A Mask-guided Attention Deep Learning Model for COVID-19 Diagnosis based on an Integrated CT Scan Images Database

<p>The global extent of COVID-19 mutations and the consequent depletion of hospital resources highlighted the necessity of effective computer-assisted medical diagnosis. COVID-19 detection mediated by deep learning models can help diagnose this highly contagious disease and lower infectivity and mortality rates. Computed tomography (CT) is the preferred imaging modality for building automatic COVID-19 screening and diagnosis models. It is well-known that the training set size significantly impacts the performance and generalization of deep learning models. However, accessing a large dataset of CT scan images from an emerging disease like COVID-19 is challenging. Therefore, data efficiency becomes a significant factor in choosing a learning model. To this end, we present a multi-task learning approach, namely, a mask-guided attention (MGA) classifier, to improve the generalization and data efficiency of COVID-19 classification on lung CT scan images.</p><p>The novelty of this method is compensating for the scarcity of data by employing more supervision with lesion masks, increasing the sensitivity of the model to COVID-19 manifestations, and helping both generalization and classification performance. Our proposed model achieves better overall performance than the single-task baseline and state-of-the-art models, as measured by various popular metrics. In our experiment with different percentages of data from our curated dataset, the classification performance gain from this multi-task learning approach is more significant for the smaller training sizes. Furthermore, experimental results demonstrate that our method enhances the focus on the lesions, as witnessed by both</p><p>attention and attribution maps, resulting in a more interpretable model.</p>

A Mask-guided Attention Deep Learning Model for COVID-19 Diagnosis based on an Integrated CT Scan Images Database

<p>The global extent of COVID-19 mutations and the consequent depletion of hospital resources highlighted the necessity of effective computer-assisted medical diagnosis. COVID-19 detection mediated by deep learning models can help diagnose this highly contagious disease and lower infectivity and mortality rates. Computed tomography (CT) is the preferred imaging modality for building automatic COVID-19 screening and diagnosis models. It is well-known that the training set size significantly impacts the performance and generalization of deep learning models. However, accessing a large dataset of CT scan images from an emerging disease like COVID-19 is challenging. Therefore, data efficiency becomes a significant factor in choosing a learning model. To this end, we present a multi-task learning approach, namely, a mask-guided attention (MGA) classifier, to improve the generalization and data efficiency of COVID-19 classification on lung CT scan images.</p><p>The novelty of this method is compensating for the scarcity of data by employing more supervision with lesion masks, increasing the sensitivity of the model to COVID-19 manifestations, and helping both generalization and classification performance. Our proposed model achieves better overall performance than the single-task baseline and state-of-the-art models, as measured by various popular metrics. In our experiment with different percentages of data from our curated dataset, the classification performance gain from this multi-task learning approach is more significant for the smaller training sizes. Furthermore, experimental results demonstrate that our method enhances the focus on the lesions, as witnessed by both</p><p>attention and attribution maps, resulting in a more interpretable model.</p>

Imaging Findings of Knee Trauma on Magnetic Resonance Imaging

Background: Knee is one of the major joints involved in kinesis. With increasing involvement in sports related activities especially in young people, Trauma related knee pathologies have increased. An accurate diagnosis regarding the type and extent of injuries is essential for early operative as well as non-operative treatment. Methods:This prospective study included total of 82 cases. The patients were referred to the department of Radiodiagnosis from indoor and outdoor departments of Guru Nanak Dev Hospital, Amritsar with suspicion of internal derangement of the knee and with history of knee trauma.Results:The most common age group involved was young males between 15-34 years. In all age groups most of the patients were males. Most common ligament to be injured was Anterior Cruciate Ligament (ACL). Partial tears were more common than complete tears. Posterior Cruciate Ligament (PCL) tears were less common. Medial Collateral Ligament (MCL) tears outnumbered Lateral Collateral Ligament (LCL) tears and grade 2 tears were more common in both. Among the meniscal injuries Medial Meniscus (MM) tears were more common than LM and grade 3 signal was more common in both. Most of the patellar retinaculum injuries were associated with Anterior Cruciate Ligament ACL tears.Conclusions:Post-traumatic pre-arthroscopic MR imaging evaluation has proved to be cost-effective. MRI is an accurate imaging modality complementing the clinical evaluation and providing a global intra-articular and extra-articular assessment of the knee.

Inter-observer agreement on the evaluation of automated breast volume scanner for breast masses

Objective: To explore inter-observer agreement on the evaluation of automated breast volume scanner (ABVS) for breast masses. Methods: A total of 846 breast masses in 630 patients underwent ABVS examinations. The imaging data were independently interpreted by senior and junior radiologists regarding the mass size ([Formula: see text][Formula: see text]cm, [Formula: see text][Formula: see text]cm and total). We assessed inter-observer agreement of BI-RADS lexicons, unique descriptors of ABVS coronal planes. Using BI-RADS 3 or 4a as a cutoff value, the diagnostic performances for 331 masses with pathological results in 253 patients were assessed. Results: The overall agreements were substantial for BI-RADS lexicons ([Formula: see text]–0.779) and the characteristics on the coronal plane of ABVS ([Formula: see text]), except for associated features ([Formula: see text]). However, the overall agreement was moderate for orientation ([Formula: see text]) for the masses [Formula: see text][Formula: see text]cm. The agreements were substantial to be perfect for categories 2, 3, 4, 5 and overall ([Formula: see text]–0.918). However, the agreements were moderate to substantial for categories 4a ([Formula: see text]), 4b ([Formula: see text]), and 4c ([Formula: see text]), except for category 4b of the masses [Formula: see text][Formula: see text]cm ([Formula: see text]). Moreover, for radiologists 1 and 2, there were no significant differences in sensitivity, specificity, accuracy, positive and negative predictive values with BI-RADS 3 or 4a as a cutoff value ([Formula: see text] for all). Conclusion: ABVS is a reliable imaging modality for the assessment of breast masses with good inter-observer agreement.

PeMNet for Pectoral Muscle Segmentation

As an important imaging modality, mammography is considered to be the global gold standard for early detection of breast cancer. Computer-Aided (CAD) systems have played a crucial role in facilitating quicker diagnostic procedures, which otherwise could take weeks if only radiologists were involved. In some of these CAD systems, breast pectoral segmentation is required for breast region partition from breast pectoral muscle for specific analysis tasks. Therefore, accurate and efficient breast pectoral muscle segmentation frameworks are in high demand. Here, we proposed a novel deep learning framework, which we code-named PeMNet, for breast pectoral muscle segmentation in mammography images. In the proposed PeMNet, we integrated a novel attention module called the Global Channel Attention Module (GCAM), which can effectively improve the segmentation performance of Deeplabv3+ using minimal parameter overheads. In GCAM, channel attention maps (CAMs) are first extracted by concatenating feature maps after paralleled global average pooling and global maximum pooling operation. CAMs are then refined and scaled up by multi-layer perceptron (MLP) for elementwise multiplication with CAMs in next feature level. By iteratively repeating this procedure, the global CAMs (GCAMs) are then formed and multiplied elementwise with final feature maps to lead to final segmentation. By doing so, CAMs in early stages of a deep convolution network can be effectively passed on to later stages of the network and therefore leads to better information usage. The experiments on a merged dataset derived from two datasets, INbreast and OPTIMAM, showed that PeMNet greatly outperformed state-of-the-art methods by achieving an IoU of 97.46%, global pixel accuracy of 99.48%, Dice similarity coefficient of 96.30%, and Jaccard of 93.33%, respectively.

Ground Penetrating Radar Algorithm to Sense the Depth of Blood Clot in Microwave Head Imaging

Background: Microwave imaging is one of the emerging non-invasive portable imaging techniques, which uses nonionized radiations to take a detailed view of biological tissues in the microwave frequency range. Brain stroke is an emergency caused by the interruption of the blood supply into parts of brain, leading to the loss of millions of brain cells. Imaging plays a major role in stroke diagnosis for prompt treatment. Objective: This work proposes a computationally efficient algorithm called the GPR algorithm to locate the blood clot with a size of 10 mm in microwave images. Methods: The electromagnetic waves are radiated, and backscattered reflections are received by Antipodal Vivaldi antenna with the parasitic patch (48 mm*21 mm). The received signals are converted to a planar 2D image, and the depth of the blood clot is identified from the B-scan image. The novelty of this work lies in applying the GPR algorithm for the accurate positioning of a blood clot in a multilayered head tissue. Results: The proposed system is effectively demonstrated using a 3D EM simulator and simulated results are verified in a Vector network analyzer (E8363B) with an experimental setup. Conclusion: This an alternative safe imaging modality compared to present imaging systems(CT and MRI)

Us and Mri Neurography Assessment of Diagnostic Criteria in Patients With Carpal Tunnel Syndrome Using Electrophysiological Tests as Gold Standard: A Prospective Study

BackgroundPeripheral neuropathy evaluation depends mainly on physical examination, patient history, electrophysiological studies, with evoked potential abnormalities. High-resolution US has the advantage of being fast, non-invasive modality with nerve dynamic assessment allowing examination of long part of nerve. MR imaging serve better in examination of deeper nerves with higher contrast resolution. It shows great benefit in patient with atypical presentation, Equivocal diagnosis and suspicious of secondary cause and post-surgical relapse.MethodsThis study was conducted prospectively on 32 patients, presented with carpal tunnel syndrome diagnosed by electrophysiological tests. Superficial US of the wrist joint was done to all participants followed by MRI within 1 weeks of the US.We aimed to assess the measurements & criteria of both US & MRN in diagnosis of CTS, depending mainly on the three-measurement assessed by Buchberger et al., then to find the agreement between US & MR Neurography (MRN)ResultsUs proved to have higher rate of CTS prediction, the three main parameters CSA measurement, distal nerve flattening and flexor retinaculum bowing indices showed positive occurrence of 93.7%,59.4% &59.4% respectively. While we found that decreased nerve echotexture was positive in 90.6% of patients.Regarding MRI it showed less diagnostic ability when using CSA measurement as it was positive in 81.2% of patients, also distal tunnel nerve increased flattening and bowed flexor retinaculum positive results were slightly decreased to 56.2% for each. In contrast to high T2 signal of median nerve which was positive in 90.6% of patients.In agreement study, we found statically significant difference supporting US as the primary diagnostic modality of CTS depending mainly on the three measurement CSA, Flattening and bowing indices. Yet, for cases of secondary CTS and detection of underlying entrapping cause as well as innervated muscle early abnormality detection and better tissue characterization, MRI was better diagnostic modality with statistically significant difference. ConclusionsOur results proved that ultrasound examination can be used as first imaging modality after physician evaluation with comparable results to electrophysiological studies in evaluating CTS and try to find the cause. MRN examination came as second step in patients with suspected muscle denervation changes that could not be elicited by US or equivocal cases for detection of secondary cause in clinically suspected patient.

Ten Good Reasons to Practice Neuroultrasound in Critical Care Setting

In the beginning, cerebral ultrasound (US) was not considered feasible because the intact skull was a seemingly impenetrable obstacle. For this reason, obtaining a clear image resolution had been a challenge since the first use of neuroultrasound (NUS) for the assessment of small deep brain structures. However, the improvements in transducer technologies and advances in signal processing have refined the image resolution, and the role of NUS has evolved as an imaging modality for the brain parenchyma within multiple pathologies. This article summarizes ten crucial applications of cerebral ultrasonography for the evaluation and management of neurocritical patients, whose transfer from and to intensive care units poses a real problem to medical care staff. This also encompasses ease of use, low cost, wide acceptance by patients, no radiation risk, and relative independence from movement artifacts. Bedsides, availability and reliability raised the interest of critical care intensivists in using it with increasing frequency. In this mini-review, the usefulness and the advantages of US in the neurocritical care setting are discussed regarding ten aspects to encourage the intensivist physician to practice this important tool.