesis is the only alternative to the conclusion that CXR screening was effective in these trials and that survival more accurately reflected the benefit of CXR screening than mortality. Mortality comparisons would be biased if randomization fails to create comparison groups with an equal probability of mortality from the target cancer. The objective of this manuscript is to review existing RPTs on CXR screening for lung cancer, and to analyze which endpoint most accurately reflects screening efficacy. We conclude that the evidence supports that CXR screening is superior to no screening, and the magnitude of overdiagnosis is minimal in the context of CXR screening.