Deep Learning Classification of Active Tuberculosis Using Chest X-Rays: Efficacy of Transfer Learning and Generalization Performance of Cross-Population Datasets

Chest X-ray based diagnosis of active Tuberculosis (TB) is one of the oldest ubiquitous tests in medical practice. Artificial Intelligence (AI) based automated detection of abnormality in chest radiography is crucial in radiology workflow. Most deep convolutional neural networks (DCNN) for diagnosing TB by transfer learning from natural images and using the same dataset to evaluate the model performance and diagnostic accuracy. However, dataset shift is a known issue in predictive models in AI, which is unexplored. In this work, we fine-tuned, validated, and tested two benchmark architectures and utilized the transfer learning methodology to measure the diagnostic accuracy on cross-population datasets. We achieved remarkable calcification accuracy of 100% and area under the receiver operating characteristic (AUC) 1.000 [1.000 – 1.000] (with a sensitivity 0.985 [0.971 – 1.000] and a specificity of 0.986 [0.971 – 1.000]) on intramural test set, but significant drop in extramural test set. Accuracy on various extramural test sets varies 50% - 70%, AUC ranges 0.527 – 0.865 (sensitivity and specificity fluctuate 0.394 – 0.995 and 0.443 – 0.864 respectively). Diagnostic performance on the intramural test set observed in this study shows that DCNN can accurately classify active TB and normal chest radiographs, however the external test set shows DCNN is less likely to generalize well on models trained on specific population dataset.

Performance of a deep learning CNN model for the automated detection of 13 common conditions on Chest X-rays

Background and aims: Chest X-rays are widely used, non-invasive, cost effective imaging tests. However, the complexity of interpretation and global shortage of radiologists have led to reporting backlogs, delayed diagnosis and a compromised quality of care. A fully automated, reliable artificial intelligence system that can quickly triage abnormal images for urgent radiologist review would be invaluable in the clinical setting. The aim was to develop and validate a deep learning Convoluted Neural Network algorithm to automate the detection of 13 common abnormalities found on Chest X-rays. Method: In this retrospective study, a VGG 16 deep learning model was trained on images from the Chest-ray 14, a large publicly available Chest X-ray dataset, containing over 112,120 images with annotations. Images were split into training, validation and testing sets and trained to identify 13 specific abnormalities. The primary performance measures were accuracy and precision. Results: The model demonstrated an overall accuracy of 88% in the identification of abnormal X-rays and 87% in the detection of 13 common chest conditions with no model bias. Conclusion: This study demonstrates that a well-trained deep learning algorithm can accurately identify multiple abnormalities on X-ray images. As such models get further refined, they can be used to ease radiology workflow bottlenecks and improve reporting efficiency. Napier Healthcare’s team that developed this model consists of medical IT professionals who specialize in AI and its practical application in acute & long-term care settings. This is currently being piloted in a few hospitals and diagnostic labs on a commercial basis.

Performance of a deep learning CNN model for the automated detection of 13 common conditions on Chest X-rays

Background and aims: Chest X-rays are widely used, non-invasive, cost effective imaging tests. However, the complexity of interpretation and global shortage of radiologists have led to reporting backlogs, delayed diagnosis and a compromised quality of care. A fully automated, reliable artificial intelligence system that can quickly triage abnormal images for urgent radiologist review would be invaluable in the clinical setting. The aim was to develop and validate a deep learning Convoluted Neural Network algorithm to automate the detection of 13 common abnormalities found on Chest X-rays. Method: In this retrospective study, a VGG 16 deep learning model was trained on images from the Chest-ray 14, a large publicly available Chest X-ray dataset, containing over 112,120 images with annotations. Images were split into training, validation and testing sets and trained to identify 13 specific abnormalities. The primary performance measures were accuracy and precision. Results: The model demonstrated an overall accuracy of 88% in the identification of abnormal X-rays and 87% in the detection of 13 common chest conditions with no model bias. Conclusion: This study demonstrates that a well-trained deep learning algorithm can accurately identify multiple abnormalities on X-ray images. As such models get further refined, they can be used to ease radiology workflow bottlenecks and improve reporting efficiency. Napier Healthcare’s team that developed this model consists of medical IT professionals who specialize in AI and its practical application in acute & long-term care settings. This is currently being piloted in a few hospitals and diagnostic labs on a commercial basis.

MSC 2021 Virtual E-Abstract

1. Case Report: Dual Antiplatelet In Capsular Warning Syndrome.2. Anxiety, Depression And Occupational Participation Of Stroke Survivors.3. Atrial Fibrillation In Hypertensive Patient With Prior Stroke: A Case Report.4. Radiology Workflow Efficiency In Managing Stroke Patient During Pandemic Covid-19: Early Experience In A Teaching Hospital.5. Are There Missed Opportunities In Reducing Risk Of Recurrent Cardiovascular Event Among Stroke Survivors Living In The Community?6. Efficiency of Hand-Arm Language Therapy7. The Rash That Solved The Diagnostic Dilemma: An Overlooked Cause Of Ischemic Stroke.

Development and web deployment of an automated neuroradiology MRI protocoling tool with natural language processing

Background A systematic approach to MRI protocol assignment is essential for the efficient delivery of safe patient care. Advances in natural language processing (NLP) allow for the development of accurate automated protocol assignment. We aim to develop, evaluate, and deploy an NLP model that automates protocol assignment, given the clinician indication text. Methods We collected 7139 spine MRI protocols (routine or contrast) and 990 head MRI protocols (routine brain, contrast brain, or other) from a single institution. Protocols were split into training (n = 4997 for spine MRI; n = 839 for head MRI), validation (n = 1071 for spine MRI, fivefold cross-validation used for head MRI), and test (n = 1071 for spine MRI; n = 151 for head MRI) sets. fastText and XGBoost were used to develop 2 NLP models to classify spine and head MRI protocols, respectively. A Flask-based web app was developed to be deployed via Heroku. Results The spine MRI model had an accuracy of 83.38% and a receiver operator characteristic area under the curve (ROC-AUC) of 0.8873. The head MRI model had an accuracy of 85.43% with a routine brain protocol ROC-AUC of 0.9463 and contrast brain protocol ROC-AUC of 0.9284. Cancer, infectious, and inflammatory related keywords were associated with contrast administration. Structural anatomic abnormalities and stroke/altered mental status were indicative of routine spine and brain MRI, respectively. Error analysis revealed increasing the sample size may improve performance for head MRI protocols. A web version of the model is provided for demonstration and deployment. Conclusion We developed and web-deployed two NLP models that accurately predict spine and head MRI protocol assignment, which could improve radiology workflow efficiency.

Rama Co-RADS: Categorical Assessment Scheme of Chest Radiographic Findings for Diagnosing Pneumonia in Patients With Confirmed COVID-19

Due to the rapid spread of COVID-19 during the third wave of infection in Thailand, the number of confirmed COVID-19 cases has increased exponentially since April 2021. As a result, the country’s healthcare facilities and personnel are overwhelmed. Hence, many new intervention strategies have been designed and implemented. In such a resource-constrained condition, multiple alternate care sites, such as converted hotels (the so-called hospitels) and mobile field medical units, have been established for quarantine and taking care of confirmed COVID-19 cases having no or mild symptoms. In this context, it is essential to have clinical and chest radiographic assessment as a baseline screening for an accurate and rapid triage of patients and early detection of COVID-19 pneumonia, which significantly impacts patient outcomes. Therefore, a clear, concise and standardized chest radiographic report is mandatory. To facilitate this process, the authors have introduced Rama Co-RADS for the categorical assessment scheme of pulmonary involvement in COVID-19. After the pilot implementation of Rama Co-RADS in the routine radiology workflow for chest radiography screening in patients with confirmed COVID-19 at the Ramathibodi Hospitels, there is a 24% reduction in the median turnaround radiology reporting time. It also enhances the radiologist’s performance in establishing the diagnosis of COVID-19 pneumonia (especially in the early phase). Furthermore, the categorical assessment scheme in Rama Co-RADS facilitates communication among healthcare personnel, guiding effective management, triage, consultation and treatment of patients with confirmed COVID-19.

The triple rule out CT in acute chest pain: a challenge for emergency radiologists?

Purpose To evaluate the feasibility of triple rule out computed tomography (TRO-CT) in an emergency radiology workflow by comparing the diagnostic performance of cardiovascular and general radiologists in the interpretation of emergency TRO-CT studies in patients with acute and atypical chest pain. Methods Between July 2017 and December 2019, 350 adult patients underwent TRO-CT studies for the assessment of atypical chest pain. Three radiologists with different fields and years of expertise (a cardioradiologist—CR, an emergency senior radiologist—SER, and an emergency junior radiologist—JER) retrospectively and independently reviewed all TRO-CT studies, by trans-axial and multiplanar reconstruction only. Concordance rates were then calculated using as reference blinded results from a different senior cardioradiologist, who previously evaluated studies using all available analysis software. Results Concordance rate was 100% for acute aortic syndrome (AAS) and pulmonary embolism (PE). About coronary stenosis (CS) for non-obstructive (<50%), CS concordance rates were 97.98%, 90.91%, and 97.18%, respectively, for CR, SER, and JER; for obstructive CS (>50%), concordance rates were respectively 88%, 85.7%, and 71.43%. Moreover, it was globally observed a better performance in the evaluation of last half of examinations compared with the first one. Conclusions Our study confirm the feasibility of the TRO-CT even in an Emergency Radiology department that cannot rely on a 24/7 availability of a dedicated skilled cardiovascular radiologist. The “undedicated” radiologists could exclude with good diagnostic accuracy the presence of obstructive stenosis, those with a clinical impact on patient management, without needing time-consuming software and/or reconstructions.

Impact of the COVID-19 pandemic on radiologists

COVID-19 pandemic is one of the biggest crises faced by health-care systems in the recent times. The aim of this study was to assess the impact of the COVID-19 pandemic on radiology workflow, working pattern, training and continuing professional development (CPD) activities, as well as personal well-being of the radiologists during the pandemic. Material and Methods: Questionnaire designed to gather the opinions regarding the impact of the COVID-19 pandemic was distributed to radiologists throughout the world in electronic format. Anonymized responses were obtained and analyzed. Two hundred radiologists, working in 17 different countries, responded to our questionnaire. Majority of the respondents were from India (72.8%) and 70% of the them were in the age group of 25–45 years. About 80% of respondents felt that they were well protected or moderately well protected in terms of the personal protective equipment (PPE), however, most of them felt that the use of PPE had affected their ability to work. Similar number of radiologists felt that there was significant reduction in the radiology workload. More than half of the respondents felt that their working patterns were altered by the pandemic with drastic impact on teaching, CPD activities, and personal well-being. COVID-19 pandemic has had profound impact on the radiologists all over the world. Learning from the experiences of the first wave should be used to provide innovative solutions to some of the challenges posed to provide better radiology services, training, and improve the well-being of radiologists if we encounter a similar situation in the future. COVID-19 pandemic had significant impact on radiologists. Radiologists felt that they were well or moderately well protected with PPEs; however, PPEs affected their ability to work. Radiology workflow was significantly reduced in the pandemic with more radiologists working from home. COVID-19 pandemic had deleterious effect on radiologist’s well-being, education, and CPD activities.