Deep neural models for automated multi-task diagnostic scan management - quality enhancement, view classification and report generation

The detailed physiological perspectives captured by medical imaging provides actionable insights to doctors to manage comprehensive care of patients. However, the quality of such diagnostic image modalities is often affected by mismanagement of the image capturing process by poorly trained technicians and older/poorly maintained imaging equipment. Further, a patient is often subjected to scanning at different orientations to capture the frontal, lateral and sagittal views of the affected areas. Due to the large volume of diagnostic scans performed at a modern hospital, adequate documentation of such additional perspectives is mostly overlooked, which is also an essential key element of quality diagnostic systems and predictive analytics systems. Another crucial challenge affecting effective medical image data management is that the diagnostic scans are essentially stored as unstructured data, lacking a well-defined processing methodology for enabling intelligent image data management for supporting applications like similar patient retrieval , automated disease prediction etc. One solution is to incorporate automated diagnostic image descriptions of the observation/findings by leveraging computer vision and natural language processing. In this work, we present multi-task neural models capable of addressing these critical challenges. We propose ESRGAN, an image enhancement technique for improving the quality and visualization of medical chest X-ray images, thereby substantially improving the potential for accurate diagnosis, automatic detection and region-of-interest segmentation. We also propose a CNN-based model called ViewNet for predicting the view orientation of the X-ray image and generating a medical report using Xception net, thus facilitating a robust medical image management system for intelligent diagnosis applications. Experimental results are demonstrated using standard metrics like BRISQUE, PIQE and BLEU scores, indicating that the proposed models achieved excellent performance. Further, the proposed deep learning approaches enable diagnosis in a lesser time and their hybrid architecture shows significant potential for supporting many intelligent diagnosis applications.

The Impact of Different Types of Orthodontic Appliances and Its Location in Producing CT Scan Artefacts

Fixed orthodontic appliances can produce metal artefacts in CT images which may degrade the diagnostic image quality. The study aimed to evaluate the artefacts based on the types and location of the metallic and non-metallic orthodontic brackets. This is an in-vitro cross-sectional study. Four different types of orthodontic brackets (stainless steel, titanium, monocrystalline, and polycrystalline ceramic bracket) were bonded consecutively in four different locations of the cadaveric skull. All scans were performed by a single operator using the same CT machine followed by a standard scanning protocol. Artefact intensity for all data sets was quantified by calculating the standard deviation (SD) of the grey values within the dataset by following a standard method. The One-way ANOVA Bonferroni test was used for the data analysis. The mean artefact score of the stainless steel bracket was significantly (p < 0.001) high in comparison with other types of the orthodontic brackets. Besides, the mean artefact score was significantly (p=0.002) low when orthodontic brackets were placed unilaterally. Stainless steel brackets produced a significant amount of noise in CT images which can degrade the diagnostic image quality. Thus, the polycrystalline ceramic bracket can be a better alternative of stainless steel brackets for patient need frequent CT scan.

Rising utilization of coronary CT angiography across Europe over the last decade: insights from a large prospective European registry

Background/Introduction The recently updated 2019 ESC guidelines for the diagnosis and management of chronic coronary syndromes endorse the use of coronary computed tomography angiography (CCTA) for exclusion of obstructive coronary artery disease in patients with a low clinical likelihood (Class I, LOE B). Higher demand for CCTA requires broad availability, inevitably involving smaller healthcare providers, such as non-academic hospitals and private practices. Nevertheless, most published data on CCTA image quality and safety rely on exams performed in high-volume academic centers, and little is known about CCTA in non-academic settings. Purpose To investigate the utilization of CCTA across Europe over the last decade, focusing on differences between academic and non-academic centers. Methods We included patients with stable chest pain and suspected coronary artery disease (CAD) who received CCTA and were included in the European Society of Cardiovascular Radiology MR/CT registry 01/2010–01/2020. We compared CT equipment, image quality, radiation dose, the incidence of periprocedural adverse events, patient characteristics, and CCTA findings between academic (high volume university hospitals) and non-academic centers (non-academic hospitals and private practices). Results Overall, 64,317 patients (41.2% women; age 60±13 years) from 212 sites across 19 European countries were included. Academic centers submitted most cases in 2010—2014 (51.6%), whereas non-academic centers accounted for 71.3% of records in 2015–2020. While non-academic centers used less advanced technology, radiation dose remained low (4.54 [interquartile range (IQR) 2.28–6.76] mSv) with a 30% decline of high-dose scans (&gt;7 mSv) over time. Diagnostic image quality was reported in 97.7% of cases, and the rate of acute scan-related events was low (0.4%) (Figure 1). From 2010–2014 to 2015–2020, CCTA nearly doubled in patients with low to intermediate pretest-probability, women &gt;50, and 40–60 years old men (Figure 2). CAD presence and extent decreased slightly over time (prevalence: 2010–2014: 41.5% vs. 2015–2020: 40.6%), (multi-vessel disease in those with CAD: 2010–2014: 61.9% vs. 2015–2020: 55.9%; all p&lt;0.01). Conclusion CCTA expands rapidly to non-academic centers across Europe, increasing availability while maintaining relatively low radiation dose, high diagnostic image quality, and safety. Broad availability of high-quality CCTA is essential for a successfully implementation of the recently updated guidelines for the diagnosis and management of chronic coronary syndromes. FUNDunding Acknowledgement Type of funding sources: None. Changes in CCTA utilization Changes in patient characteristics

Design of a Long-Pass Filter with Effects on Fluorescence Image Observation for Surgical Fluorescence Microscope Applications

The goal of oncological surgery is to completely remove the tumor. Tumors are often difficult to observe with the naked eye because of the presence of numerous blood vessels and the fact the colors of the tumor and blood vessels are similar. Therefore, a fluorescent contrast medium using a surgical microscope is used to observe the removal status of the tumor. To observe the tumor removal status using a fluorescent contrast agent, fluorescence is expressed in the tumor by irradiating with an external light source, and the expressed tumor can be confirmed through a surgical microscope. However, not only fluorescence-expressed tumors are observed under a surgical microscope, but images from an external light source are also mixed and observed. Therefore, since the surgical microscope is connected to a filter, the quality of the diagnostic image is not uniform, and it is difficult to achieve a clear observation. As a result, an asymmetric image quality phenomenon occurs in the diagnostic images. In this paper, a filter with high clarity that provides a symmetrical observation of diagnostic images is developed and manufactured.

Low-count whole-body PET with deep learning in a multicenter and externally validated study

More widespread use of positron emission tomography (PET) imaging is limited by its high cost and radiation dose. Reductions in PET scan time or radiotracer dosage typically degrade diagnostic image quality (DIQ). Deep-learning-based reconstruction may improve DIQ, but such methods have not been clinically evaluated in a realistic multicenter, multivendor environment. In this study, we evaluated the performance and generalizability of a deep-learning-based image-quality enhancement algorithm applied to fourfold reduced-count whole-body PET in a realistic clinical oncologic imaging environment with multiple blinded readers, institutions, and scanner types. We demonstrate that the low-count-enhanced scans were noninferior to the standard scans in DIQ (p < 0.05) and overall diagnostic confidence (p < 0.001) independent of the underlying PET scanner used. Lesion detection for the low-count-enhanced scans had a high patient-level sensitivity of 0.94 (0.83–0.99) and specificity of 0.98 (0.95–0.99). Interscan kappa agreement of 0.85 was comparable to intrareader (0.88) and pairwise inter-reader agreements (maximum of 0.72). SUV quantification was comparable in the reference regions and lesions (lowest p-value=0.59) and had high correlation (lowest CCC = 0.94). Thus, we demonstrated that deep learning can be used to restore diagnostic image quality and maintain SUV accuracy for fourfold reduced-count PET scans, with interscan variations in lesion depiction, lower than intra- and interreader variations. This method generalized to an external validation set of clinical patients from multiple institutions and scanner types. Overall, this method may enable either dose or exam-duration reduction, increasing safety and lowering the cost of PET imaging.

Stretchable self-tuning MRI receive coils based on liquid metal technology (LiquiTune)

Magnetic resonance imaging systems rely on signal detection via radiofrequency coil arrays which, ideally, need to provide both bendability and form-fitting stretchability to conform to the imaging volume. However, most commercial coils are rigid and of fixed size with a substantial mean offset distance of the coil from the anatomy, which compromises the spatial resolution and diagnostic image quality as well as patient comfort. Here, we propose a soft and stretchable receive coil concept based on liquid metal and ultra-stretchable polymer that conforms closely to a desired anatomy. Moreover, its smart geometry provides a self-tuning mechanism to maintain a stable resonance frequency over a wide range of elongation levels. Theoretical analysis and numerical simulations were experimentally confirmed and demonstrated that the proposed coil withstood the unwanted frequency detuning typically observed with other stretchable coils (0.4% for the proposed coil as compared to 4% for a comparable control coil). Moreover, the signal-to-noise ratio of the proposed coil increased by more than 60% as compared to a typical, rigid, commercial coil.

On the way to routine cardiac MRI at 7 Tesla - a pilot study on consecutive 84 examinations

Introduction Cardiac magnetic resonance (CMR) at ultrahigh field (UHF) offers the potential of high resolution and fast image acquisition. Both technical and physiological challenges associated with CMR at 7T require specific hardware and pulse sequences. This study aimed to assess the current status and existing, publicly available technology regarding the potential of a clinical application of 7T CMR. Methods Using a 7T MRI scanner and a commercially available radiofrequency coil, a total of 84 CMR examinations on 72 healthy volunteers (32 males, age 19–70 years, weight 50–103 kg) were obtained. Both electrocardiographic and acoustic triggering were employed. The data were analyzed regarding the diagnostic image quality and the influence of patient and hardware dependent factors. 50 complete short axis stacks and 35 four chamber CINE views were used for left ventricular (LV) and right ventricular (RV), mono-planar LV function, and RV fractional area change (FAC). Twenty-seven data sets included aortic flow measurements that were used to calculate stroke volumes. Subjective acceptance was obtained from all volunteers with a standardized questionnaire. Results Functional analysis showed good functions of LV (mean EF 56%), RV (mean EF 59%) and RV FAC (mean FAC 52%). Flow measurements showed congruent results with both ECG and ACT triggering. No significant influence of experimental parameters on the image quality of the LV was detected. Small fractions of 5.4% of LV and 2.5% of RV segments showed a non-diagnostic image quality. The nominal flip angle significantly influenced the RV image quality. Conclusion The results demonstrate that already now a commercially available 7T MRI system, without major methods developments, allows for a solid morphological and functional analysis similar to the clinically established CMR routine approach. This opens the door towards combing routine CMR in patients with development of advanced 7T technology.

Clinical comparison of sub-mm high-resolution non-contrast coronary CMR angiography against coronary CT angiography in patients with low-intermediate risk of coronary artery disease: a single center trial

Background The widespread clinical application of coronary cardiovascular magnetic resonance (CMR) angiography (CMRA) for the assessment of coronary artery disease (CAD) remains limited due to low scan efficiency leading to prolonged and unpredictable acquisition times; low spatial-resolution; and residual respiratory motion artefacts resulting in limited image quality. To overcome these limitations, we have integrated highly undersampled acquisitions with image-based navigators and non-rigid motion correction to enable high resolution (sub-1 mm3) free-breathing, contrast-free 3D whole-heart coronary CMRA with 100% respiratory scan efficiency in a clinically feasible and predictable acquisition time. Objectives To evaluate the diagnostic performance of this coronary CMRA framework against coronary computed tomography angiography (CTA) in patients with suspected CAD. Methods Consecutive patients (n = 50) with suspected CAD were examined on a 1.5T CMR scanner. We compared the diagnostic accuracy of coronary CMRA against coronary CTA for detecting a ≥ 50% reduction in luminal diameter. Results The 50 recruited patients (55 ± 9 years, 33 male) completed coronary CMRA in 10.7 ± 1.4 min. Twelve (24%) had significant CAD on coronary CTA. Coronary CMRA obtained diagnostic image quality in 95% of all, 97% of proximal, 97% of middle and 90% of distal coronary segments. The sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy were: per patient (100%, 74%, 55%, 100% and 80%), per vessel (81%, 88%, 46%, 97% and 88%) and per segment (76%, 95%, 44%, 99% and 94%) respectively. Conclusions The high diagnostic image quality and diagnostic performance of coronary CMRA compared against coronary CTA demonstrates the potential of coronary CMRA as a robust and safe non-invasive alternative for excluding significant disease in patients at low-intermediate risk of CAD.

A comprehensive review of imaging findings in COVID-19 - status in early 2021

Medical imaging methods are assuming a greater role in the workup of patients with COVID-19, mainly in relation to the primary manifestation of pulmonary disease and the tissue distribution of the angiotensin-converting-enzyme 2 (ACE 2) receptor. However, the field is so new that no consensus view has emerged guiding clinical decisions to employ imaging procedures such as radiography, computer tomography (CT), positron emission tomography (PET), and magnetic resonance imaging, and in what measure the risk of exposure of staff to possible infection could be justified by the knowledge gained. The insensitivity of current RT-PCR methods for positive diagnosis is part of the rationale for resorting to imaging procedures. While CT is more sensitive than genetic testing in hospitalized patients, positive findings of ground glass opacities depend on the disease stage. There is sparse reporting on PET/CT with [18F]-FDG in COVID-19, but available results are congruent with the earlier literature on viral pneumonias. There is a high incidence of cerebral findings in COVID-19, and likewise evidence of gastrointestinal involvement. Artificial intelligence, notably machine learning is emerging as an effective method for diagnostic image analysis, with performance in the discriminative diagnosis of diagnosis of COVID-19 pneumonia comparable to that of human practitioners.