57 research outputs found
Graph Node Based Interpretability Guided Sample Selection for Active Learning.
No full textWhile supervised learning techniques have demonstrated state-of-the-art performance in many medical image analysis tasks, the role of sample selection is important. Selecting the most informative samples contributes to the system attaining optimum performance with minimum labeled samples, which translates to fewer expert interventions and cost. Active Learning (AL) methods for informative sample selection are effective in boosting performance of computer aided diagnosis systems when limited labels are available. Conventional approaches to AL have mostly focused on the single label setting where a sample has only one disease label from the set of possible labels. These approaches do not perform optimally in the multi-label setting where a sample can have multiple disease labels (e.g. in chest X-ray images). In this paper we propose a novel sample selection approach based on graph analysis to identify informative samples in a multi-label setting. For every analyzed sample, each class label is denoted as a separate node of a graph. Building on findings from interpretability of deep learning models, edge interactions in this graph characterize similarity between corresponding interpretability saliency map model encodings. We explore different types of graph aggregation to identify informative samples for active learning. We apply our method to public chest X-ray and medical image datasets, and report improved results over state-of-the-art AL techniques in terms of model performance, learning rates, and robustness
The Octopus Sign-A New HRCT Sign in Pulmonary Langerhans Cell Histiocytosis.
No full textBackground: Fibrosis in pulmonary Langerhans cell histiocytosis (PLCH) histologically comprises a central scar with septal strands and associated airspace enlargement that produce an octopus-like appearance. The purpose of this study was to identify the octopus sign on high-resolution computed tomography (HRCT) images to determine its frequency and distribution across stages of the disease. Methods: Fifty-seven patients with confirmed PLCH were included. Two experienced chest radiologists assessed disease stages as early, intermediate, or late, as well as the lung parenchyma for nodular, cystic, or fibrotic changes and for the presence of the octopus sign. Statistical analysis included Cohen's kappa for interrater agreement and Fisher's exact test for the frequency of the octopus sign. Results: Interobserver agreement was substantial for the octopus sign (kappa = 0.747). Significant differences in distribution of the octopus sign between stages 2 and 3 were found with more frequent octopus signs in stage 2 and fewer in stage 3. In addition, we only found the octopus sign in cases of nodular und cystic lung disease. Conclusions: The octopus sign in PLCH can be identified not only on histological images, but also on HRCT images. Its radiological presence seems to depend on the stage of PLCH
Interpretability-Driven Sample Selection Using Self Supervised Learning For Disease Classification And Segmentation.
No full textIn supervised learning for medical image analysis, sample selection methodologies are fundamental to attain optimum system performance promptly and with minimal expert interactions (e.g. label querying in an active learning setup). In this paper we propose a novel sample selection methodology based on deep features leveraging information contained in interpretability saliency maps. In the absence of ground truth labels for informative samples, we use a novel self supervised learning based approach for training a classifier that learns to identify the most informative sample in a given batch of images. We demonstrate the benefits of the proposed approach, termed Interpretability-Driven Sample Selection (IDEAL), in an active learning setup aimed at lung disease classification and histopathology image segmentation. We analyze three different approaches to determine sample informativeness from interpretability saliency maps: (i) an observational model stemming from findings on previous uncertainty-based sample selection approaches, (ii) a radiomics-based model, and (iii) a novel data-driven self-supervised approach. We compare IDEAL to other baselines using the publicly available NIH chest X-ray dataset for lung disease classification, and a public histopathology segmentation dataset (GLaS), demonstrating the potential of using interpretability information for sample selection in active learning systems. Results show our proposed self supervised approach outperforms other approaches in selecting informative samples leading to state of the art performance with fewer samples
Penetrating chest trauma after attempted suicide: An extraordinary behavior of a posttraumatic pulmonary artery pseudoaneurysm
No full text
Comparison of distinctive models for calculating an interlobar emphysema heterogeneity index in patients prior to endoscopic lung volume reduction
No full textDorothea Theilig,1 Felix Doellinger,1 Alexander Poellinger,1 Vera Schreiter,1 Konrad Neumann,2 Ralf-Harto Hubner31Department of Radiology, Charité Campus Virchow Klinikum, Charité, Universitätsmedizin Berlin, Berlin, Germany; 2Institute of Biometrics and Clinical Epidemiology, Charité Campus Benjamin Franklin, Charité, Universitätsmedizin Berlin, Berlin, Germany; 3Department of Pneumology, Charité Campus Virchow Klinikum, Charité, Universitätsmedizin Berlin, Berlin, GermanyBackground: The degree of interlobar emphysema heterogeneity is thought to play an important role in the outcome of endoscopic lung volume reduction (ELVR) therapy of patients with advanced COPD. There are multiple ways one could possibly define interlobar emphysema heterogeneity, and there is no standardized definition.Purpose: The aim of this study was to derive a formula for calculating an interlobar emphysema heterogeneity index (HI) when evaluating a patient for ELVR. Furthermore, an attempt was made to identify a threshold for relevant interlobar emphysema heterogeneity with regard to ELVR.Patients and methods: We retrospectively analyzed 50 patients who had undergone technically successful ELVR with placement of one-way valves at our institution and had received lung function tests and computed tomography scans before and after treatment. Predictive accuracy of the different methods for HI calculation was assessed with receiver-operating characteristic curve analysis, assuming a minimum difference in forced expiratory volume in 1 second of 100 mL to indicate a clinically important change.Results: The HI defined as emphysema score of the targeted lobe (TL) minus emphysema score of the ipsilateral nontargeted lobe disregarding the middle lobe yielded the best predicative accuracy (AUC =0.73, P=0.008). The HI defined as emphysema score of the TL minus emphysema score of the lung without the TL showed a similarly good predictive accuracy (AUC =0.72, P=0.009). Subgroup analysis suggests that the impact of interlobar emphysema heterogeneity is of greater importance in patients with upper lobe predominant emphysema than in patients with lower lobe predominant emphysema.Conclusion: This study reveals the most appropriate ways of calculating an interlobar emphysema heterogeneity with regard to ELVR. Keywords: CT-quantitative, COPD, emphysema heterogeneity, endoscopic lung volume reductio
Computer-aided diagnosis through medical image retrieval in radiology.
No full textCurrently, radiologists face an excessive workload, which leads to high levels of fatigue, and consequently, to undesired diagnosis mistakes. Decision support systems can be used to prioritize and help radiologists making quicker decisions. In this sense, medical content-based image retrieval systems can be of extreme utility by providing well-curated similar examples. Nonetheless, most medical content-based image retrieval systems work by finding the most similar image, which is not equivalent to finding the most similar image in terms of disease and its severity. Here, we propose an interpretability-driven and an attention-driven medical image retrieval system. We conducted experiments in a large and publicly available dataset of chest radiographs with structured labels derived from free-text radiology reports (MIMIC-CXR-JPG). We evaluated the methods on two common conditions: pleural effusion and (potential) pneumonia. As ground-truth to perform the evaluation, query/test and catalogue images were classified and ordered by an experienced board-certified radiologist. For a profound and complete evaluation, additional radiologists also provided their rankings, which allowed us to infer inter-rater variability, and yield qualitative performance levels. Based on our ground-truth ranking, we also quantitatively evaluated the proposed approaches by computing the normalized Discounted Cumulative Gain (nDCG). We found that the Interpretability-guided approach outperforms the other state-of-the-art approaches and shows the best agreement with the most experienced radiologist. Furthermore, its performance lies within the observed inter-rater variability
Lung perfusion and emphysema distribution affect the outcome of endobronchial valve therapy
No full textChristian Thomsen,1 Dorothea Theilig,2 Dominik Herzog,1 Alexander Poellinger,2 Felix Doellinger,2 Nils Schreiter,3 Vera Schreiter,2 Dirk Schürmann,1 Bettina Temmesfeld-Wollbrueck,1 Stefan Hippenstiel,1 Norbert Suttorp,1 Ralf-Harto Hubner1 1Department of Internal Medicine/Infectious Diseases and Respiratory Medicine, 2Institute of Radiology, 3Institute of Nuclear Medicine, Charité – Universitätsmedizin Berlin, Berlin, Germany Abstract: The exclusion of collateral ventilation (CV) and other factors affect the clinical success of endoscopic lung volume reduction (ELVR). However, despite its benefits, the outcome of ELVR remains difficult to predict. We investigated whether clinical success could be predicted by emphysema distribution assessed by computed tomography scan and baseline perfusion assessed by perfusion scintigraphy. Data from 57 patients with no CV in the target lobe (TL) were retrospectively analyzed after ELVR with valves. Pulmonary function tests (PFT), St George’s Respiratory Questionnaire (SGRQ), and 6-minute walk tests (6MWT) were performed on patients at baseline. The sample was grouped into high and low levels at the median of TL perfusion, ipsilateral nontarget lobe (INL) perfusion, and heterogeneity index (HI). These groups were analyzed for association with changes in outcome parameters from baseline to 3 months follow-up. Compared to baseline, patients showed significant improvements in PFT, SGRQ, and 6MWT (all P≤0.001). TL perfusion was not associated with changes in the outcome. High INL perfusion was significantly associated with increases in 6MWT (P=0.014), and high HI was associated with increases in forced expiratory volume in 1 second (FEV1), (P=0.012). Likewise, there were significant correlations for INL perfusion and improvement of 6MWT (r=0.35, P=0.03) and for HI and improvement in FEV1 (r=0.45, P=0.001). This study reveals new attributes that associate with positive outcomes for patient selection prior to ELVR. Patients with high perfusions in INL demonstrated greater improvements in 6MWT, while patients with high HI were more likely to respond in FEV1. Keywords: endoscopic lung volume reduction, COPD, valves, lung perfusion, emphysema distribution  
- …
