editorial computer-aided detection and diagnosis in...

Hindawi Publishing CorporationComputational and Mathematical Methods in MedicineVolume 2013, Article ID 790608, 2 pageshttp://dx.doi.org/10.1155/2013/790608

EditorialComputer-Aided Detection and Diagnosis in Medical Imaging

Chung-Ming Chen,1 Yi-Hong Chou,2 Norio Tagawa,3 and Younghae Do4

1 Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan2Department of Radiology, Taipei Veterans General Hospital and National Yang Ming University, Taipei, Taiwan3Division of Information and Communication Systems, Tokyo Metropolitan University, Tokyo, Japan4Department of Mathematics, Kyungpook National University, Kyungpook, Republic of Korea

Correspondence should be addressed to Chung-Ming Chen; [email protected]

Received 28 July 2013; Accepted 28 July 2013

Copyright © 2013 Chung-Ming Chen et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Medical images nowadays play an essential role in detec-tion and diagnosis of numerous diseases. Ranging fromanatomical information, functional activities, to the molec-ular and cellular expressions, medical imaging providesdirect visualization means to see through the human bodiesand observe the minute anatomical changes and biologicalprocesses characterized by different physical and biologicalparameters. Informative as they are, medical imaging usuallyrequires experienced medical doctors to best interpret theinformation revealed in the images. However, because ofvarious subjective factors as well as limited analysis time andtools, it is quite common that different medical doctors maycome up with diverse interpretations, leading to differentdiagnoses. Moreover, for the same set of medical imaging,a medical doctor may make different diagnosis results atdifferent time.

To attain a more reliable and accurate diagnosis, recently,varieties of computer-aided detection (CAD) and diagnosis(CADx) approaches have been developed to assist interpre-tation of the medical images. At least four types, denoted asTypes I–IV, of efforts may be identified among these CADand CADx approaches. Type I is to assist visual detection,qualitative analysis, and interactive quantitative analysis ofthe objects of interest in the medical images by eitherenhancing the salient features of the objects or suppressingthe background noises. Type II is to assist feature extractionof the objects of interest for further quantitative analysesby such techniques as boundary delineation, tree-structurereconstruction, fiber tracking, texture analysis, and so on.

Type III is to automatically detect and classify the objectsof interest by integrating the data mining, medical imageanalysis, and signal processing technologies. Type IV is toestimate the anatomical and functional tissue properties notexplicitly revealed in themedical images based onmathemat-ical modeling, for example, physiology, biomechanics, heattransfer, and so forth.

This special issue presents one review paper and fifteenpapers of latest research results on computer-aided detectionand diagnosis in medical imaging covering all four typesof works, including one paper of Type I, six papers ofType II, seven papers of Type III, and two papers of TypeIV. The distribution of these four types of research works,though onlywith a limited number of papers, does reasonablyaccount for the amount of research efforts in the various areasof CAD/CADx in medical imaging.

The Type I paper in this special issue, presented by Y.Dai et el., proposed a volume-rendering-based interactive3D measurement framework for quantitative analysis of 3Dmedical images. The idea is to integrate 3D widgets andvolume clipping into volume rendering, using 3D planewidgets, 3D line widgets, and 3D angle widgets to measurethe areas, distances, and angles of interesting objects.

To assist feature extraction and quantitative analysis, thesix papers of Type II in this special issue may be furtherdivided into three groups, namely, registration, texture anal-ysis, and segmentation, representing three essential tasks inCAD/CADx in medical imaging. The registration group hasonly one paper in which R. Xu et al. addressed one of the key

2 Computational and Mathematical Methods in Medicine

issues in medical image registration, that is, determinationof corresponding points. A particle-system-based methodwas proposed to obtain adaptive sampling positions on theunit sphere for the construction of statistical shape models.In the group of texture analysis, D. Avola et al. presented acustomizing approach to deriving a set of first- and second-order statistics-based operators for texture analysis of MRIimages. J. Deng et al. proposed a robust statistical texturemodel for medical volumes based on a linear tensor coding(LTC) algorithm.Medical volumes are represented by a linearcombination of mutually independent bases, from whichdistinctive bases may be selected for classification. To extractthe morphological information, in the segmentation group,Y. Ujihara et al. developed a two-step region growingmethodto reconstruct cell geometry from confocal fluorescencemicroscopy images of the cytoskeleton. C.-F. Jiang et al.integrated the GVF-snake model and a hybrid registrationtechnique to extract regions from MR T1-weighted images,mapping them into the corresponding SPECT images. Toautomatically recognize the vertebral column in a SPECTimage, S.-F. Huang et al. formulated the bone segmentationproblem as a graph clustering problem and proposed a “bonegraph” image description method to facilitate manipulationof morphological relationships in the skeleton.

To assist nodule detection and/or differential diagnosis ofvarious diseases, the seven papers of Type III in this specialissue presented CAD/CADx approaches for six clinical appli-cations. They are mammographic lesion detection and diag-nosis, lung nodule detection, as well as differential diagnosisof Alzheimer disease (AD) versus mild cognitive impairment(MCI), cerebral lymphomas versus glioblastomas, pancreaticdiseases, and trigger finger disease. As a review paper, T.Ayer et al. provided an informative overview of artificialneural networks-based mammography interpretation anddiagnostic decision making. Both of B. Li et al. and A.Tartar et al. aimed to achieve computer-aided detection oflung nodules in CT images. Nevertheless, A. Tartar et al.focused on classification of a candidate nodule into truenodule or nonnodule by selecting the best features fromthree conventional methods. On the other hand, B. Li at al.proposed a complete framework for nodule detection basedon a fuzzy integrated active contour model and a hybridparametric mixture model of the juxtavascular nodules. Fordifferential diagnosis, S.-T. Yang et al. proposed an MRI-based classification framework to distinguish the patientswith AD and MCI from the normal participants, usingparticle swarm optimization for feature selection and supportvector machine as the classifier. To differentiate cerebrallymphomas fromglioblastomas, T. Yamasaki et al. presented atumor classification system, classifying typical cases by lumi-nance range thresholding and apparent diffusion coefficientsthresholding and nontypical by a support vector machine(SVM). A. Jiang et al. developed a classification method forpancreatic diseases, using multilinear principal componentanalysis to extract the eigen tensors and SVM as the classifierwith the parameter optimized by a quantum simulatedannealing algorithm. Y.-C. Liu et al. proposed two parametersas the pathological progression indices for evaluation oftrigger finger disease from the microscopic pulley images.

These two parameters are the size ratio of the abnormal tissueregions and the number ratio of the abnormal nuclei, derivedfrom a color-based image segmentation system.

Both of the two Type IV papers in this special issue aimedto estimate the anatomical and functional tissue propertiesfrom MRI images. With dynamic MR images, Y.-H. Kaoet al. investigated the respiratory and cardiac pulsations inthe brain of normal subjects based on transfer functionanalysis. W. Swastika et al. analyzed two statistical modelsof diaphragm motion constructed by using regular principalcomponent analysis (PCA) and generalized N-dimensionalPCA (GND-PCA), the results of which showed that theGND-PCA model was superior to the PCA model.

Chung-Ming ChenYi-Hong ChouNorio TagawaYounghae Do

Submit your manuscripts athttp://www.hindawi.com

Stem CellsInternational

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinson’s Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com

editorial computer-aided detection and diagnosis in...

Documents