To assess the potential of compressed-sensing parallel-imaging four-dimensional (4D) phase-contrast magnetic resonance (MR) imaging and specialized imaging software in the evaluation of valvular insufficiency and intracardiac shunts in patients with congenital heart disease.Institutional review board approval was obtained for this HIPAA-compliant study. Thirty-four consecutive retrospectively identified patients in whom a compressed-sensing parallel-imaging 4D phase-contrast sequence was performed as part of routine clinical cardiac MR imaging between March 2010 and August 2011 and who had undergone echocardiography were included. Multiplanar, volume-rendered, and stereoscopic three-dimensional velocity-fusion visualization algorithms were developed and implemented in Java and OpenGL. Two radiologists independently reviewed 4D phase-contrast studies for each of 34 patients (mean age, 6 years; age range, 10 months to 21 years) and tabulated visible shunts and valvular regurgitation. These results were compared with color Doppler echocardiographic and cardiac MR imaging reports, which were generated without 4D phase-contrast visualization. Cohen κ statistics were computed to assess interobserver agreement and agreement with echocardiographic results.The 4D phase-contrast acquisitions were performed, on average, in less than 10 minutes. Among 123 valves seen in 34 4D phase-contrast studies, 29 regurgitant valves were identified, with good agreement between observers (k=0.85). There was also good agreement with the presence of at least mild regurgitation at echocardiography (observer 1, κ=0.76; observer 2, κ=0.77) with high sensitivity (observer 1, 75%; observer 2, 82%) and specificity (observer 1, 97%; observer 2, 95%) relative to the reference standard. Eight intracardiac shunts were identified, four of which were not visible with conventional cardiac MR imaging but were detected with echocardiography. No intracardiac shunts were found with echocardiography alone.With velocity-fusion visualization, the compressed-sensing parallel-imaging 4D phase-contrast sequence can augment conventional cardiac MR imaging by improving sensitivity for and depiction of hemodynamically significant shunts and valvular regurgitation.
We argue that the dominant approach to explainable AI for explaining image classification, annotating images with heatmaps, provides little value for users unfamiliar with deep learning. We argue that explainable AI for images should produce output like experts produce when communicating with one another, with apprentices, and with novices. We provide an expanded set of goals of explainable AI systems and propose a Turing Test for explainable AI.
The study demonstrates that convolutional neural networks can accurately diagnose and stage chronic obstructive pulmonary disease with inspiratory-only chest CT combined with clinical data.
4D flow MRI is a promising method for providing global quantification of cardiac flow in a single acquisition, yet its use in clinical application suffers from low velocity-to-noise ratio. In this work, we present a novel noise reduction processing for 4D flow MRI data using divergence-free wavelet transform. Divergence-free wavelets have the advantage of enforcing soft divergence-free conditions when discretization and partial voluming result in numerical non-divergence-free components and at the same time, provide sparse representation of flow in a generally divergence-free field. Efficient denoising is achieved by appropriate shrinkage of divergence-free and non-divergence-free wavelet coefficients. To verify its performance, divergence-free wavelet denoising was performed on simulated flow and compared with existing methods. The proposed processing was also applied on in vivo data and was demonstrated to improve visualization of flow data while preserving quantifications of flow data.
BRIEF INTRO The ongoing coronavirus (COVID-19) outbreak beginning in December 2019 in Wuhan, China, has spread rapidly, with confirmed cases in multiple countries. This virus causes a severe lower respiratory tract infection, with ∼75% of COVID-19+ hospitalized patients developing a viral pneumonia. Seventeen percent of hospitalized patients go on to develop acute respiratory distress syndrome, and often fatal lung injury representing diffuse alveolar damage on pathologic examination.1 The 2% mortality rate associated with COVID-19+ in China is less than that seen with previous zoonotic coronavirus outbreaks such as SARS (10% mortality) and MERS (30% mortality); it is 20-fold higher than that associated with seasonal influenza according to CDC estimates for 2019-2020.2 Chest radiographs are often obtained as part of the diagnostic workup to triage and daily follow-up of patients with suspected pneumonia, including COVID-19 infection. The rapid recognition of pneumonia in these patients may allow for early isolation precautions and administration of supportive therapies. Deep learning (DL), a form of artificial intelligence, is beginning to show promise for supporting the diagnostic interpretation of chest x-rays. We recently described a DL approach to augment radiographs with a color probability overlay to improve the diagnosis of pneumonia.3 In contrast to common whole-image classification approaches, our method explicitly learns pixel-level likelihoods of pneumonia across the lung parenchyma. This provides natural transparency and explainability. We were interested in assessing the generalizability of our algorithm on frontal chest x-ray images recently published related to the recent COVID-19 outbreak. METHODS A total of 10 frontal chest radiographs from 5 patients treated in China and the United States were sourced from 5 recent COVID-19 epidemiologic and case-study publications.1,4–7 Publication figures with frontal chest radiographs were downloaded as JPEG files and manually cropped to only include the frontal radiograph. These images were used as inputs for our DL algorithm, implemented as a U-Net trained with 22K radiologist-annotated radiographs, which produces pneumonia probability maps overlaid onto an input radiograph. RESULTS Radiographs and the corresponding pneumonia probability maps are shown for each x-ray in Figure 1. Figure 1A shows serial chest radiographs of a COVID-19+ patient from the United States consistent with the evolving atypical pneumonia and progression over several days.4 Our algorithm predicted and consistently localized areas of pneumonia with increasing likelihood, as the subtle airspace opacities increased over time. It is worth noting that each radiograph was analyzed by the algorithm independently without awareness of the time course or relationship of previous films.FIGURE 1: DL-based localization of pneumonia. Radiographs obtained from multiple published COVID-19 case series were analyzed by our algorithm. A, Serial chest radiographs from a US patient with foci of infection that progress over several days. Initially, subtle perihilar airspace opacities are highlighted by the algorithm with low likelihood, which become less apparent on day 3, and continue to progress on days 5 and 6. B, Additional radiographs from 4 Chinese patients. Increasingly confluent airspace opacities in all 4 patients are each highlighted by the algorithm. (Source images were adapted with permissions from the journals publishing Holshue et al,4 Chen et al,1 Song et al,5 Ng et al,6 and Kong et al.7 Therefore, in order to reprint this adapted figure, authorization must be obtained both from the owner of the copyright in the original work and from the owner of copyright in the translation or adaptation).Figure 1B shows 6 additional radiographs for 4 COVID-19+ patients acquired in Chinese hospitals spanning 4 other publications. The group of 3 radiographs on the left side of the panel is from 1 patient over a 7-day span showing progression of a mostly right basilar and perihilar airspace opacities. The 3 radiographs on the right are from different patients. Two illustrate cases showing diffuse bilateral airspace opacities consistent with pneumonia1,5 and another case showing a right infrahilar consolidation subsequently confirmed by computed tomography.7 In each case, the predicted probability map correctly localizes the findings and assigns likelihoods that mirror the severity of the imaging findings. COMMENT These results illustrate a surprising degree of generalizability and robustness of the DL approach that we recently proposed, suggesting that it may have utility in early diagnosis and longitudinal follow-up of suspected pneumonia, including patients with COVID-19 pneumonia. Although our results are not an exhaustive proof of cross-hospital performance, these results imply that cross-institutional generalizability is feasible, standing in contrast to what is generally perceived in the field.8 This is despite considerable variation in the respiratory effort, image contrast, technique, and resolution between each of these published images. It is possible that the decrease in pneumonia likelihood on day 3 in panel A is related to the change inspiratory effort or in the x-ray technique between the outpatient and inpatient settings. A larger study will be necessary to assess the generalizability of this algorithm across institutions. Nevertheless, these results support the idea that DL algorithms will become increasingly valuable as they become further integrated into the clinical diagnostic workflow. Our application to the current COVID-19 outbreak provides a tangible example of how physicians and radiologists can work with artificial intelligence. This has the potential to augment the diagnostic abilities of physicians at the point of care, highlighting subtle abnormalities that may be missed by less experienced physicians, and triage patients for computed tomography. It may also help physicians track the daily evolution of the pulmonary manifestations over a patient's hospitalization before development of diffuse alveolar damage or acute respiratory distress syndrome. As viral epidemics such as COVID-19 place a greater strain on the health care system, it may also provide a mechanism of workload relief and earlier advanced interpretation. Although further study is required to evaluate the effectiveness of this algorithm across multiple institutions, these results provide further evidence that this approach could be a powerful tool for physicians and other health care providers to provide more reliable early diagnosis of infection.
Background Cardiac MRI is routinely performed for quantification of shunt flow in patients with anomalous pulmonary veins, but can be technically‐challenging to perform. Four‐dimensional phase‐contrast (4D‐PC) MRI has potential to simplify this exam. We sought to determine whether 4D‐PC may be a viable clinical alternative to conventional 2D phase‐contrast MR imaging. Methods With institutional review board approval and HIPAA‐compliance, we retrospectively identified all patients with anomalous pulmonary veins who underwent cardiac MRI at either 1.5 Tesla (T) or 3T with parallel‐imaging compressed‐sensing (PI‐CS) 4D‐PC between April, 2011 and October, 2013. A total of 15 exams were included (10 male, 5 female). Algorithms for interactive streamline visualization were developed and integrated into in‐house software. Blood flow was measured at the valves, pulmonary arteries and veins, cavae, and any associated shunts. Pulmonary veins were mapped to their receiving atrial chamber with streamlines. The intraobserver, interobserver, internal consistency of flow measurements, and consistency with conventional MRI were then evaluated with Pearson correlation and Bland‐Altman analysis. Results Triplicate measurements of blood flow from 4D‐PC were highly consistent, particularly at the aortic and pulmonary valves (cv 2–3%). Flow measurements were reproducible by a second observer ( ρ = 0.986–0.999). Direct measurements of shunt volume from anomalous veins and intracardiac shunts matched indirect estimates from the outflow valves ( ρ = 0.966). Measurements of shunt fraction using 4D‐PC using any approach were more consistent with ventricular volumetric displacements than conventional 2D‐PC ( ρ = 0.972–0.991 versus 0.929). Conclusion Shunt flow may be reliably quantified with 4D‐PC MRI, either indirectly or with detailed delineation of flow from multiple shunts. The 4D‐PC may be a more accurate alternative to conventional MRI. J. MAGN. RESON. IMAGING 2015;42:1765–1776.
