Machine learning concepts, concerns and opportunities for a pediatric radiologist

Moore, Michael J.; Slonimsky, Einat; Long, Aaron; Sze, Raymond W.; Iyer, Ramesh S.

doi:10.1007/s00247-018-4277-7

Cited by 46 publications

(32 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Utilization of deep learning in pediatric radiology has several limitations. First is the relatively small number of data for consideration in comparison with available adult data 4 . In addition, the images can be different according to age and growth of the children 4 .…”

Section: Discussionmentioning

confidence: 99%

“…First is the relatively small number of data for consideration in comparison with available adult data 4 . In addition, the images can be different according to age and growth of the children 4 . However, because large datasets are fundamental for development of algorithms 18 , we tried to have as many intussusception radiographs available as possible.…”

Section: Discussionmentioning

confidence: 99%

“…Deep learning represents an increasingly disruptive technology and is being adopted earlier in radiology for reasons such as detection, segmentation, and classification using imaging data 4–7 . Also, among children, deep learning-based algorithms have been used for analysis of bone and chest radiographs for assessment of bone age, fracture, effusion, and differentiation of various projection views 4,8–10 . However, there is no study that has assessed whether deep learning could aid with detection of focal lesions on abdominal radiographs of young children.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Performance of deep learning-based algorithm for detection of ileocolic intussusception on abdominal radiographs of young children

Kim

Yoon

Lee

et al. 2019

Sci Rep

View full text Add to dashboard Cite

The purpose of this study was to develop and test the performance of a deep learning-based algorithm to detect ileocolic intussusception using abdominal radiographs of young children. For the training set, children (≤5 years old) who underwent abdominal radiograph and ultrasonography (US) for suspicion of intussusception from March 2005 to December 2017 were retrospectively included and divided into control and intussusception groups according to the US results. A YOLOv3-based algorithm was developed to recognize the rectangular area of the right abdomen and to diagnose intussusception. For the validation set, children (≤5 years old) who underwent both radiograph and US from January to August 2018 with the suspicion of intussusception were included. Diagnostic performances of an algorithm and radiologists were compared. Total 681 children including 242 children in intussusception group were included in the training set and 75 children including 25 children in intussusception group were included in the validation set. The sensitivity of the algorithm was higher compared with that of the radiologists (0.76 vs. 0.46, p = 0.013), while specificity was not different between the algorithm and the radiologists (0.96 vs. 0.92, p = 0.32). Deep learning-based algorithm can aid screening of intussusception using abdominal radiography in young children.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Performance of deep learning-based algorithm for detection of ileocolic intussusception on abdominal radiographs of young children

Kim

Yoon

Lee

et al. 2019

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Due to the simplicity of our proposed models and the limitation of the small sample size, five samples per predictor was chosen in this study and this required a minimum of thirty investigator contours per model. Datasets are typically split 70%/30% into training and testing datasets [16], where the training dataset is used to train the model and the testing dataset is used to validate the model to determine its predictive accuracy, sensitivity and specificity. For smaller datasets, a k-fold cross validation technique can be applied to test and validate the data and produce each model's predictive accuracy, sensitivity and specificity.…”

Section: Supervised Machine Learningmentioning

confidence: 99%

Automatic evaluation of contours in radiotherapy planning utilising conformity indices and machine learning

Terparia

Mir

Tsang

et al. 2020

Physics and Imaging in Radiation Oncology

View full text Add to dashboard Cite

Background and purpose: Peer-review of Target Volume (TV) and Organ at Risk (OAR) contours in radiotherapy planning are typically conducted visually; this can be time consuming and subject to interobserver variation. This study investigated automatic evaluation of contouring using conformity indices and supervised machine learning. Methods: A total of 393 contours from 253 Stereotactic Ablative Body Radiotherapy (SABR) benchmark cases (adrenal gland, liver, pelvic lymph node and spine), delineated by 132 clinicians from 25 centres, were visually evaluated for conformity against gold standard contours. Contours were scored as "pass" or "fail" on visual peer review and six Conformity Indices (CIs) were applied. CI values were mapped to pass/fail scores for each contour and used to train supervised machine learning models. A 5-fold cross validation method was employed to determine the predictive accuracies of each model. Results: The stomach structure produced models with the highest predictive accuracy overall (96% using Support Vector Machine and Ensemble models), whilst the liver GTV produced models with the lowest predictive accuracy (76% using Logistic Regression). Predictive accuracies across all models ranged from 68-96% (68-87% for TV and 71-96% for OARs). Conclusions: Although a final visual review by an experienced clinician is still required, the automatic contour evaluation method could reduce the time for benchmark case reviews by identifying gross contouring errors. This method could be successfully implemented to support departmental training and the continuous assessment of outlining for clinical staff in the peer-review process, to reduce interobserver variability in contouring and improve interpretation of radiological anatomy.

show abstract

“…En imagenología pediátrica, el aprendizaje automático se ha utilizado para desarrollar modelos de detección y segmentación de imágenes 32 . Así, por ejemplo, se estudiaron 14.036 radiografías infantiles de mano y sus respectivos informes para elaborar un modelo, basado en analisis de redes neuronales, para hacer estimaciones de edad ósea.…”

Section: Aplicaciones En Pediatríaunclassified

Los Big Data: conceptos relacionados y algunas aplicaciones en pediatría

Zepeda

2019

Rev Chil Pediatr

View full text Add to dashboard Cite

El avance de la tecnología médica, el registro de salud electrónico (EHR, por sus siglas en inglés) y la producción compleja de datos biomoleculares están generando grandes volúmenes de información, en varios formatos y de múltiples fuentes, que se conocen como “Big Data”. El análisis integrado de estos datos ha abierto una amplia posibilidad para explorar respuestas a problemas de salud. En pediatría, se han incrementado los estudios se analizan Big Data o se utilizan las herramientas informáticas que se han desarrollado para analizar estos datos. Los propósitos de esos estudios han sido variados, por ejemplo: en la detección y prevención temprana de una amplia gama de afecciones médicas, mejoramiento de los diagnósticos, para especificar tratamientos o anticipar el resultado de alguna patología, etc. El presente documento tiene como objetivo revisar los conceptos principales involucrados en el análisis de Big Data o en las tecnologías informáticas asociadas, así como también examinar sus aplicaciones, potencialidades y limitaciones actuales. Este estudio se realizó sobre la base de una revisión bibliográfica no sistemática, centrada en el campo de la pediatría. En la selección de los ejemplos de aplicación, se consideró que eran fuentes primarias, publicadas en los últimos cinco años y con poblaciones infanto-juveniles.

show abstract

Machine learning concepts, concerns and opportunities for a pediatric radiologist

Cited by 46 publications

References 14 publications

Performance of deep learning-based algorithm for detection of ileocolic intussusception on abdominal radiographs of young children

Performance of deep learning-based algorithm for detection of ileocolic intussusception on abdominal radiographs of young children

Automatic evaluation of contours in radiotherapy planning utilising conformity indices and machine learning

Los Big Data: conceptos relacionados y algunas aplicaciones en pediatría

Contact Info

Product

Resources

About