How an Automated System to Identify Children at Risk for High Radiation Exposure from Computed Tomography Can Be Used to Inform Research and Clinical Practice*

Background Over half a million individuals are diagnosed with head and neck cancer each year globally. Radiotherapy is an important curative treatment for this disease, but it requires manual time to delineate radiosensitive organs at risk. This planning process can delay treatment while also introducing interoperator variability, resulting in downstream radiation dose differences. Although auto-segmentation algorithms offer a potentially time-saving solution, the challenges in defining, quantifying, and achieving expert performance remain. Objective Adopting a deep learning approach, we aim to demonstrate a 3D U-Net architecture that achieves expert-level performance in delineating 21 distinct head and neck organs at risk commonly segmented in clinical practice. Methods The model was trained on a data set of 663 deidentified computed tomography scans acquired in routine clinical practice and with both segmentations taken from clinical practice and segmentations created by experienced radiographers as part of this research, all in accordance with consensus organ at risk definitions. Results We demonstrated the model’s clinical applicability by assessing its performance on a test set of 21 computed tomography scans from clinical practice, each with 21 organs at risk segmented by 2 independent experts. We also introduced surface Dice similarity coefficient, a new metric for the comparison of organ delineation, to quantify the deviation between organ at risk surface contours rather than volumes, better reflecting the clinical task of correcting errors in automated organ segmentations. The model’s generalizability was then demonstrated on 2 distinct open-source data sets, reflecting different centers and countries to model training. Conclusions Deep learning is an effective and clinically applicable technique for the segmentation of the head and neck anatomy for radiotherapy. With appropriate validation studies and regulatory approvals, this system could improve the efficiency, consistency, and safety of radiotherapy pathways.

Download Full-text

Clinician Judgment Versus a Decision Rule for Identifying Children at Risk of Traumatic Brain Injury on Computed Tomography After Blunt Head Trauma

Yearbook of Diagnostic Radiology ◽

10.1016/s0098-1672(10)79146-3 ◽

2010 ◽

Vol 2010 ◽

pp. 151-152

Author(s):

A.C. Offiah

Keyword(s):

Computed Tomography ◽

Traumatic Brain Injury ◽

At Risk ◽

Brain Injury ◽

Decision Rule ◽

Head Trauma ◽

Children At Risk ◽

Blunt Head Trauma ◽

Clinician Judgment

Download Full-text

Clinician Judgment Versus a Decision Rule for Identifying Children at Risk of Traumatic Brain Injury on Computed Tomography After Blunt Head Trauma

Pediatric Emergency Care ◽

10.1097/pec.0b013e318196e9c0 ◽

2009 ◽

Vol 25 (2) ◽

pp. 61-65 ◽

Cited By ~ 15

Author(s):

Michael J. Palchak ◽

James F. Holmes ◽

Nathan Kuppermann

Keyword(s):

Computed Tomography ◽

Traumatic Brain Injury ◽

At Risk ◽

Brain Injury ◽

Decision Rule ◽

Head Trauma ◽

Children At Risk ◽

Blunt Head Trauma ◽

Clinician Judgment

Download Full-text

Clinically Applicable Segmentation of Head and Neck Anatomy for Radiotherapy: Deep Learning Algorithm Development and Validation Study (Preprint)

10.2196/preprints.26151 ◽

2020 ◽

Author(s):

Stanislav Nikolov ◽

Sam Blackwell ◽

Alexei Zverovitch ◽

Ruheena Mendes ◽

Michelle Livne ◽

...

Keyword(s):

Computed Tomography ◽

At Risk ◽

Deep Learning ◽

Clinical Practice ◽

Head And Neck ◽

Organs At Risk ◽

Data Set ◽

Computed Tomography Scans ◽

Neck Anatomy ◽

Organ At Risk

BACKGROUND Over half a million individuals are diagnosed with head and neck cancer each year globally. Radiotherapy is an important curative treatment for this disease, but it requires manual time to delineate radiosensitive organs at risk. This planning process can delay treatment while also introducing interoperator variability, resulting in downstream radiation dose differences. Although auto-segmentation algorithms offer a potentially time-saving solution, the challenges in defining, quantifying, and achieving expert performance remain. OBJECTIVE Adopting a deep learning approach, we aim to demonstrate a 3D U-Net architecture that achieves expert-level performance in delineating 21 distinct head and neck organs at risk commonly segmented in clinical practice. METHODS The model was trained on a data set of 663 deidentified computed tomography scans acquired in routine clinical practice and with both segmentations taken from clinical practice and segmentations created by experienced radiographers as part of this research, all in accordance with consensus organ at risk definitions. RESULTS We demonstrated the model’s clinical applicability by assessing its performance on a test set of 21 computed tomography scans from clinical practice, each with 21 organs at risk segmented by 2 independent experts. We also introduced surface Dice similarity coefficient, a new metric for the comparison of organ delineation, to quantify the deviation between organ at risk surface contours rather than volumes, better reflecting the clinical task of correcting errors in automated organ segmentations. The model’s generalizability was then demonstrated on 2 distinct open-source data sets, reflecting different centers and countries to model training. CONCLUSIONS Deep learning is an effective and clinically applicable technique for the segmentation of the head and neck anatomy for radiotherapy. With appropriate validation studies and regulatory approvals, this system could improve the efficiency, consistency, and safety of radiotherapy pathways.

Download Full-text

Research With Young Children at Risk for Speech/Language Development Disorders

Perspectives on Augmentative and Alternative Communication ◽

10.1044/aac8.2.1 ◽

1999 ◽

Vol 8 (2) ◽

pp. 1-2

Keyword(s):

At Risk ◽

Young Children ◽

Language Development ◽

Children At Risk ◽

Language Development Disorders

Download Full-text

Discriminant Validation of the Test of Reading Performance (TORP) for Identifying Children at Risk of Reading Difficulties

European Journal of Psychological Assessment ◽

10.1027//1015-5759.16.2.139 ◽

2000 ◽

Vol 16 (2) ◽

pp. 139-146 ◽

Cited By ~ 17

Author(s):

Padeliadu Susana ◽

Georgios D. Sideridis

Keyword(s):

At Risk ◽

Elementary School Students ◽

Reading Difficulties ◽

Reading Performance ◽

Low Achievers ◽

Children At Risk ◽

School Students ◽

Students At Risk ◽

Group A ◽

The Mean

Abstract This study investigated the discriminant validation of the Test of Reading Performance (TORP), a new scale designed to evaluate the reading performance of elementary-school students. The sample consisted of 181 elementary-school students drawn from public elementary schools in northern Greece using stratified random procedures. The TORP was hypothesized to measure six constructs, namely: “letter knowledge,” “phoneme blending,” “word identification,” “syntax,” “morphology,” and “passage comprehension.” Using standard deviations (SD) from the mean, three groups of students were formed as follows: A group of low achievers in reading (N = 9) including students who scored between -1 and -1.5 SD from the mean of the group. A group of students at risk of reading difficulties (N = 6) including students who scored between -1.5 and -2 SDs below the mean of the group. A group of students at risk of serious reading difficulties (N = 6) including students who scored -2 or more SDs below the mean of the group. The rest of the students (no risk, N = 122) comprised the fourth group. Using discriminant analyses it was evaluated how well the linear combination of the 15 variables that comprised the TORP could discriminate students of different reading ability. Results indicated that correct classification rates for low achievers, those at risk for reading problems, those at risk of serious reading problems, and the no-risk group were 89%, 100%, 83%, and 97%, respectively. Evidence for partial validation of the TORP was provided through the use of confirmatory factor analysis and indices of sensitivity and specificity. It is concluded that the TORP can be ut ilized for the identification of children at risk for low achievement in reading. Analysis of the misclassified cases indicated that increased variability might have been responsible for the existing misclassification. More research is needed to determine the discriminant validation of TORP with samples of children with specific reading disabilities.

Download Full-text