Abstract:
:Bone age assessment (BAA) is a commonly performed diagnostic study in pediatric radiology to assess skeletal maturity. The most commonly utilized method for assessment of BAA is the Greulich and Pyle method (Pediatr Radiol 46.9:1269-1274, 2016; Arch Dis Child 81.2:172-173, 1999) atlas. The evaluation of BAA can be a tedious and time-consuming process for the radiologist. As such, several computer-assisted detection/diagnosis (CAD) methods have been proposed for automation of BAA. Classical CAD tools have traditionally relied on hard-coded algorithmic features for BAA which suffer from a variety of drawbacks. Recently, the advent and proliferation of convolutional neural networks (CNNs) has shown promise in a variety of medical imaging applications. There have been at least two published applications of using deep learning for evaluation of bone age (Med Image Anal 36:41-51, 2017; JDI 1-5, 2017). However, current implementations are limited by a combination of both architecture design and relatively small datasets. The purpose of this study is to demonstrate the benefits of a customized neural network algorithm carefully calibrated to the evaluation of bone age utilizing a relatively large institutional dataset. In doing so, this study will aim to show that advanced architectures can be successfully trained from scratch in the medical imaging domain and can generate results that outperform any existing proposed algorithm. The training data consisted of 10,289 images of different skeletal age examinations, 8909 from the hospital Picture Archiving and Communication System at our institution and 1383 from the public Digital Hand Atlas Database. The data was separated into four cohorts, one each for male and female children above the age of 8, and one each for male and female children below the age of 10. The testing set consisted of 20 radiographs of each 1-year-age cohort from 0 to 1 years to 14-15+ years, half male and half female. The testing set included left-hand radiographs done for bone age assessment, trauma evaluation without significant findings, and skeletal surveys. A 14 hidden layer-customized neural network was designed for this study. The network included several state of the art techniques including residual-style connections, inception layers, and spatial transformer layers. Data augmentation was applied to the network inputs to prevent overfitting. A linear regression output was utilized. Mean square error was used as the network loss function and mean absolute error (MAE) was utilized as the primary performance metric. MAE accuracies on the validation and test sets for young females were 0.654 and 0.561 respectively. For older females, validation and test accuracies were 0.662 and 0.497 respectively. For young males, validation and test accuracies were 0.649 and 0.585 respectively. Finally, for older males, validation and test set accuracies were 0.581 and 0.501 respectively. The female cohorts were trained for 900 epochs each and the male cohorts were trained for 600 epochs. An eightfold cross-validation set was employed for hyperparameter tuning. Test error was obtained after training on a full data set with the selected hyperparameters. Using our proposed customized neural network architecture on our large available data, we achieved an aggregate validation and test set mean absolute errors of 0.637 and 0.536 respectively. To date, this is the best published performance on utilizing deep learning for bone age assessment. Our results support our initial hypothesis that customized, purpose-built neural networks provide improved performance over networks derived from pre-trained imaging data sets. We build on that initial work by showing that the addition of state-of-the-art techniques such as residual connections and inception architecture further improves prediction accuracy. This is important because the current assumption for use of residual and/or inception architectures is that a large pre-trained network is required for successful implementation given the relatively small datasets in medical imaging. Instead we show that a small, customized architecture incorporating advanced CNN strategies can indeed be trained from scratch, yielding significant improvements in algorithm accuracy. It should be noted that for all four cohorts, testing error outperformed validation error. One reason for this is that our ground truth for our test set was obtained by averaging two pediatric radiologist reads compared to our training data for which only a single read was used. This suggests that despite relatively noisy training data, the algorithm could successfully model the variation between observers and generate estimates that are close to the expected ground truth.
journal_name
J Digit Imagingjournal_title
Journal of digital imagingauthors
Mutasa S,Chang PD,Ruzal-Shapiro C,Ayyala Rdoi
10.1007/s10278-018-0053-3subject
Has Abstractpub_date
2018-08-01 00:00:00pages
513-519issue
4eissn
0897-1889issn
1618-727Xpii
10.1007/s10278-018-0053-3journal_volume
31pub_type
杂志文章abstract::The study investigates the effect of a substantial dose reduction on the variability of lung nodule volume measurements by assessing and comparing nodule volumes using a dedicated semiautomated segmentation software on ultralow-dose computed tomography (ULD-CT) and standard-dose computed tomography (SD-CT) data. In 20...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-008-9157-5
更新日期:2010-02-01 00:00:00
abstract::Extensive research efforts have been devoted to the feasibility of picture archiving and communication systems (PACS) in recent years. The advantages of PACS are numerous but mainly include reduced cost and improvement in the operational efficiency of a PACS-based radiology department. In digital radiography, images a...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03168839
更新日期:1997-11-01 00:00:00
abstract::Clinical picture archiving and communications systems provide convenient, efficient access to digital medical images from multiple modalities but can prove challenging to deploy, configure and use. MRIdb is a self-contained image database, particularly suited to the storage and management of magnetic resonance imaging...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-013-9604-9
更新日期:2013-10-01 00:00:00
abstract::Even though facilities using computed radiography (CR) operate in an electronic environment, the production of hard-copy films is still necessary during the transition period, as well as for particular needs following complete implementation. We have implemented a quantitative technique to match the response of printe...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03167622
更新日期:2000-05-01 00:00:00
abstract::The Integrating the Healthcare Enterprise (IHE) Teaching File and Clinical Trial Export (TCE) integration profile describes a standard workflow for exporting key images from an image manager/archive to a teaching file, clinical trial, or electronic publication application. Two specific digital imaging and communicatio...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-007-9052-5
更新日期:2008-12-01 00:00:00
abstract::In today's radiology workflow, free-text reporting is established as the most common medium to capture, store, and communicate clinical information. Radiologists routinely refer to prior radiology reports of a patient to recall critical information for new diagnosis, which is quite tedious, time consuming, and prone t...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-018-0116-5
更新日期:2019-02-01 00:00:00
abstract::Frequently when patient and study identification information (patient name, patient identification, date of birth, sex, and accession number) are manually entered at a modality, typographical errors occur that have to be corrected before the acquired images can be matched to the proper patient and study on a picture a...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03190323
更新日期:2001-06-01 00:00:00
abstract::Perfusion CT of the liver typically involves scanning the liver at least 20 times, resulting in a large radiation dose. We developed and validated a simplified model of tumor blood supply that can be applied to standard triphasic scans and evaluated whether this can be used to distinguish benign and malignant liver le...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-014-9725-9
更新日期:2015-04-01 00:00:00
abstract::The effect of percutaneous, surgical, and medical therapies for vascular malformations (VMs) is often difficult to quantify volumetrically using cross-sectional imaging. Volumetric measurement is often estimated with serial, expensive MRI examinations which may require sedation or anesthesia. We aim to explore whether...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-019-00183-6
更新日期:2019-04-01 00:00:00
abstract::Web educational resources integrating interactive simulation tools provide students with an in-depth understanding of the medical imaging process. The aim of this work was the development of a purely Web-based, open access, interactive application, as an ancillary learning tool in graduate and postgraduate medical ima...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-014-9689-9
更新日期:2015-02-01 00:00:00
abstract::The purpose of this study is to assess the accuracy, precision, and rapidity of liver volumes calculated by using a freehand electromagnetic pen tablet contourtracing method as compared with the volumes calculated by using the standard optical mouse contourtracing method. The imaging data used as input for accuracy an...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-010-9291-8
更新日期:2011-04-01 00:00:00
abstract::Surface morphology and shape in general are important predictors for the behavior of solid-type lung nodules detected on CT. More broadly, shape analysis is useful in many areas of computer-aided diagnosis and essentially all scientific and engineering disciplines. Automated methods for shape detection have all previo...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-014-9732-x
更新日期:2015-04-01 00:00:00
abstract::The recognizability of facial images extracted from publically available medical scans raises patient privacy concerns. This study examined how accurately facial images extracted from computed tomography (CT) scans are objectively matched with corresponding photographs of the scanned individuals. The test subjects wer...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-016-9932-7
更新日期:2017-04-01 00:00:00
abstract::Content-based medical image retrieval (CBMIR) is a powerful resource to improve differential computer-aided diagnosis. The major problem with CBMIR applications is the semantic gap, a situation in which the system does not follow the users' sense of similarity. This gap can be bridged by the adequate modeling of simil...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-015-9809-1
更新日期:2016-02-01 00:00:00
abstract::The authors present a unique medical technical application for illustrating the success and/or failure of the physiological healing process as a dynamically morphed video. Two examples used in this report include the healing of a severely fractured humerus from an explosion in Iraq and the other of dramatic tissue des...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-006-1050-5
更新日期:2007-09-01 00:00:00
abstract::Digital modalities such as CT, MRI, Ultrasound and Computerized Radiography systems, generating softcopy images to be used by a Picture Archiving and Communication System (PACS), need to identify the images properly in order to retrieve and manage them. In many cases, a technologist re-enters patient demographic and s...
journal_title:Journal of digital imaging
pub_type: 杂志文章,评审
doi:10.1007/BF03168256
更新日期:1998-08-01 00:00:00
abstract::In this paper, we compare five common classifier families in their ability to categorize six lung tissue patterns in high-resolution computed tomography (HRCT) images of patients affected with interstitial lung diseases (ILD) and with healthy tissue. The evaluated classifiers are naive Bayes, k-nearest neighbor, J48 d...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-008-9158-4
更新日期:2010-02-01 00:00:00
abstract::We propose the design of a teaching system named Interpretation and Diagnosis of Mammograms (INDIAM) for training students in the interpretation of mammograms and diagnosis of breast cancer. The proposed system integrates an illustrated tutorial on radiology of the breast, that is, mammography, which uses education te...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-008-9111-6
更新日期:2009-08-01 00:00:00
abstract::Different digital medical images have been printed on paper with a continuous ink jet printer, and the quality has been evaluated. The emphasis has been on digital chest radiographs from a computed radiography system. The ink jet printing technique is described as well as the handling of the image data from image sour...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03167806
更新日期:1992-11-01 00:00:00
abstract::Monitoring patients' imaging-related radiation is currently a hot topic, but there are many obstacles to accurate, patient-specific dose estimation. While some, such as easier access to dose data and parameters, have been overcome, the challenge remains as to how accurately these dose estimates reflect the actual dose...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-012-9567-2
更新日期:2013-08-01 00:00:00
abstract::Devising a method that can select cases based on the performance levels of trainees and the characteristics of cases is essential for developing a personalized training program in radiology education. In this paper, we propose a novel hybrid prediction algorithm called content-boosted collaborative filtering (CBCF) to...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-014-9678-z
更新日期:2014-08-01 00:00:00
abstract::Widespread integration of networking technology will change the hospital imaging environment, making computational, archiving, and film production services available to network-connected modalities. This change will be reflected in the designs of the image acquisition modalities themselves. A view of the future is pre...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03168154
更新日期:1991-11-01 00:00:00
abstract::This manuscript reviews Touch Surgery, a novel online platform geared towards innovating professional training for surgical procedures. In other industries, such as aviation, simulation has already been shown to reduce costs and improve outcomes in crisis (JETS 3(4):348-352, 2010). Studies involving simulation-based l...
journal_title:Journal of digital imaging
pub_type: 杂志文章,评审
doi:10.1007/s10278-018-0102-y
更新日期:2018-10-01 00:00:00
abstract::To solve the problems of image displays in filmless radiology conferences for the purpose of teaching, we made an experimental design of a conference system with dual 50-in. plasma monitors for displaying larger images and a shared folder containing shortcuts to images for quick display during conferences on the deskt...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-007-9095-7
更新日期:2009-12-01 00:00:00
abstract::New consumer technologies and interoperability standards have dated the first standardized curriculum for imaging informatics fellowships suggested by the Society for Computer Applications in Radiology (SCAR) in 2004 (Journal of digital imaging 17(4):244-248, 2004). Last year, analysis from this institution characteri...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-019-00292-2
更新日期:2020-04-01 00:00:00
abstract::There is increasing prevalence of digital diagnostic imaging in veterinary medicine with a progressive need to use medical imaging software. As Digital Imaging and Communications in Medicine (DICOM)-viewers for veterinary use do not require medical device approval in many countries, freeware viewers might be a practic...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-019-00194-3
更新日期:2020-02-01 00:00:00
abstract::Recent healthcare policies have influenced the manner in which patient data is handled in research projects, and the regulations concerning protected health information have become significantly tighter. Thus, new procedures are needed to facilitate research while protecting the confidentiality of patient data and ens...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-010-9350-1
更新日期:2011-10-01 00:00:00
abstract::Digital or computed radiography (CR) using photostimulable storage phosphor plate technology is becoming increasingly popular in certain clinical applications, such as bedside radiography, where it possesses clear advantages over conventional screen-film imaging. The majority of CR systems in clinical use have been ma...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/BF03168504
更新日期:1994-08-01 00:00:00
abstract::Optimization of brightness distribution in the template used for detection of cancerous masses in mammograms by means of correlation coefficient is presented. This optimization is performed by the evolutionary algorithm using an auxiliary mass classifier. Brightness along the radius of the circularly symmetric templat...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-011-9402-1
更新日期:2012-02-01 00:00:00
abstract:BACKGROUND:Orthopedic trauma care relies on two-dimensional radiograms both before and during the operation. Understanding the three-dimensional nature of complex fractures on plain radiograms is challenging. Modern fluoroscopes can acquire three-dimensional volume datasets even during an operation, but the device limi...
journal_title:Journal of digital imaging
pub_type: 杂志文章
doi:10.1007/s10278-007-9078-8
更新日期:2009-10-01 00:00:00