Research Article Volume 6 Issue No. 3

Transcription

1 DOI / ISSN IJESC Research Article Volume 6 Issue No. 3 Detection of Pulmonary Nodules in Thoracic CT Images Using A HVQ Scheme N. Bala Sundara Ganapathy 1, B.Naresh Kumar 2, K.Prabhakaran 3, S.Praveeen 4, P.Jayaseelan 5 Assistant Professor 1 2, 3, 4, UG Scholar Department of IT Panimalar Engineering College, Chennai, India bornnaresh@gmail.com 2 Abstract: Computer-aided detection (CADe) of pulmonary nodules is critical to assisting radiologists in early identification lung cancer from computed tomography (CT) scans. This paper proposes a novel CADe system based on a hierarchical vector quantization (VQ) scheme. Compared with the commonly-used simple thresholding approach, the high-level VQ yields a more accurate segmentation of the lungs from the chest volume. In identifying initial nodule candidates (INCs) within the lungs, the low-level VQ proves to be effective for INCs detection and segmentation, as well as computationally efficient compared to existing approaches. The proposed system was validated on 205 patient cases from the publically available online Lung Image Database Consortium database, with each case having at least one juxta-pleural nodule annotation. Experimental results demonstrated that our CADe system obtained an overall sensitivity of 82.7% at a specificity of 4 FPs/scan. Especially for the Performance on juxtapleural nodules, we observed 89.2% sensitivity at 4.14 FPs/scan. With respect to comparable CADe systems, the proposed system shows outperformance and demonstrates its potential for fast and adaptive detection of pulmonary nodules via CT imaging. Index Terms: Computer-aided detection (CADe), computed tomography(ct) imaging, false positive (FP) reduction, lung nodules, vector quantization (VQ). I. INTRODUCTION According to the up-to-date statistics from the American Cancer Society [1], lung cancer is the leading cause of cancer-related deaths with over deaths estimated for the United States alone in 2013, and the overall five-year survival rate for lung cancer is merely 16%. The survival rate increases to 52% if it is localized, and decreases to 4% if it has metastasized. Detection of pulmonary nodules has a crucial effect on the diagnosis of lung cancer, but the detection is a nontrivial task, not only because the appearance of pulmonary nodules varies in a wide range, but also because nodule densities have low contrast against adjacent vessel segments and other lung tissues. Computed tomography (CT) has been shown as the most popular imaging modality for nodule detection, because it has the ability to provide reliable image textures for the detection of small nodules. The development of lung nodule CADe systems using CT imaging modality has made good progress over the past decade. Generally, such CADe systems consist of three stages: 1) image preprocessing, 2) initial nodule candidates (INCs) identification, and 3) false positive (FP) reduction of the INCs with preservation of the true positives (TPs). II. METHODS This section will provide details of the proposed CADe system for lung nodules. The top-level block diagram of the proposed CADe system is depicted in Fig. 1. In the preprocessor block, the chest volume is extracted from the field-of-view of the image volume by simple thresholding (where the outside of the chest volume does not have anatomical structures). In the detector block, the two lungs are first separated from their surrounding anatomical structures via high level VQ and a connected component analysis. A. Self-Adaptive VQ Algorithm for Image Segmentation VQ was originally used for data compression in signal processing, and became popular in a variety of research fields such as speech recognition, face detection, image compression and classification, and image segmentation.. The general VQ framework evolves two processes: 1) the training process which determines the set of codebook vector according to the probability of the input data; and 2) the encoding process which assigns input vectors to the codebook vectors. The proposed VQ-based image segmentation algorithm is outlined as follows: 1) Perform PCA to obtain the K L transformation matrix for the target VOI, determine the reduced dimension P for the local intensity vector space, and calculate the K L transformed local intensity vector ωi = {ωi1, ωi2,..., ωip} for each voxel i = 1,..., I. International Journal of Engineering Science and Computing, March

2 2) Set the classification threshold T as the maximum PC variance, and set a value for the maximum class number K based on prior anatomical knowledge. 3) i = 1, set the first voxel label v1 = 1, its local intensity vector ω1 as the representative vector c1 for the first class, n1 = 1 as the number of voxels belonging to class 1, and K_ = 1 as the current number of classes. 4) i = i + 1, calculate the squared Euclidean distance d(ωi, ck ) between the local intensity vector ωi of thecurrent voxel and the representative vector ck for each existing class k = 1,..., K_. 5) Let d(ωi, cm) = min1 k K_{d(ωi, ck )}, if d(ωi, cm) <T or K_ = K, the label for the ith voxel is vi = m. cm is updated by cm = (nm cm + ωi)/(nm + 1), and nm = nm + 1.Otherwise, a newclassk_ = K_ + 1is generated with representative vector ck_ = ωi, and the current voxel is labeled as vi = K_ s.t. K_ <= K. 6) Repeat from step 4) until i = I to complete a whole scan. 7) If K_ < K, repeat steps 1) to 6) for another whole scan while setting the classification threshold T to be the variance of the second or higher-order PC until reaching the desired number of tissue types K_ = K. In this paper, we mainly focus on demonstrating the merits of VQ in the detection of INCs, from where a novel CADe system is proposed to efficiently detect pulmonary nodules. The proposed hierarchical VQ scheme is detailed below. B. INCs Detection via a Hierarchical VQ Scheme A very important but difficult task in the CADe of lung nodules is the detection of INCs, which aims to search for suspicious 3-D objects as nodule candidates using specific strategies. This step is required to be characterized by a sensitivity that is as close to 100% as possible, in order to avoid setting a priori upper bound on the CADe system performance. Meanwhile, the INCs should minimize the number of FPs to ease the following P reduction step. This section presents our hierarchical VQ scheme for automatic detection and segmentation of INCs. 1. Lung Segmentation by High Level VQ: First of all, since the chest CT images in the LIDC database were acquired under different scanning protocols, a standardization of CT intensities was carried out to preprocess our target datasets. Then by imposing n empirical threshold of 500 hounsfield unit (HU) on the entire CT scan, we could separate the chest body volume from the surrounding materials, such as air, fastening belts, CT bed, and other background materials. n order to ensure nodule detection being performed within the lung volume, the segmentation of lungs from the body volume is desired. Fig. 3 shows the histogram plot of the intensity of chest body volume from a typical lung CT scan, where we observe a clear separation of two major classes (i.e., the air tissue and other dense body tissue). The proposed high level VQ scheme can avoid this failure on lung segmentation, as shown by the rectangular regions marked in Fig. 4. Comparison of simple thresholding and VQ for lung segmentation. Panel (a) shows a raw CT image with dense pathologies on the left lung lobe. Panel (b) illustrates the lung mask obtained by simple thresholding, and panel (c) is the lung mask obtained by the two-class high level VQ, which is robust to intensity inhomogeneity. 2.INCs Detection by Low Level VQ: After extraction of the lungs from the entire chest volume, the low level VQ aims to simultaneously detect and segment nodules within the much smaller VOI the lung volume. Hereby, low level also indicates the operation of VQ with a more diversified classification compared to the high level VQ. In contrast to the high-level operation, the low-level classification becomes more challenging when it comes to the determination of an appropriate value for the maximum class number K The overall flowchart of the proposed hierarchical VQ scheme for INCs detection is illustrated through Fig. 5. International Journal of Engineering Science and Computing, March

3 curve, we can obtain the classification sensitivity and specificity, which are defined as follows Sensitivity = TP/TP + FN Specificity =TN/TN + FP Here TP, FN, TN, and FP denote the number of true positives, false negatives, true negatives, and false positives, respectively. RESULTS The proposed CADe system was validated on a subset of the largest publicly available database LIDC IDRI. The LIDC IDRI images were collected under different scanner manufacturers, variable spatial resolution and different X- ray imaging parameters (e.g., slice thicknesses: mm; in-plane pixel size: mm; total slice number: slices; tube peak potential energies: kv; and tube current: ma).. Therefore, a total of 205 chest CT scans in the LIDC database were utilized to evaluate the previously presented CADe system. Fig. 5. Flowchart of the proposed hierarchical VQ scheme for INCs detection B.False Positive Reduction from INCs 1) Rule-Based Filtering Operations: It is challenging to thoroughly separate nodules from attached structures due to their similar intensities, especially for the juxta-vascular nodules (the nodules attached to blood vessels). For the INCs obtained at each opening level, obvious FPs are first filtered out using the size rule on volume-equivalent diameter, which is calculated by R1 = [(INC voxels voxel volume)/(4π/3)] (1/3). (3) Because blood vessels are elongated in shape, we also establish the elongation rule to exclude vessel-like structures with large elongation values. And the elongation in 3-D space is defined by Fig. 6 These nodules were annotated with an agreement level of as low as one (i.e., each nodule was annotated by at least one out of the four radiologists). A. Evaluation of the PCA for Feature Extraction The purpose of local intensity feature extraction is to retain the major image information for reducing the computational burden. R2 = max(dx, dy, dz)/min(dx, dy, dz). (4) In addition, because pulmonary nodules are usually compact in shape, we define the following compactness rule: R3 = (dx dy dz)/ _[max(dx, dy, dz)] 3_. (5) In (4) and (5), dx, dy, and dz are the maximal projection lengths (in mm) of an INC object along X, Y, and Z axes, respectively. Rules defined in (3) (5) together can capture the major shape characteristics of pulmonary nodules. In this study, the threshold values for R1, R2, and R3 were determined experimentally. Given the true nodule (or TP) annotations in the studied dataset, we can compute the aforementioned three shape features for each true nodule. R1 < 3.0 _R1 > 30.0 _R2 > 3.0 _R3 < 0.1. (6) 2) Feature-Based SVM Classification: A supervised learning strategy is carried out using the SVM classifier to further reduce FPs.. Both 2-D and 3-D features are extracted for the first three groups of features.by searching the optimal operating point on the averaged ROC Fig. 7 shows an example of one original lung CT image and the corresponding three-component images in the K L domain selected by PCA, with a decreasing order of significance. It is clear that the first three PCs represent the dominant information. International Journal of Engineering Science and Computing, March

4 B. INCs Detection and Rule-Based Filtering Performance For INCs detection, all of the 490 nodules from the 205 chest CT scans were detected by the proposed hierarchical VQ approach. Hence, the sensitivity is 100%, and the number of FPs at this prescreening stage is about 1526 per scan. The subsequent rule-based filtering operations preserved 475 of the nodules detected by VQ and brought the overall detection sensitivity to 96.9%. But the average number of FPs was significantly reduced from 1526 to per scan, which corresponds to a FP reduction rate of more than ten times. Three examples showing the performance of hierarchical INCs detection followed by rule-based filtering operation are illustrated through D. Performance on Detection of Juxta-Pleural Nodules In particular, we further conducted an experiment to examine the performance of our CADe system to the detection of juxtapleural nodules, where the SVM classification was exclusively applied on the juxta-pleural INCs. Technically, the corrected lung mask for each scan is firstly eroded by five layers, and any INC with at least one voxel outside this eroded mask will be treated as a juxtapleural candidate. Each fold consisted of 158 true nodules and 5512 FPs. This random cross-validation process was also repeated for 50 times to generate a series of AUC and ROC results. Fig. 10. Averaged ROC curves of the step-wise selected features for the SVM classification of juxta-pleural nodule candidates. Fig. 8. Examples on the performance of INCs detection and rule-based filtering operations. From top to bottom row: chest CT image; high-level VQ outcome; border-corrected lung mask; low-level VQ segmentation result; INCs after VQ; INCs after rule-based filtering. C. Performance of FP Reduction by SVM Classification After rule-based pruning of the obvious FPs, we further reduced the FPs via feature-based SVM classification. Twofold cross validation was used to estimate the generalization capabilities of the SVM classifier. This two-fold cross validation was repeated for 50 times to generate the averaged SVM classification result. E. Comparison with Existing Methods: We compared the detection accuracy and speed of our CADe system with those of existing systems that also evaluated the LIDC IDRI database. We selected five CADe systems that showed detection capability or reported detection speed. DISCUSSIONS Since LIDC IDRI images were collected from several different institutions, the evaluation of such databases was much more challenging than evaluation on images acquired from a single institution. Another challenge to the capability of our CADe system is to deal with juxta-pleural datasets. This is clinically valuable in the sense that the detection of small nodules is crucial to detecting lung cancer at early stages. Besides the substantial presence of small nodules, the consideration of nodules with low agreement level also places a challenge to the detection power of our CADe system. Fig. 9. Averaged ROC curves showing the SVM classifier performance as a function of the step-wise selected features. Fig. 11. Examples of juxta-pleural nodules that were missed by 2-D closing operation but recovered by 3-D closing operation. International Journal of Engineering Science and Computing, March

5 III. CONCLUSION In this paper, a novel CADe system was proposed for fast and adaptive detection of pulmonary nodules in chest CT scans. Based on our previous work of self-adaptive online VQ for image segmentation, we developed a hierarchical VQ scheme for INCs detection. The outcome from our CADe system, with an overall sensitivity of 89.2% at a specificity level f 4.14 FPs per scan, is promising for tackling this challenging detection task. In a nutshell, the proposed hierarchical INCs detection approach is fast, adaptive, and fully automatic. The presented CADe system yields comparable detection accuracy and more computational efficiency than existing systems, which demonstrates the feasibility of our CADe system for clinicalutility. [8] S. Ukil and J. M. Reinhardt, Anatomy-guided lung lobe segmentation in X-ray CT images, IEEE Trans. Med. Imag., vol. 28, no. 2, pp , Feb [9] I. Sluimer,M. Prokop, and B. van Ginneken, Toward automated segmentation of the pathological lung in CT, IEEE Trans. Med. Imag., vol. 24, no. 8, pp , Aug [10] E. M. van Rikxoort, B. de Hoop, M. A. Viergever, M. Prokop, and B. van Ginneken, Automatic lung segmentation from thoracic computed tomography scans using a hybrid approach with error detection, Med. Phys., vol. 36, no. 7, pp , REFERENCES [1] R. Siegel, D. Naishadham, and A. Jemal, Cancer statistics, CA Cancer J. Clin., vol. 63, pp , [2] C. I. Henschke, D. I. McCauley, D. F. Yankelevitz, D. P. Naidich,G. McGuinness, O. S. Miettinen, D. M. Libby, M. W. Pasmantier, J. Koizumi, N. K. Altorki, and J. P. Smith, Early lung cancer action project: Overall design and findings from baseline screening, Lancet, vol. 354, no. 9173, pp , [3] A. El-Baz and J. Suri, Lung Imaging and Computer Aided Diagnosis.Boca Raton, FL, USA: CRC Press, [4] H. MacMahon, J. H. M. Austin, G. Gamsu, C. J. Herold, J. R. Jett,D. P. Naidich, E. F. Patz, and S. J. Swensen, Guidelines for management of small pulmonary nodules detected on CT scans: A statement from the Fleischner society, Radiology, vol. 237, no. 2, pp , [5] B. Van Ginneken, S. G. Armato 3rd, B. de Hoop, S. van Amelsvoort-van de Vorst, T. Duindam, M. Niemeijer, K. Murphy,A. Schilham, A. Retico, M. E. Fantacci, N. Camarlinghi, F. Bagagli, I. Gori, T. Hara, H. Fujita, G. Gargano, R. Bellotti, S. Tangaro, L. Bola nos, F. De Carlo, P. Cerello, S. Cristian Cheran, E. Lopez Torres, M. Prokop, Comparing and combining algorithms for computer-aided detection of pulmonary nodules in computed tomography scans: The ANODE09 study, Med. Image Anal., vol. 14, no. 6, pp , [6] A. El-Baz, G. M. Beache, G. Gimel farb, K. Suzuki, K. Okada, A. Elnakib, A. Soliman, and B. Abdollahi, Computer-aided diagnosis systems for lung cancer: Challenges and methodologies, Int. J. Biomed. Imag., vol. 2013, pp. 1 46, [7] J. Pu, J. K. Leader, B. Zheng, F. Knollmann, C. Fuhrman, F. C. Sciurba, and D. Gur, A computational geometry approach to automated pulmonary fissure segmentation in CT examinations, IEEE Trans. Med. Imag., vol. 28, no. 5, pp , May International Journal of Engineering Science and Computing, March