Content-based medical image retrieval system for lung diseases using deep CNNs

Agrawal, Shubham; Chowdhary, Aastha; Agarwala, Saurabh; Mayya, Veena; S, Sowmya Kamath

doi:10.1007/s41870-022-01007-7

Cited by 25 publications

(13 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These images were obtained from clinical diagnostic indications with different CT scanner settings. Several methods for extracting image information for classification have been proposed (Shah et al 2016, Agrawal et al 2022. However, previous disease classification systems targeted only a single disease; thus, their scope is substantially narrower than that of clinical practice.…”

Section: Discussionmentioning

confidence: 99%

Interactive content-based image retrieval with deep learning for CT abdominal organ recognition

Lo,

Wang,

Hung

2024

Phys. Med. Biol.

View full text Add to dashboard Cite

Objective: Recognizing the most relevant seven organs in an abdominal computed tomography (CT) slice requires sophisticated knowledge. This study proposed automatically extracting relevant features and applying them in a content-based image retrieval (CBIR) system to provide similar evidence for clinical use.Approach: A total of 2827 abdominal CT slices, including 638 liver, 450 stomach, 229 pancreas, 442 spleen, 362 right kidney, 424 left kidney and 282 gallbladder tissues, were collected to evaluate the proposed CBIR in the present study. Upon fine-tuning, high-level features used to automatically interpret the differences among the seven organs were extracted via deep learning architectures, including DenseNet, Vision Transformer (ViT), and Swin Transformer v2 (SwinViT). Three images with different annotations were employed in the classification and query.Main results: The resulting performances included the classification accuracy (94%-99%) and retrieval result (0.98-0.99). Considering global features and multiple resolutions, SwinViT performed better than ViT. ViT also benefited from a better receptive field to outperform DenseNet. Additionally, the use of hole images can obtain almost perfect results regardless of which deep learning architectures are used.Significance: The experiment showed that using pretrained deep learning architectures and fine-tuning with enough data can achieve successful recognition of seven abdominal organs. The CBIR system can provide more convincing evidence for recognizing abdominal organs via similarity measurements, which could lead to additional possibilities in clinical practice.

show abstract

Section: Discussionmentioning

confidence: 99%

Interactive content-based image retrieval with deep learning for CT abdominal organ recognition

Lo,

Wang,

Hung

2024

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

“…Further, the last image representation was produced by incorporating the removed class probability vectors from the constructed ensemble. Agrawal et al [20] introduced a CBIR technique for retrieving medical images (CBMIR) to enable earlier identification and classification of lung disease depending on X-ray scans. This developed CBMIR method was made on the predicted ability of DNN to detect and classify disease-related features employing TL-based techniques trained on traditional COVID19 chest X-ray (CXR) databases.…”

Section: Related Workmentioning

confidence: 99%

Archimedes Optimization Algorithm With Deep Learning Assisted Content-Based Image Retrieval in Healthcare Sector

Issaoui,

Alohali,

Mtouaa

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Content-based image retrieval (CBIR) in the healthcare field is an advanced technology that leverages the visual or content features of medical images to retrieve similar images from vast data. This technology has significant applications in healthcare, including medical diagnosis, research, and treatment planning. It is a diagnostic tool that enhances the explainability of computer-aided diagnoses (CAD) systems and provides decision-making support for healthcare professionals. A conventional way to CBIR is learning a distance metric by converting images into feature space where the distance between samples is a similarity measurement. CBIR systems can autonomously extract intricate features from medical images by leveraging convolutional neural networks (CNN) and other advanced deep learning techniques, enabling accurate and swift retrieval of relevant patient cases and diagnostic references. This manuscript designs an Archimedes Optimization Algorithm with Deep Learning Assisted Content-Based Image Retrieval in Healthcare Sector (AOADL-CBIRH) technique. The AOADL-CBIRH technique intends to retrieve similar images based on query images in the healthcare sector. In the presented AOADL-CBIRH technique, image pre-processing is initially performed using an adaptive bilateral filtering (ABF) approach to enhance the image quality. For deep feature extraction, the AOADL-CBIRH method applies the Efficient Channel Spatial Model (ECSM) with the ResNet50 model. To improve the retrieval performance, the AOADL-CBIRH technique employs AOA-based hyperparameter tuning for the ECSM-ResNet50 model. Lastly, the Manhattan distance metric determines the similarity between the images and retrieves them. The experimental evaluation of the AOADL-CBIRH algorithm is tested by using benchmark image dataset. The stimulation values signified the enhanced retrieval results of the AOADL-CBIRH technique over other models. INDEX TERMSContent-based image retrieval; Healthcare; Similarity measurement; Deep learning; Archimedes optimization algorithm I. INTRODUCTIONRecently, the accessibility and application of numerous medical imaging approaches namely positron emission tomography (PET), magnetic resonance imaging (MRI), and X-ray scans have led to a dramatic expansion in the size of medical image sources [1]. It highlights the requirement for emerging beneficial healthcare information management systems (HIMS), which are assisted in easy-to-use and instinctive applications like query-based image retrieval, similar image retrieval, and image classification to allow This article has been accepted for publication in IEEE Access.

show abstract

“…Agrawal et al [5] proposed an ImR framework to retrieve chest X-ray images of lungs with COVID-19 infections. Their proposed framework extracted deep features from CNN models (i.e., VGG19 and ResNet50) and utilised distance-based metrics (i.e., chi-square, Euclidean, and cosine) to compute the similarity between images.…”

Section: Recent Work To Retrieve Industrial and Healthcare Images Wit...mentioning

confidence: 99%

“…Their retrieval performance achieved 50.4% in mAP across all classes using the ResNet50 model and cosine metric. Agrawal et al [5] also highlighted that large-scale datasets and advanced CNN architectures can further improve retrieval performance. In 2023, Gassner et al [75] proposed a saliency-enhanced CBIR algorithm to retrieve medical images containing skin lesions.…”

Section: Recent Work To Retrieve Industrial and Healthcare Images Wit...mentioning

confidence: 99%

“…An irregular pattern is a region that differs in consistency or uniformity compared to surrounding areas or a typical example baseline. Irregular pattern retrieval and analysis have been used for the detection of industrial defects [2][3][4], chest infections in medical scans [5][6][7], and ice or snow on lakes [8,9], serving industry, healthcare, and environmental monitoring. An accurate ImR system can aid experts (e.g., manufacturing engineers, doctors, and quality inspectors) during decision-making.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Efficient Retrieval of Images with Irregular Patterns Using Morphological Image Analysis: Applications to Industrial and Healthcare Datasets

Zhang,

Cosma,

Bugby

et al. 2023

J. Imaging

View full text Add to dashboard Cite

Image retrieval is the process of searching and retrieving images from a datastore based on their visual content and features. Recently, much attention has been directed towards the retrieval of irregular patterns within industrial or healthcare images by extracting features from the images, such as deep features, colour-based features, shape-based features, and local features. This has applications across a spectrum of industries, including fault inspection, disease diagnosis, and maintenance prediction. This paper proposes an image retrieval framework to search for images containing similar irregular patterns by extracting a set of morphological features (DefChars) from images. The datasets employed in this paper contain wind turbine blade images with defects, chest computerised tomography scans with COVID-19 infections, heatsink images with defects, and lake ice images. The proposed framework was evaluated with different feature extraction methods (DefChars, resized raw image, local binary pattern, and scale-invariant feature transforms) and distance metrics to determine the most efficient parameters in terms of retrieval performance across datasets. The retrieval results show that the proposed framework using the DefChars and the Manhattan distance metric achieves a mean average precision of 80% and a low standard deviation of ±0.09 across classes of irregular patterns, outperforming alternative feature–metric combinations across all datasets. Our proposed ImR framework performed better (by 8.71%) than Super Global, a state-of-the-art deep-learning-based image retrieval approach across all datasets.

show abstract

Content-based medical image retrieval system for lung diseases using deep CNNs

Cited by 25 publications

References 31 publications

Interactive content-based image retrieval with deep learning for CT abdominal organ recognition

Interactive content-based image retrieval with deep learning for CT abdominal organ recognition

Archimedes Optimization Algorithm With Deep Learning Assisted Content-Based Image Retrieval in Healthcare Sector

Efficient Retrieval of Images with Irregular Patterns Using Morphological Image Analysis: Applications to Industrial and Healthcare Datasets

Contact Info

Product

Resources

About