Exploiting Cascaded Ensemble of Features for the Detection of Tuberculosis Using Chest Radiographs

Saif, Ahmed; Imtiaz, Tamjid; Shahnaz, Celia; Zhu, Wei-Ping; Ahmad, M. Omair

doi:10.1109/access.2021.3102077

Cited by 12 publications

(3 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the work of Saif et al [52] the AUROC in tuberculosis classification is set at 0.997 for the Montgomery dataset and 0.981 for the Shenzhen dataset. This was achieved through ensemble voting of different handcrafted and deep-learned features with data augmentation.…”

Section: Different Model Variants Comparisonmentioning

confidence: 99%

Deep metric learning for few-shot X-ray image classification

Prokop,

Tordera,

Jaworek-Korjakowska

et al. 2023

Preprint

View full text Add to dashboard Cite

Deep learning models have proven the potential to aid professionals with medical image analysis, including many image classification tasks. However, the scarcity of data in medical imaging poses a significant challenge, as the limited availability of diverse and comprehensive datasets hinders the development and evaluation of accurate and robust imaging algorithms and models. Few-shot learning approaches have emerged as a potential solution to address this issue. In this research, we propose to deploy the Generalized Metric Learning Model for Few-Shot X-ray Image Classification. The model comprises a feature extractor to embed images into a lower-dimensional space and a distance-based classifier for label assignment based on the relative distance of these embeddings. We extensively evaluate the model using various pre-trained convolutional neural networks (CNNs) and vision transformers (ViTs) as feature extractors. We also assess the performance of the commonly used distance-based classifiers in several few-shot settings. Finally, we analyze the potential to adapt the feature encoders to the medical domain with both supervised and self-supervised frameworks. Our model achieves 0.689 AUROC in 2-way 5-shot COVID-19 recognition task when combined with REMEDIS (Robust and Efficient Medical Imaging with Self-supervision) domain-adapted model as feature extractor, and 0.802 AUROC in 2-way 5-shot tuberculosis recognition task with domain-adapted DenseNet-121 model. Moreover, the simplicity and flexibility of our approach allows for easy improvement in the feature, either by incorporating other few-shot methods or new, powerful architectures into the pipeline.

show abstract

Section: Different Model Variants Comparisonmentioning

confidence: 99%

Deep metric learning for few-shot X-ray image classification

Prokop,

Tordera,

Jaworek-Korjakowska

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Sputum smear microscopy, chest X-rays [3], [4], [5], [6], [7], [8], [9], [10], [11], [12] rapid molecular tests, MRI [13], and culture methods are diagnostic laboratory tests for TB. Sputum smear microscopy is the most common and essential method used in Indonesia because it is the most reliable and cost-effective approach recommended by the WHO for first-line laboratory diagnosis of TB [14].…”

Section: Introductionmentioning

confidence: 99%

A Novel Digitized Microscopic Images of ZN-Stained Sputum Smear and Its Classification Based on IUATLD Grades

Aulia,

Suksmono,

Mengko

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Microscopic detection of acid-fast bacilli (AFB) from mycobacterium tuberculosis (MTB) in Ziehl-Neelsen (ZN)-stained sputum samples is a crucial step in the detection of TB (tuberculosis) disease. Pathologists encounter many challenges that may result in incorrect diagnoses, such as the heterogeneous shape and irregular appearance of MTB, low-quality ZN staining, and errors in in scanning each of the field of view (FoV) using a conventional microscope. Additionally, multiple manual observations may cause fatigue that leads to human error. Several studies have created microscopic imaging databases of sputum samples, aiding researchers in creating computer-aided diagnosis (CAD) for tuberculosis, which is a promising method that offers timely, reliable, and repeatable assistance. Nevertheless, the implementation of CAD systems for TB diagnosis remains an area of ongoing research and development owing to the lack of microscopic image datasets of sputum samples, which represent whole-slide imaging (WSI) that follows the WHO (World Health Organization) regulations. To address this issue, this study developed a novel digitized microscopic image from sputum smear samples of Indonesian patients in the WSI that conform to the WHO regulation. These images are collected as a Microscopic Imaging Database of Tuberculosis Indonesia (MIDTI). This study also proposed a method based on the YOLOv7 (You Only Look Once seventh version) algorithm to develop a CAD for tuberculosis diagnosis by classifying ZN-stained sputum smear samples into International Union Against Tuberculosis and Lung Disease (IUATLD) grades, which has never been revealed in any previous studies.

show abstract

“…In today's world, the introduction and proliferation of deep learning models can make help more effective because these models have demonstrated exceptional achievements in a variety of domains. Radiologists are increasingly turning to and requesting a strategy known as deep learning because it helps them make more precise diagnoses and improves their ability to accurately anticipate patient outcomes [22]. It may be possible to achieve higher classification accuracy by developing ensemble methods that combine a number of different deep learning approaches.…”

Section: Introductionmentioning

confidence: 99%

Tuberculosis Disease Detection through CXR Images based on Deep Neural Network Approach

Patil,

Gadekar

2023

IJRITCC

View full text Add to dashboard Cite

Tuberculosis (TB) is a disease that, if left untreated for an extended period of time, can ultimately be fatal. Early TB detection can be aided by using a deep learning ensemble. In previous work, ensemble classifiers were only trained on images that shared similar characteristics. It is necessary for an ensemble to produce a diverse set of errors in order for it to be useful; this can be accomplished by making use of a number of different classifiers and/or features. In light of this, a brand-new framework has been constructed in this study for the purpose of segmenting and identifying TB in human Chest X-ray. It was determined that searching traditional web databases for chest X-ray was necessary. At this point, we pass the photos that we have collected over to Swin ResUnet3 so that they may be segmented. After the segmented chest X-ray have been provided to it, the Multi-scale Attention-based Densenet with Extreme Learning Machine (MAD-ELM) model will be applied in the detection stage in order to effectively diagnose tuberculosis from human chest X-ray. This will be done in order to maximize efficiency. Because it increased the variety of errors made by the basic classifiers, the supplied variation of the approach that was proposed was able to detect tuberculosis more effectively. The proposed ensemble method produced results with an accuracy of 94.2 percent, which are comparable to those obtained by past efforts.

show abstract

Exploiting Cascaded Ensemble of Features for the Detection of Tuberculosis Using Chest Radiographs

Cited by 12 publications

References 38 publications

Deep metric learning for few-shot X-ray image classification

Deep metric learning for few-shot X-ray image classification

A Novel Digitized Microscopic Images of ZN-Stained Sputum Smear and Its Classification Based on IUATLD Grades

Tuberculosis Disease Detection through CXR Images based on Deep Neural Network Approach

Contact Info

Product

Resources

About