Papers by Hridayanand Goswami
Scientific Reports, Jan 17, 2023
Chest X-rays are the most economically viable diagnostic imaging test for active pulmonary tuberc... more Chest X-rays are the most economically viable diagnostic imaging test for active pulmonary tuberculosis screening despite the high sensitivity and low specificity when interpreted by clinicians or radiologists. Computer aided detection (CAD) algorithms, especially convolution based deep learning architecture, have been proposed to facilitate the automation of radiography imaging modalities. Deep learning algorithms have found success in classifying various abnormalities in lung using chest X-ray. We fine-tuned, validated and tested EfficientNetB4 architecture and utilized the transfer learning methodology for multilabel approach to detect lung zone wise and image wise manifestations of active pulmonary tuberculosis using chest X-ray. We used Area Under Receiver Operating Characteristic (AUC), sensitivity and specificity along with 95% confidence interval as model evaluation metrics. We also utilized the visualisation capabilities of convolutional neural networks (CNN), Gradient-weighted Class Activation Mapping (Grad-CAM) as post-hoc attention method to investigate the model and visualisation of Tuberculosis abnormalities and discuss them from radiological perspectives. EfficientNetB4 trained network achieved remarkable AUC, sensitivity and specificity of various pulmonary tuberculosis manifestations in intramural test set and external test set from different geographical region. The grad-CAM visualisations and their ability to localize the abnormalities can aid the clinicians at primary care settings for screening and triaging of tuberculosis where resources are constrained or overburdened. Global TB report articulates 10.6 million active cases of tuberculosis (TB) [9.9-11.0, 95% uncertainty interval (UI)], 1.4 million deaths (1.3-1.5, 95% UI) and an extra 187,000 (158,000-218,000, 95% UI) deaths (TB with HIV) by a single infectious agent despite being preventable and curable 1. Even though screening, triaging, and diagnostic methods have been developed to detect TB, early detection of Pulmonary TB continues to be challenging in clinical settings where confirmatory tests (Molecular methods, smear microscopy and culture) are limited or unavailable. Timely detection of active TB is crucial for treating an individual and for public health intervention to control TB. Chest X-Ray (CXR) is the oldest and primary radiologic evaluation method for detecting pulmonary TB. It has high sensitivity and low specificity and may appear normal even when active TB is present 2,3. World Health Organization mandates the use of chest X-ray in triaging and screening for active TB 4. The integrated diagnostic algorithm in National Tuberculosis Elimination Programme in India prioritize the use of CXR as a screening method to increase the detection of active pulmonary TB followed by Nucleic Acid Amplification Test for Universal Drug Susceptible Testing. In this context, it is important for clinicians or radiologists to detect and determine the visual signals of active tuberculosis from CXR for early detection and treatment. In developing countries with high burden of TB, lack of trained clinicians in primary care settings or radiologists and inter/
Scientific Reports
Chest X-rays are the most economically viable diagnostic imaging test for active pulmonary tuberc... more Chest X-rays are the most economically viable diagnostic imaging test for active pulmonary tuberculosis screening despite the high sensitivity and low specificity when interpreted by clinicians or radiologists. Computer aided detection (CAD) algorithms, especially convolution based deep learning architecture, have been proposed to facilitate the automation of radiography imaging modalities. Deep learning algorithms have found success in classifying various abnormalities in lung using chest X-ray. We fine-tuned, validated and tested EfficientNetB4 architecture and utilized the transfer learning methodology for multilabel approach to detect lung zone wise and image wise manifestations of active pulmonary tuberculosis using chest X-ray. We used Area Under Receiver Operating Characteristic (AUC), sensitivity and specificity along with 95% confidence interval as model evaluation metrics. We also utilized the visualisation capabilities of convolutional neural networks (CNN), Gradient-weig...
Uploads
Papers by Hridayanand Goswami