Search

Computer-aided diagnosis systems in adult chest radiography (CXR) have recently achieved great success thanks to the availability of large-scale, annotated datasets and the advent of high-performance supervised learning algorithms. However, the development of diagnostic models for detecting and diagnosing pediatric diseases in CXR scans is undertaken due to the lack of high-quality physician-annotated datasets. To overcome this challenge, we introduce and release VinDr-PCXR, a new pediatric CXR dataset of 9,125 studies retrospectively collected from a major pediatric hospital in Vietnam between 2020 and 2021. Each scan was manually annotated by a pediatric radiologist who has more than ten years of experience. The dataset was labeled for the presence of 36 critical findings and 15 diseases. In particular, each abnormal finding was identified via a rectangle bounding box on the image. To the best of our knowledge, this is the first and largest pediatric CXR dataset containing lesion-level annotations and image-level labels for the detection of multiple findings and diseases. For algorithm development, the dataset was divided into a training set of 7,728 and a test set of 1,397. To encourage new advances in pediatric CXR interpretation using data-driven approaches, we provide a detailed description of the VinDr-PCXR data sample and make the dataset publicly available on https://physionet.org/.

Interpretation of chest radiographs (CXR) is a difficult but essential task for detecting thoracic abnormalities. Recent artificial intelligence (AI) algorithms have achieved radiologist-level performance on various medical classification tasks. However, only a few studies addressed the localization of abnormal findings from CXR scans, which is essential in explaining the image-level classification to radiologists. Additionally, the actual impact of AI algorithms on the diagnostic performance of radiologists in clinical practice remains relatively unclear. To bridge these gaps, we developed an explainable deep learning system called VinDr-CXR that can classify a CXR scan into multiple thoracic diseases and, at the same time, localize most types of critical findings on the image. VinDr-CXR was trained on 51,485 CXR scans with radiologist-provided bounding box annotations. It demonstrated a comparable performance to experienced radiologists in classifying 6 common thoracic diseases on a retrospective validation set of 3,000 CXR scans, with a mean area under the receiver operating characteristic curve (AUROC) of 0.967 (95% confidence interval [CI]: 0.958–0.975). The sensitivity, specificity, F1-score, false-positive rate (FPR), and false-negative rate (FNR) of the system at the optimal cutoff value were 0.933 (0.898–0.964), 0.900 (0.887–0.911), 0.631 (0.589–0.672), 0.101 (0.089– 0.114) and 0.067 (0.057–0.102), respectively. For the localization task with 14 types of lesions, our free-response receiver operating characteristic (FROC) analysis showed that the VinDr-CXR achieved a sensitivity of 80.2% at the rate of 1.0 false-positive lesion identified per scan. A prospective study was also conducted to measure the clinical impact of the VinDr-CXR in assisting six experienced radiologists. The results indicated that the proposed system, when used as a diagnosis supporting tool, significantly improved the agreement between radiologists themselves with an increase of 1.5% in mean Fleiss’ Kappa. We also observed that, after the radiologists consulted VinDr-CXR’s suggestions, the agreement between each of them and the system was remarkably increased by 3.3% in mean Co-hen’s Kappa. Altogether, our results highlight the potentials of the proposed deep learning system as an effective assistant to radiologists in clinical practice. Part of the dataset used for developing the VinDr-CXR system has been made publicly available at https://physionet.org/content/vindr-cxr/1.0.0/.

X-ray imaging in DICOM format is the most commonly used imaging modality in clinical practice, resulting in vast, non-normalized databases. This leads to an obstacle in deploying AI solutions for analyzing medical images, which often requires identifying the right body part before feeding the image into a specified AI model. This challenge raises the need for an automated and efficient approach to classifying body parts from X-ray scans. Unfortunately, to the best of our knowledge, there is no open tool or framework for this task to date. To fill this lack, we introduce a DICOM Imaging Router that deploys deep CNNs for categorizing unknown DICOM X-ray images into five anatomical groups: abdominal, adult chest, pediatric chest, spine, and others. To this end, a large-scale X-ray dataset consisting of 16,093 images has been collected and manually classified. We then trained a set of state-of-the-art deep CNNs using a training set of 11,263 images. These networks were then evaluated on an independent test set of 2,419 images and showed superior performance in classifying the body parts. Specifically, our best performing model achieved a recall of 0.982 (95% CI, 0.977-0.988), a precision of 0.985 (95% CI, 0.975-0.989) and a F1-score of 0.981 (95% CI, 0.976-0.987), whilst requiring less computation for inference (0.0295 second per image). Our external validity on 1,000 X-ray images shows the robustness of the proposed approach across hospitals. These remarkable performances indicate that deep CNNs can accurately and effectively differentiate human body parts from X-ray scans, thereby providing potential benefits for a wide range of applications in clinical settings. The dataset, codes, and trained deep learning models from this study will be made publicly available on our project website at https://vindr.ai/., This is a preprint of our paper, which was accepted for publication to ICCV Workshop 2021

Chest radiography is one of the most common types of diagnostic radiology exams, which is critical for screening and diagnosis of many different thoracic diseases. Specialized algorithms have been developed to detect several specific pathologies such as lung nodule or lung cancer. However, accurately detecting the presence of multiple diseases from chest X-rays (CXRs) is still a challenging task. This paper presents a supervised multi-label classification framework based on deep convolutional neural networks (CNNs) for predicting the risk of 14 common thoracic diseases. We tackle this problem by training state-of-the-art CNNs that exploit dependencies among abnormality labels. We also propose to use the label smoothing technique for a better handling of uncertain samples, which occupy a significant portion of almost every CXR dataset. Our model is trained on over 200,000 CXRs of the recently released CheXpert dataset and achieves a mean area under the curve (AUC) of 0.940 in predicting 5 selected pathologies from the validation set. This is the highest AUC score yet reported to date. The proposed method is also evaluated on the independent test set of the CheXpert competition, which is composed of 500 CXR studies annotated by a panel of 5 experienced radiologists. The performance is on average better than 2.6 out of 3 other individual radiologists with a mean AUC of 0.930, which ranks first on the CheXpert leaderboard at the time of writing this paper., Comment: This is a pre-print of our paper that was accepted by Neurocomputing - Its shorter version has been accepted by Medical Imaging with Deep Learning conference (MIDL 2020)

The classical assumption in sampling and spline theories is that the input signal is square-integrable, which prevents us from applying such techniques to signals that do not decay or even grow at infinity. In this paper, we develop a sampling theory for multidimensional non-decaying signals living in weighted $ L _{ p } $ spaces. The sampling and reconstruction of an analog signal can be done by a projection onto a shift-invariant subspace generated by an interpolating kernel. We show that, if this kernel and its biorthogonal counterpart are elements of appropriate hybrid-norm spaces, then both the sampling and the reconstruction are stable. This is an extension of earlier results by Aldroubi and Gröchenig. The extension is required because it allows us to develop the theory for the ideal sampling of non-decaying signals in weighted Sobolev spaces. When the d-dimensional signal and its d∕p + ε derivatives, for arbitrarily small ε > 0, grow no faster than a polynomial in the $ L _{ p } $ sense, the sampling operator is shown to be bounded even without a sampling kernel. As a consequence, the signal can also be interpolated from its samples with a nicely behaved interpolating kernel.

We propose a novel method that combines a convolutional neural network (CNN) with a long short-term memory (LSTM) mechanism for accurate prediction of intracranial hemorrhage on computed tomography (CT) scans. The CNN plays the role of a slice-wise feature extractor while the LSTM is responsible for linking the features across slices. The whole architecture is trained end-to-end with input being an RGB-like image formed by stacking 3 different viewing windows of a single slice. We validate the method on the recent RSNA Intracranial Hemorrhage Detection challenge and on the CQ500 dataset. For the RSNA challenge, our best single model achieves a weighted log loss of 0.0522 on the leaderboard, which is comparable to the top 3% performances, almost all of which make use of ensemble learning. Importantly, our method generalizes very well: the model trained on the RSNA dataset significantly outperforms the 2D model, which does not take into account the relationship between slices, on CQ500. Our codes and models is publicly avaiable at https://github.com/VinBDI-MedicalImagingTeam/midl2020-cnnlstm-ich.

Tyrosine is mainly degraded in the liver by a series of enzymatic reactions. Abnormal expression of the tyrosine catabolic enzyme tyrosine aminotransferase (TAT) has been reported in patients with hepatocellular carcinoma (HCC). Despite this, aberration in tyrosine metabolism has not been investigated in cancer development. In this work, we conduct comprehensive cross-platform study to obtain foundation for discoveries of potential therapeutics and preventative biomarkers of HCC. We explore data from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), Gene Expression Profiling Interactive Analysis (GEPIA), Oncomine and Kaplan Meier plotter (KM plotter) and performed integrated analyses to evaluate the clinical significance and prognostic values of the tyrosine catabolic genes in HCC. We find that five tyrosine catabolic enzymes are downregulated in HCC compared to normal liver at mRNA and protein level. Moreover, low expression of these enzymes correlates with poorer survival in patients with HCC. Notably, we identify pathways and upstream regulators that might involve in tyrosine catabolic reprogramming and further drive HCC development. In total, our results underscore tyrosine metabolism alteration in HCC and lay foundation for incorporating these pathway components in therapeutics and preventative strategies.

Chest radiography is one of the most common types of diagnostic radiology exams, which is critical for screening and diagnosis of many different thoracic diseases. Specialized algorithms have been developed to detect several specific pathologies such as lung nodule or lung cancer. However, accurately detecting the presence of multiple diseases from chest X-rays (CXRs) is still a challenging task. This paper presents a supervised multi-label classification framework based on deep convolutional neural networks (CNNs) for predicting the risk of 14 common thoracic diseases. We tackle this problem by training state-of-the-art CNNs that exploit dependencies among abnormality labels. We also propose to use the label smoothing technique for a better handling of uncertain samples, which occupy a significant portion of almost every CXR dataset. Our model is trained on over 200,000 CXRs of the recently released CheXpert dataset and achieves a mean area under the curve (AUC) of 0.940 in predicting 5 selected pathologies from the validation set. This is the highest AUC score yet reported to date. The proposed method is also evaluated on the independent test set of the CheXpert competition, which is composed of 500 CXR studies annotated by a panel of 5 experienced radiologists. The performance is on average better than 2.6 out of 3 other individual radiologists with a mean AUC of 0.930, which ranks first on the CheXpert leaderboard at the time of writing this paper.

The Poisson summation formula (PSF) describes the equivalence between the sampling of an analog signal and the periodization of its frequency spectrum. In engineering textbooks, the PSF is usually stated formally without explicit conditions on the signal for the formula to hold. By contrast, in the mathematics literature, the PSF is commonly stated and proven in the pointwise sense for various types of \(L_1\) signals. This \(L_1\) assumption is, however, too restrictive for many signal-processing tasks that demand the sampling of possibly growing signals. In this paper, we present two generalized versions of the PSF for d-dimensional signals of polynomial growth. In the first generalization, we show that the PSF holds in the space of tempered distributions for every continuous and polynomially growing signal. In the second generalization, the PSF holds in a particular negative-order Sobolev space if we further require that \(d/2+\varepsilon \) derivatives of the signal are bounded by some polynomial in the \(L_2\) sense.

In our recent work, the sampling and reconstruction of non-decaying signals, modeled as members of weighted- $$L_p$$ spaces, were shown to be stable with an appropriate choice of the generating kernel for the shift-invariant reconstruction space. In this paper, we extend the Strang–Fix theory to show that, for d-dimensional signals whose derivatives up to order L are all in some weighted- $$L_p$$ space, the weighted norm of the approximation error can be made to go down as $$O(h^L)$$ when the sampling step h tends to 0. The sufficient condition for this decay rate is that the generating kernel belongs to a particular hybrid-norm space and satisfies the Strang–Fix conditions of order L. We show that the $$O(h^L)$$ behavior of the error is attainable for both approximation schemes using projection (when the signal is prefiltered with the dual kernel) and interpolation (when a prefilter is unavailable). The requirement on the signal for the interpolation method, however, is slightly more stringent than that of the projection because we need to increase the smoothness of the signal by a margin of $$d/p+\varepsilon $$ , for arbitrary $$\varepsilon >0$$ . This extra amount of derivatives is used to make sure that the direct sampling is stable.