Evaluation of geometric feature descriptors for detection and classification of lung nodules in low dose CT scans of the chest

Farag, A.A.M.; Ali, Asem; Graham, James H.; Farag, Aly A.; Elshazly, Salwa; Falk, Robert

doi:10.1109/isbi.2011.5872380

Cited by 45 publications

(35 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This creates diffiuclties for such applications as object detection and recognistions; and complicates fundamental image analysis tasks such as segementation and registartion. In this paper we focus on biomedical objects that the authors have worked on during the past several years (e.g., [6] [7]). …”

Section: Intoductionmentioning

confidence: 99%

See 1 more Smart Citation

Performance Evaluation of Geometric Feature Descriptors with Application to Classification of Small-Size Lung Nodules in Low Dose CT

Farag¹

2013

Proceedings of 3rd International Conference on Multimedia Technology(ICMT-13)

View full text Add to dashboard Cite

Object modeling is a multifaceted and active area of research in image analysis, manufacturing, and computational geometry. In image analysis, the focus of this paper, objects may be described in terms of their shape and appearance. Geometric features that are invariant to scale, rotation and translation are essential for proper modeling of shape. Deformable random objects, common in biomedical imaging, do not carry distinct features and suffer from multiple sources of uncertainty, including resolution, size, location and occlusion. Registration involves generating the transformation parameters for mapping a source to a target, while categorization involves matching entities into classes. Both processes, involve mapping attributes of objects with respect to each other, and robust features together with proper similarity measures are keys for good performance. This article examines the issues of features, similarity measures for registration and categorization of small-size deformable random objects; a case study of lung nodules in chest CT is used as an example. We show that feature descriptors capturing textural contents are more effective for these types of objects. The paper includes survey and evaluations of several modern feature descriptors that have been introduced in the computer vision and image analysis literature.

show abstract

Section: Intoductionmentioning

confidence: 99%

“…The classification stage using various feature descriptors was analyzed in the author's work (e.g., [7][8]). The most significant classification results were obtained when the shape based signed distance transform was combined to the texture based LBP approach.…”

Section: Feature-based Nodule Classificationmentioning

confidence: 99%

Performance Evaluation of Geometric Feature Descriptors with Application to Classification of Small-Size Lung Nodules in Low Dose CT

Farag¹

2013

Proceedings of 3rd International Conference on Multimedia Technology(ICMT-13)

View full text Add to dashboard Cite

show abstract

“…Conventionally, the classification of lung nodules was performed using hand-crafted imaging features such as histograms [2], Scale Invariant Feature Transform (SIFT) [3], Local Binary Patterns (LBP) [4] and Histogram of Oriented Gradients (HOG) [5]. The extracted sets of features were then classified using a variety of classifiers such as Support Vector Machines (SVM) [6] and Random Forests (RF) [7].…”

Section: Introductionmentioning

confidence: 99%

TumorNet: Lung nodule characterization using multi-view Convolutional Neural Network with Gaussian Process

Hussein

Gillies

Cao

et al. 2017

2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017)

View full text Add to dashboard Cite

Characterization of lung nodules as benign or malignant is one of the most important tasks in lung cancer diagnosis, staging and treatment planning. While the variation in the appearance of the nodules remains large, there is a need for a fast and robust computer aided system. In this work, we propose an end-to-end trainable multi-view deep Convolutional Neural Network (CNN) for nodule characterization. First, we use median intensity projection to obtain a 2D patch corresponding to each dimension. The three images are then concatenated to form a tensor, where the images serve as different channels of the input image. In order to increase the number of training samples, we perform data augmentation by scaling, rotating and adding noise to the input image. The trained network is used to extract features from the input image followed by a Gaussian Process (GP) regression to obtain the malignancy score. We also empirically establish the significance of different high level nodule attributes such as calcification, sphericity and others for malignancy determination. These attributes are found to be complementary to the deep multi-view CNN features and a significant improvement over other methods is obtained.

show abstract

“…This work resulted in 93.3%, specificity of 100% and sensitivity of 91.4%. In [7], the effectiveness of geometric feature descriptors is examined , for false positive reduction and for classification of lung nodules in low dose CT (LDCT) scans using active appearance models (AAM); which are then used to detect candidate nodules based on optimum similarity measured by the normalized cross -correlation (NCC). In [8], weighted multi-scale convergence index (WMCI) and fisher linear discriminant (FLD) were combined.…”

Section: Introductionmentioning

confidence: 99%

Lung Cancer Recognition using Radon Transform and Adaptive Neuro Fuzzy Inference System

Obayya¹,

Ghandour²

2015

IJCA

View full text Add to dashboard Cite

In this paper, we represent computer aided diagnosis (CAD) system for recognition of lung cancer by analyzing CT images of chest. CAD system helps to improve the diagnostic performance of radiologists in their image interpretations. The proposed system relies on three stages mainly; firstly, the CT image is enhanced. Secondly, the lung and tumor are segmented from the input CT image by separating them from other organs in the CT scan. This is done using region growing Algorithm for segmenting the lung parenchyma and a set of morphological operations to detect the tumor. Thirdly, the geometrical information and transformed based features such as Radon transform based features obtained from the extracted tumor are used to classify the lung tumor into benign and malignant employing adaptive neuro fuzzy inference system (ANFIS) classifier. Correct Classification rate of 98% is obtained by using geometric features.

show abstract

Evaluation of geometric feature descriptors for detection and classification of lung nodules in low dose CT scans of the chest

Cited by 45 publications

References 9 publications

Performance Evaluation of Geometric Feature Descriptors with Application to Classification of Small-Size Lung Nodules in Low Dose CT

Performance Evaluation of Geometric Feature Descriptors with Application to Classification of Small-Size Lung Nodules in Low Dose CT

TumorNet: Lung nodule characterization using multi-view Convolutional Neural Network with Gaussian Process

Lung Cancer Recognition using Radon Transform and Adaptive Neuro Fuzzy Inference System

Contact Info

Product

Resources

About