close
close
Posted in

AI brings x-ray scoliosis monitoring into the modern age

An AI model has provided a long-awaited update to pediatric bone growth predictions used in X-ray imaging to monitor scoliosis, according to a study published April 30 in Radiology.

The model’s growth curve predictions are derived from standing radiographs of a diverse pediatric population and proved to be more accurate than current methods for guiding predictions, noted lead author John Zech, MD, of New York University Langone Health in New York City . and colleagues.

“Although the Anderson-Green standards are still widely used today, they are based on a sample of only 100 children whose growth was assessed more than 60 years ago and who were not racially or ethnically diverse,” the wrote group.

Whole-body biplanar slot scanning is a form of low-dose digital X-ray imaging used to monitor scoliosis. In most patients, imaging occurs at six-month intervals over a period of one to five years. This allows doctors to estimate growth over time and compare it to statistical averages using the Anderson-Green method, the authors explain.

Additionally, although manually measuring femoral and tibial length on radiographs has been shown to be reliable, it is a tedious and time-consuming task, the authors said.

In this study, the authors aimed to use AI to improve on these methods by training a convolutional neural network (CNN) model to automatically measure femorotibial length on radiographs from a racially diverse group of pediatric patients.

The study data included 1,874 examinations in 523 pediatric patients aged 0 to 21 years who underwent at least two slot-scanning radiographs during routine clinical care. Forty percent of patients identified themselves as white and not Hispanic or Latino and 60% identified themselves as belonging to another racial or ethnic group, the authors noted.

The lower extremity measurement pipeline is illustrated using a representative slot scanning radiograph.  The AI ​​model was trained to segment the femur and tibia: (A) the original radiograph and (B) the same radiograph showing the segmentations (boxes) produced by the model.  (C) The segmentations in B were used by the model to identify the top of the femoral head, medial tibial condyle, and tibial ceiling (lines).  These locations were used by the model to measure the total length of the femur, tibia, and lower extremities.  Image courtesy of Radiology.The lower extremity measurement pipeline is illustrated using a representative slot scanning radiograph. The AI ​​model was trained to segment the femur and tibia: (A) the original radiograph and (B) the same radiograph showing the segmentations (boxes) produced by the model. (C) The segmentations in B were used by the model to identify the top of the femoral head, medial tibial condyle, and tibial ceiling (lines). These locations were used by the model to measure the total length of the femur, tibia, and lower extremities. Image courtesy of Radiology.

The CNN model was trained to segment the femur and tibia on the radiographs and measure the total length of the leg, femur, and tibia. The results showed that the average absolute error measurements of the model were 0.25 cm for the femur, 0.27 cm for the tibia, and 0.33 cm for composite lower extremities.

The AI ​​measurements were then used to create femoral and tibial growth curves and the researchers compared these to those derived from the Anderson-Green method.

According to the analysis, AI growth curves represented lower limb growth more accurately in an external test set (n=154) than the Anderson-Green method. Growth reference ranges generated using the AI ​​model showed a higher coverage probability than those generated using the Anderson-Green method (86.7% vs. 73.4%; p < 0.001), the authors reported.

Ultimately, the authors noted that this higher probability of coverage demonstrates the potential value of updated standards based on a diverse population. The higher coverage may reflect greater height and greater variability in height in the modern child population compared to the children on whom the Anderson-Green standards were based, they suggested.

“The AI ​​model provides an easily scalable method to obtain growth data from additional patients, allowing for the creation of stratified standards that better match a child’s individual growth profile,” the researchers concluded.

The full study is available here.